{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 13980,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 4.056429511946979,
      "learning_rate": 2.3809523809523813e-08,
      "loss": 0.3022,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.254068828741834,
      "learning_rate": 4.7619047619047627e-08,
      "loss": 0.3284,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.7081250197058817,
      "learning_rate": 7.142857142857144e-08,
      "loss": 0.3585,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.03035489196616,
      "learning_rate": 9.523809523809525e-08,
      "loss": 0.3019,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.8736544756581472,
      "learning_rate": 1.1904761904761906e-07,
      "loss": 0.3255,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.3222832104609203,
      "learning_rate": 1.4285714285714287e-07,
      "loss": 0.3255,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.4210795474931173,
      "learning_rate": 1.6666666666666668e-07,
      "loss": 0.3276,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.5634613739343233,
      "learning_rate": 1.904761904761905e-07,
      "loss": 0.2726,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.6637944779237985,
      "learning_rate": 2.142857142857143e-07,
      "loss": 0.3569,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.5471335549365794,
      "learning_rate": 2.3809523809523811e-07,
      "loss": 0.2951,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.977026469672543,
      "learning_rate": 2.6190476190476194e-07,
      "loss": 0.3717,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.2639479730612186,
      "learning_rate": 2.8571428571428575e-07,
      "loss": 0.331,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.5942630021409734,
      "learning_rate": 3.0952380952380955e-07,
      "loss": 0.3077,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.4934757744517793,
      "learning_rate": 3.3333333333333335e-07,
      "loss": 0.3157,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.3732383934025227,
      "learning_rate": 3.5714285714285716e-07,
      "loss": 0.2967,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.6149605542926895,
      "learning_rate": 3.80952380952381e-07,
      "loss": 0.3424,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.5210734613974997,
      "learning_rate": 4.047619047619048e-07,
      "loss": 0.276,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.061422767153501,
      "learning_rate": 4.285714285714286e-07,
      "loss": 0.3026,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.160475698653178,
      "learning_rate": 4.523809523809524e-07,
      "loss": 0.3297,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.5495914807491262,
      "learning_rate": 4.7619047619047623e-07,
      "loss": 0.2812,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.6927500779564375,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.3275,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.5205250700987505,
      "learning_rate": 5.238095238095239e-07,
      "loss": 0.3052,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.742367553400832,
      "learning_rate": 5.476190476190477e-07,
      "loss": 0.2851,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.9650903686604932,
      "learning_rate": 5.714285714285715e-07,
      "loss": 0.3266,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.4156996750251345,
      "learning_rate": 5.952380952380953e-07,
      "loss": 0.3231,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.3692584103692957,
      "learning_rate": 6.190476190476191e-07,
      "loss": 0.2806,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.6932628521462156,
      "learning_rate": 6.428571428571428e-07,
      "loss": 0.3239,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.3137713650870695,
      "learning_rate": 6.666666666666667e-07,
      "loss": 0.2365,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.829652477374651,
      "learning_rate": 6.904761904761906e-07,
      "loss": 0.3574,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.6553668991478743,
      "learning_rate": 7.142857142857143e-07,
      "loss": 0.2845,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.495345310085186,
      "learning_rate": 7.380952380952381e-07,
      "loss": 0.7697,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.389002003838203,
      "learning_rate": 7.61904761904762e-07,
      "loss": 0.2958,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2865110242169084,
      "learning_rate": 7.857142857142857e-07,
      "loss": 0.3121,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.4238725380166968,
      "learning_rate": 8.095238095238096e-07,
      "loss": 0.258,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.355608078103788,
      "learning_rate": 8.333333333333333e-07,
      "loss": 0.2702,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.427643609197508,
      "learning_rate": 8.571428571428572e-07,
      "loss": 0.9491,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9928679067193384,
      "learning_rate": 8.80952380952381e-07,
      "loss": 0.2227,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2340918480144483,
      "learning_rate": 9.047619047619048e-07,
      "loss": 0.2885,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.193974440968795,
      "learning_rate": 9.285714285714287e-07,
      "loss": 0.3291,
      "step": 39
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1196758770255992,
      "learning_rate": 9.523809523809525e-07,
      "loss": 0.2951,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1490244515586836,
      "learning_rate": 9.761904761904764e-07,
      "loss": 0.2745,
      "step": 41
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.0259501280518046,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.2335,
      "step": 42
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.386078433373896,
      "learning_rate": 1.023809523809524e-06,
      "loss": 0.2754,
      "step": 43
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.3222612620184395,
      "learning_rate": 1.0476190476190478e-06,
      "loss": 0.268,
      "step": 44
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9842027523787453,
      "learning_rate": 1.0714285714285714e-06,
      "loss": 0.2357,
      "step": 45
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9404390155680744,
      "learning_rate": 1.0952380952380954e-06,
      "loss": 0.2369,
      "step": 46
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1724755446218986,
      "learning_rate": 1.1190476190476192e-06,
      "loss": 0.2839,
      "step": 47
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.147205523844692,
      "learning_rate": 1.142857142857143e-06,
      "loss": 0.2807,
      "step": 48
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1451415010931316,
      "learning_rate": 1.1666666666666668e-06,
      "loss": 0.2399,
      "step": 49
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2977228806377155,
      "learning_rate": 1.1904761904761906e-06,
      "loss": 0.2535,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.1811249474428,
      "learning_rate": 1.2142857142857144e-06,
      "loss": 0.6552,
      "step": 51
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1343221115649977,
      "learning_rate": 1.2380952380952382e-06,
      "loss": 0.2532,
      "step": 52
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2639059478688472,
      "learning_rate": 1.261904761904762e-06,
      "loss": 0.2586,
      "step": 53
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.0089478289042684,
      "learning_rate": 1.2857142857142856e-06,
      "loss": 0.2402,
      "step": 54
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.24336603317913,
      "learning_rate": 1.3095238095238096e-06,
      "loss": 0.2846,
      "step": 55
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9587757608266414,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 0.2065,
      "step": 56
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.930728490515616,
      "learning_rate": 1.3571428571428572e-06,
      "loss": 0.2133,
      "step": 57
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.183490676647861,
      "learning_rate": 1.3809523809523812e-06,
      "loss": 0.2805,
      "step": 58
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1172476465345884,
      "learning_rate": 1.4047619047619048e-06,
      "loss": 0.237,
      "step": 59
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.5514224885111245,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.7453,
      "step": 60
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1124935644479677,
      "learning_rate": 1.4523809523809526e-06,
      "loss": 0.2218,
      "step": 61
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.170312311251715,
      "learning_rate": 1.4761904761904762e-06,
      "loss": 0.2409,
      "step": 62
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.0894495745829857,
      "learning_rate": 1.5e-06,
      "loss": 0.1823,
      "step": 63
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8889374506086778,
      "learning_rate": 1.523809523809524e-06,
      "loss": 0.2266,
      "step": 64
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.984488536257302,
      "learning_rate": 1.5476190476190479e-06,
      "loss": 0.211,
      "step": 65
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.145016408706811,
      "learning_rate": 1.5714285714285714e-06,
      "loss": 0.7319,
      "step": 66
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1654520463013207,
      "learning_rate": 1.5952380952380953e-06,
      "loss": 0.2173,
      "step": 67
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.244806015594005,
      "learning_rate": 1.6190476190476193e-06,
      "loss": 0.2628,
      "step": 68
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8942822683828673,
      "learning_rate": 1.642857142857143e-06,
      "loss": 0.1977,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8954494215887965,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.1887,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0499942197816883,
      "learning_rate": 1.6904761904761907e-06,
      "loss": 0.2128,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9144317231484271,
      "learning_rate": 1.7142857142857145e-06,
      "loss": 0.1934,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.104567411561711,
      "learning_rate": 1.738095238095238e-06,
      "loss": 0.2257,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9643248646333762,
      "learning_rate": 1.761904761904762e-06,
      "loss": 0.1816,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2148682540820457,
      "learning_rate": 1.7857142857142859e-06,
      "loss": 0.2361,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2483964630906637,
      "learning_rate": 1.8095238095238097e-06,
      "loss": 0.2452,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1479332406203637,
      "learning_rate": 1.8333333333333333e-06,
      "loss": 0.2634,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0963707561854488,
      "learning_rate": 1.8571428571428573e-06,
      "loss": 0.2121,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.173608815933427,
      "learning_rate": 1.880952380952381e-06,
      "loss": 0.2404,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.885364430718631,
      "learning_rate": 1.904761904761905e-06,
      "loss": 0.7518,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.017850365944102,
      "learning_rate": 1.928571428571429e-06,
      "loss": 0.2323,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0934319220645303,
      "learning_rate": 1.9523809523809527e-06,
      "loss": 0.2107,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0157776224705963,
      "learning_rate": 1.976190476190476e-06,
      "loss": 0.2161,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0963356587727358,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.2073,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1453558286787144,
      "learning_rate": 2.023809523809524e-06,
      "loss": 0.2714,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.105244076484882,
      "learning_rate": 2.047619047619048e-06,
      "loss": 0.1831,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0469816531104628,
      "learning_rate": 2.0714285714285717e-06,
      "loss": 0.1842,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1131678021217355,
      "learning_rate": 2.0952380952380955e-06,
      "loss": 0.2319,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2234116155027306,
      "learning_rate": 2.1190476190476194e-06,
      "loss": 0.2364,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2898981677302066,
      "learning_rate": 2.1428571428571427e-06,
      "loss": 0.2098,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.865692206738114,
      "learning_rate": 2.166666666666667e-06,
      "loss": 0.1888,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2130298092942216,
      "learning_rate": 2.1904761904761908e-06,
      "loss": 0.2547,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.72766461879791,
      "learning_rate": 2.2142857142857146e-06,
      "loss": 0.7045,
      "step": 93
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.977836159104701,
      "learning_rate": 2.2380952380952384e-06,
      "loss": 0.2262,
      "step": 94
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1714672727232327,
      "learning_rate": 2.261904761904762e-06,
      "loss": 0.1557,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9299618073119207,
      "learning_rate": 2.285714285714286e-06,
      "loss": 0.224,
      "step": 96
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.046524239303243,
      "learning_rate": 2.3095238095238098e-06,
      "loss": 0.1843,
      "step": 97
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0839290248878695,
      "learning_rate": 2.3333333333333336e-06,
      "loss": 0.2689,
      "step": 98
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.120963721697617,
      "learning_rate": 2.3571428571428574e-06,
      "loss": 0.1751,
      "step": 99
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9036105564820034,
      "learning_rate": 2.380952380952381e-06,
      "loss": 0.2208,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0485115765607276,
      "learning_rate": 2.404761904761905e-06,
      "loss": 0.2307,
      "step": 101
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.60262679923109,
      "learning_rate": 2.428571428571429e-06,
      "loss": 0.7797,
      "step": 102
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.837861455828818,
      "learning_rate": 2.4523809523809526e-06,
      "loss": 0.2522,
      "step": 103
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1194774167042114,
      "learning_rate": 2.4761904761904764e-06,
      "loss": 0.194,
      "step": 104
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.022736300741432,
      "learning_rate": 2.5e-06,
      "loss": 0.2121,
      "step": 105
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9780281422325785,
      "learning_rate": 2.523809523809524e-06,
      "loss": 0.2147,
      "step": 106
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8383683180089418,
      "learning_rate": 2.547619047619048e-06,
      "loss": 0.2183,
      "step": 107
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9918997649579477,
      "learning_rate": 2.571428571428571e-06,
      "loss": 0.2279,
      "step": 108
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9099623977560096,
      "learning_rate": 2.595238095238096e-06,
      "loss": 0.229,
      "step": 109
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1413284891047724,
      "learning_rate": 2.6190476190476192e-06,
      "loss": 0.2221,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0913816480066716,
      "learning_rate": 2.642857142857143e-06,
      "loss": 0.2277,
      "step": 111
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3357881099876483,
      "learning_rate": 2.666666666666667e-06,
      "loss": 0.2501,
      "step": 112
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.909182560515798,
      "learning_rate": 2.6904761904761906e-06,
      "loss": 0.5005,
      "step": 113
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9189708493261897,
      "learning_rate": 2.7142857142857144e-06,
      "loss": 0.2104,
      "step": 114
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.09162065524797,
      "learning_rate": 2.7380952380952387e-06,
      "loss": 0.2177,
      "step": 115
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0144457732401135,
      "learning_rate": 2.7619047619047625e-06,
      "loss": 0.263,
      "step": 116
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8133612519967228,
      "learning_rate": 2.785714285714286e-06,
      "loss": 0.2118,
      "step": 117
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8419847635786342,
      "learning_rate": 2.8095238095238096e-06,
      "loss": 0.2098,
      "step": 118
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1396979749275866,
      "learning_rate": 2.8333333333333335e-06,
      "loss": 0.1928,
      "step": 119
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0846192623694746,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.2996,
      "step": 120
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0703470958859502,
      "learning_rate": 2.880952380952381e-06,
      "loss": 0.2084,
      "step": 121
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.159704343366871,
      "learning_rate": 2.9047619047619053e-06,
      "loss": 0.2049,
      "step": 122
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0614930122275426,
      "learning_rate": 2.928571428571429e-06,
      "loss": 0.2517,
      "step": 123
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.262587549707648,
      "learning_rate": 2.9523809523809525e-06,
      "loss": 0.2032,
      "step": 124
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.262565782524033,
      "learning_rate": 2.9761904761904763e-06,
      "loss": 0.2264,
      "step": 125
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.056499286779882,
      "learning_rate": 3e-06,
      "loss": 0.2335,
      "step": 126
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0216787118216217,
      "learning_rate": 3.023809523809524e-06,
      "loss": 0.2281,
      "step": 127
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5201661765670345,
      "learning_rate": 3.047619047619048e-06,
      "loss": 0.2485,
      "step": 128
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2942754671277363,
      "learning_rate": 3.071428571428572e-06,
      "loss": 0.2233,
      "step": 129
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.977533385058842,
      "learning_rate": 3.0952380952380957e-06,
      "loss": 0.2035,
      "step": 130
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1238003805874146,
      "learning_rate": 3.1190476190476195e-06,
      "loss": 0.1839,
      "step": 131
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9678074238799244,
      "learning_rate": 3.142857142857143e-06,
      "loss": 0.204,
      "step": 132
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9627148032828434,
      "learning_rate": 3.1666666666666667e-06,
      "loss": 0.2337,
      "step": 133
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1960777229916686,
      "learning_rate": 3.1904761904761905e-06,
      "loss": 0.215,
      "step": 134
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7854201466840234,
      "learning_rate": 3.2142857142857147e-06,
      "loss": 0.1838,
      "step": 135
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8595115145218097,
      "learning_rate": 3.2380952380952385e-06,
      "loss": 0.2047,
      "step": 136
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9915969651414422,
      "learning_rate": 3.2619047619047623e-06,
      "loss": 0.2611,
      "step": 137
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1160146592796214,
      "learning_rate": 3.285714285714286e-06,
      "loss": 0.2288,
      "step": 138
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1093082399019334,
      "learning_rate": 3.3095238095238095e-06,
      "loss": 0.223,
      "step": 139
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9661795407430658,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.2658,
      "step": 140
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8064197669404958,
      "learning_rate": 3.357142857142857e-06,
      "loss": 0.1927,
      "step": 141
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.978841467097062,
      "learning_rate": 3.3809523809523814e-06,
      "loss": 0.2037,
      "step": 142
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.204542478982373,
      "learning_rate": 3.404761904761905e-06,
      "loss": 0.2557,
      "step": 143
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7884326802506234,
      "learning_rate": 3.428571428571429e-06,
      "loss": 0.1888,
      "step": 144
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8491485952298592,
      "learning_rate": 3.4523809523809528e-06,
      "loss": 0.1853,
      "step": 145
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.189118700654106,
      "learning_rate": 3.476190476190476e-06,
      "loss": 0.4302,
      "step": 146
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0070526993070854,
      "learning_rate": 3.5e-06,
      "loss": 0.1866,
      "step": 147
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9238891857286395,
      "learning_rate": 3.523809523809524e-06,
      "loss": 0.1902,
      "step": 148
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3488600553246397,
      "learning_rate": 3.547619047619048e-06,
      "loss": 0.2199,
      "step": 149
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2937836740323254,
      "learning_rate": 3.5714285714285718e-06,
      "loss": 0.2366,
      "step": 150
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.955484153435743,
      "learning_rate": 3.5952380952380956e-06,
      "loss": 0.2138,
      "step": 151
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9696059878227126,
      "learning_rate": 3.6190476190476194e-06,
      "loss": 0.222,
      "step": 152
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.036668231835895,
      "learning_rate": 3.642857142857143e-06,
      "loss": 0.2076,
      "step": 153
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2371897387757533,
      "learning_rate": 3.6666666666666666e-06,
      "loss": 0.2002,
      "step": 154
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1150825877853476,
      "learning_rate": 3.690476190476191e-06,
      "loss": 0.2398,
      "step": 155
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.051675139066703,
      "learning_rate": 3.7142857142857146e-06,
      "loss": 0.2697,
      "step": 156
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9987071450358536,
      "learning_rate": 3.7380952380952384e-06,
      "loss": 0.2921,
      "step": 157
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.852216888343916,
      "learning_rate": 3.761904761904762e-06,
      "loss": 0.2074,
      "step": 158
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9109294703981274,
      "learning_rate": 3.785714285714286e-06,
      "loss": 0.2229,
      "step": 159
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9827822056153939,
      "learning_rate": 3.80952380952381e-06,
      "loss": 0.2044,
      "step": 160
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8958194523778151,
      "learning_rate": 3.833333333333334e-06,
      "loss": 0.2281,
      "step": 161
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0352702433879086,
      "learning_rate": 3.857142857142858e-06,
      "loss": 0.1967,
      "step": 162
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.193590453573038,
      "learning_rate": 3.880952380952381e-06,
      "loss": 0.2713,
      "step": 163
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.087677938534779,
      "learning_rate": 3.9047619047619055e-06,
      "loss": 0.2218,
      "step": 164
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.925880552360889,
      "learning_rate": 3.928571428571429e-06,
      "loss": 0.5797,
      "step": 165
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3067196692705023,
      "learning_rate": 3.952380952380952e-06,
      "loss": 0.2556,
      "step": 166
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.481123179032876,
      "learning_rate": 3.9761904761904764e-06,
      "loss": 0.2198,
      "step": 167
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8364567899800972,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.2337,
      "step": 168
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9577327048730095,
      "learning_rate": 4.023809523809524e-06,
      "loss": 0.2237,
      "step": 169
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3386135226840246,
      "learning_rate": 4.047619047619048e-06,
      "loss": 0.2615,
      "step": 170
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1206970562637233,
      "learning_rate": 4.071428571428572e-06,
      "loss": 0.2118,
      "step": 171
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9515176070657816,
      "learning_rate": 4.095238095238096e-06,
      "loss": 0.1839,
      "step": 172
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9422732216718517,
      "learning_rate": 4.119047619047619e-06,
      "loss": 0.2069,
      "step": 173
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8946655266765382,
      "learning_rate": 4.1428571428571435e-06,
      "loss": 0.1613,
      "step": 174
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7844880333011413,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.1955,
      "step": 175
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2916838035958893,
      "learning_rate": 4.190476190476191e-06,
      "loss": 0.245,
      "step": 176
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9348749130091802,
      "learning_rate": 4.2142857142857145e-06,
      "loss": 0.24,
      "step": 177
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9060851987544007,
      "learning_rate": 4.238095238095239e-06,
      "loss": 0.1964,
      "step": 178
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1361682290003974,
      "learning_rate": 4.261904761904762e-06,
      "loss": 0.2268,
      "step": 179
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0883582444581195,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 0.2392,
      "step": 180
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7727517307423128,
      "learning_rate": 4.30952380952381e-06,
      "loss": 0.1869,
      "step": 181
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.010109935693985,
      "learning_rate": 4.333333333333334e-06,
      "loss": 0.2169,
      "step": 182
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1480669786310616,
      "learning_rate": 4.357142857142857e-06,
      "loss": 0.286,
      "step": 183
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0812727334033108,
      "learning_rate": 4.3809523809523815e-06,
      "loss": 0.2386,
      "step": 184
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.841835771092403,
      "learning_rate": 4.404761904761905e-06,
      "loss": 0.2043,
      "step": 185
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7745460577151568,
      "learning_rate": 4.428571428571429e-06,
      "loss": 0.1829,
      "step": 186
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0917374449433015,
      "learning_rate": 4.4523809523809525e-06,
      "loss": 0.284,
      "step": 187
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9414563378563814,
      "learning_rate": 4.476190476190477e-06,
      "loss": 0.2107,
      "step": 188
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9333551533085978,
      "learning_rate": 4.5e-06,
      "loss": 0.2124,
      "step": 189
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.080567276770741,
      "learning_rate": 4.523809523809524e-06,
      "loss": 0.2392,
      "step": 190
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.012896259654425,
      "learning_rate": 4.547619047619048e-06,
      "loss": 0.2048,
      "step": 191
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.105208492002958,
      "learning_rate": 4.571428571428572e-06,
      "loss": 0.2606,
      "step": 192
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.027416340272993,
      "learning_rate": 4.595238095238095e-06,
      "loss": 0.2133,
      "step": 193
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.87599091564929,
      "learning_rate": 4.6190476190476196e-06,
      "loss": 0.1843,
      "step": 194
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9495957404961308,
      "learning_rate": 4.642857142857144e-06,
      "loss": 0.2022,
      "step": 195
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.148344534365086,
      "learning_rate": 4.666666666666667e-06,
      "loss": 0.2678,
      "step": 196
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8479919100385038,
      "learning_rate": 4.6904761904761905e-06,
      "loss": 0.183,
      "step": 197
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.277673980381613,
      "learning_rate": 4.714285714285715e-06,
      "loss": 0.3075,
      "step": 198
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9395407154116269,
      "learning_rate": 4.738095238095238e-06,
      "loss": 0.2314,
      "step": 199
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.9375029931816763,
      "learning_rate": 4.761904761904762e-06,
      "loss": 0.4656,
      "step": 200
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9769617439412623,
      "learning_rate": 4.785714285714287e-06,
      "loss": 0.2318,
      "step": 201
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0826605803702134,
      "learning_rate": 4.80952380952381e-06,
      "loss": 0.2225,
      "step": 202
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9190098513987295,
      "learning_rate": 4.833333333333333e-06,
      "loss": 0.2028,
      "step": 203
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.19378248589066,
      "learning_rate": 4.857142857142858e-06,
      "loss": 0.6534,
      "step": 204
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8999833167107292,
      "learning_rate": 4.880952380952381e-06,
      "loss": 0.1908,
      "step": 205
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6608307540657135,
      "learning_rate": 4.904761904761905e-06,
      "loss": 0.1698,
      "step": 206
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.050021744810209,
      "learning_rate": 4.928571428571429e-06,
      "loss": 0.2399,
      "step": 207
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9810315170285808,
      "learning_rate": 4.952380952380953e-06,
      "loss": 0.2454,
      "step": 208
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8959122311030974,
      "learning_rate": 4.976190476190477e-06,
      "loss": 0.1822,
      "step": 209
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9266170929302338,
      "learning_rate": 5e-06,
      "loss": 0.2214,
      "step": 210
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8576383563935714,
      "learning_rate": 5.023809523809524e-06,
      "loss": 0.1747,
      "step": 211
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.420005028304979,
      "learning_rate": 5.047619047619048e-06,
      "loss": 0.2087,
      "step": 212
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1053757678492335,
      "learning_rate": 5.071428571428571e-06,
      "loss": 0.2518,
      "step": 213
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1075845919874414,
      "learning_rate": 5.095238095238096e-06,
      "loss": 0.2524,
      "step": 214
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0010763461542953,
      "learning_rate": 5.119047619047619e-06,
      "loss": 0.1987,
      "step": 215
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1973253566392272,
      "learning_rate": 5.142857142857142e-06,
      "loss": 0.24,
      "step": 216
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0411437423876895,
      "learning_rate": 5.1666666666666675e-06,
      "loss": 0.1894,
      "step": 217
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8602203050490493,
      "learning_rate": 5.190476190476192e-06,
      "loss": 0.216,
      "step": 218
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9745010237795677,
      "learning_rate": 5.214285714285715e-06,
      "loss": 0.2325,
      "step": 219
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1845890495625513,
      "learning_rate": 5.2380952380952384e-06,
      "loss": 0.2343,
      "step": 220
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.159794967578708,
      "learning_rate": 5.261904761904763e-06,
      "loss": 0.2306,
      "step": 221
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8229650941306363,
      "learning_rate": 5.285714285714286e-06,
      "loss": 0.2058,
      "step": 222
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0138766720966954,
      "learning_rate": 5.30952380952381e-06,
      "loss": 0.2167,
      "step": 223
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.83581223570138,
      "learning_rate": 5.333333333333334e-06,
      "loss": 0.1959,
      "step": 224
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.009385116397086,
      "learning_rate": 5.357142857142857e-06,
      "loss": 0.8172,
      "step": 225
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.01239880463524,
      "learning_rate": 5.380952380952381e-06,
      "loss": 0.2237,
      "step": 226
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9691087214986362,
      "learning_rate": 5.404761904761905e-06,
      "loss": 0.2073,
      "step": 227
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.906422073689916,
      "learning_rate": 5.428571428571429e-06,
      "loss": 0.2032,
      "step": 228
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.080456628303908,
      "learning_rate": 5.452380952380952e-06,
      "loss": 0.2177,
      "step": 229
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1691464980549333,
      "learning_rate": 5.476190476190477e-06,
      "loss": 0.2563,
      "step": 230
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9609442970542161,
      "learning_rate": 5.500000000000001e-06,
      "loss": 0.2382,
      "step": 231
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.063500837006452,
      "learning_rate": 5.523809523809525e-06,
      "loss": 0.2757,
      "step": 232
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9990118444146079,
      "learning_rate": 5.547619047619048e-06,
      "loss": 0.2847,
      "step": 233
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.040029602463559,
      "learning_rate": 5.571428571428572e-06,
      "loss": 0.7314,
      "step": 234
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9620435655273838,
      "learning_rate": 5.595238095238096e-06,
      "loss": 0.2261,
      "step": 235
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.274342496679555,
      "learning_rate": 5.619047619047619e-06,
      "loss": 0.2366,
      "step": 236
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9279924210380983,
      "learning_rate": 5.6428571428571435e-06,
      "loss": 0.2102,
      "step": 237
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.020322634043729,
      "learning_rate": 5.666666666666667e-06,
      "loss": 0.2393,
      "step": 238
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9948923549138908,
      "learning_rate": 5.690476190476191e-06,
      "loss": 0.2242,
      "step": 239
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.004488668018897,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 0.1864,
      "step": 240
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9465692076813665,
      "learning_rate": 5.738095238095238e-06,
      "loss": 0.281,
      "step": 241
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9236574326408011,
      "learning_rate": 5.761904761904762e-06,
      "loss": 0.2273,
      "step": 242
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.943027503615834,
      "learning_rate": 5.785714285714286e-06,
      "loss": 0.2182,
      "step": 243
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8202266039934019,
      "learning_rate": 5.8095238095238106e-06,
      "loss": 0.2681,
      "step": 244
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8943427872082612,
      "learning_rate": 5.833333333333334e-06,
      "loss": 0.1831,
      "step": 245
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9163628908792816,
      "learning_rate": 5.857142857142858e-06,
      "loss": 0.2219,
      "step": 246
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.02467163952195,
      "learning_rate": 5.8809523809523816e-06,
      "loss": 0.1822,
      "step": 247
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9948307918894201,
      "learning_rate": 5.904761904761905e-06,
      "loss": 0.2588,
      "step": 248
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8784218583060284,
      "learning_rate": 5.928571428571429e-06,
      "loss": 0.1999,
      "step": 249
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9053908951342011,
      "learning_rate": 5.9523809523809525e-06,
      "loss": 0.1974,
      "step": 250
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8344519715845773,
      "learning_rate": 5.976190476190477e-06,
      "loss": 0.1896,
      "step": 251
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.061751098664898,
      "learning_rate": 6e-06,
      "loss": 0.2566,
      "step": 252
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.457791234632552,
      "learning_rate": 6.023809523809524e-06,
      "loss": 0.7121,
      "step": 253
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.290405650754611,
      "learning_rate": 6.047619047619048e-06,
      "loss": 0.6195,
      "step": 254
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.915236806669019,
      "learning_rate": 6.071428571428571e-06,
      "loss": 0.2534,
      "step": 255
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9135861660087692,
      "learning_rate": 6.095238095238096e-06,
      "loss": 0.2378,
      "step": 256
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.77449740006332,
      "learning_rate": 6.11904761904762e-06,
      "loss": 0.9165,
      "step": 257
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9314814238403286,
      "learning_rate": 6.142857142857144e-06,
      "loss": 0.225,
      "step": 258
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8597824236462694,
      "learning_rate": 6.166666666666667e-06,
      "loss": 0.2243,
      "step": 259
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9440300785829745,
      "learning_rate": 6.1904761904761914e-06,
      "loss": 0.1999,
      "step": 260
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8683095191981387,
      "learning_rate": 6.214285714285715e-06,
      "loss": 0.2386,
      "step": 261
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0570951335021994,
      "learning_rate": 6.238095238095239e-06,
      "loss": 0.2346,
      "step": 262
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9807046290463075,
      "learning_rate": 6.261904761904762e-06,
      "loss": 0.2873,
      "step": 263
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9933947520265711,
      "learning_rate": 6.285714285714286e-06,
      "loss": 0.216,
      "step": 264
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9983609559808428,
      "learning_rate": 6.30952380952381e-06,
      "loss": 0.2141,
      "step": 265
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0267206624022767,
      "learning_rate": 6.333333333333333e-06,
      "loss": 0.2014,
      "step": 266
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9637057491999363,
      "learning_rate": 6.357142857142858e-06,
      "loss": 0.2377,
      "step": 267
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.912161557180021,
      "learning_rate": 6.380952380952381e-06,
      "loss": 0.2117,
      "step": 268
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9922123192772643,
      "learning_rate": 6.404761904761904e-06,
      "loss": 0.2168,
      "step": 269
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0524865993741486,
      "learning_rate": 6.4285714285714295e-06,
      "loss": 0.3244,
      "step": 270
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9520151412476405,
      "learning_rate": 6.452380952380954e-06,
      "loss": 0.2487,
      "step": 271
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8674499915952019,
      "learning_rate": 6.476190476190477e-06,
      "loss": 0.2079,
      "step": 272
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.701869163842095,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 0.227,
      "step": 273
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8613229362056498,
      "learning_rate": 6.523809523809525e-06,
      "loss": 0.2185,
      "step": 274
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8961958538148296,
      "learning_rate": 6.547619047619048e-06,
      "loss": 0.2477,
      "step": 275
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.048977212661111,
      "learning_rate": 6.571428571428572e-06,
      "loss": 0.2186,
      "step": 276
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8252064171242572,
      "learning_rate": 6.595238095238096e-06,
      "loss": 0.2038,
      "step": 277
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.020613005565305,
      "learning_rate": 6.619047619047619e-06,
      "loss": 0.2012,
      "step": 278
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.267345323856519,
      "learning_rate": 6.642857142857143e-06,
      "loss": 0.8847,
      "step": 279
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.566668027407411,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.5696,
      "step": 280
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.857475019185172,
      "learning_rate": 6.690476190476191e-06,
      "loss": 0.2412,
      "step": 281
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9371231880100228,
      "learning_rate": 6.714285714285714e-06,
      "loss": 0.2066,
      "step": 282
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.750408234656862,
      "learning_rate": 6.738095238095239e-06,
      "loss": 0.1983,
      "step": 283
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0621317569036024,
      "learning_rate": 6.761904761904763e-06,
      "loss": 0.2746,
      "step": 284
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8704400263261065,
      "learning_rate": 6.785714285714287e-06,
      "loss": 0.1551,
      "step": 285
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2745572060712784,
      "learning_rate": 6.80952380952381e-06,
      "loss": 0.2485,
      "step": 286
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0296284764475665,
      "learning_rate": 6.833333333333334e-06,
      "loss": 0.2083,
      "step": 287
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.663487325561779,
      "learning_rate": 6.857142857142858e-06,
      "loss": 0.6986,
      "step": 288
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9826412010368553,
      "learning_rate": 6.880952380952381e-06,
      "loss": 0.2323,
      "step": 289
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.890359106406398,
      "learning_rate": 6.9047619047619055e-06,
      "loss": 0.2125,
      "step": 290
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.159655777152774,
      "learning_rate": 6.928571428571429e-06,
      "loss": 0.5949,
      "step": 291
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0622253517354467,
      "learning_rate": 6.952380952380952e-06,
      "loss": 0.2269,
      "step": 292
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8794407809613203,
      "learning_rate": 6.9761904761904765e-06,
      "loss": 0.2496,
      "step": 293
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8273633787133798,
      "learning_rate": 7e-06,
      "loss": 0.2123,
      "step": 294
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2604015118849006,
      "learning_rate": 7.023809523809524e-06,
      "loss": 0.2249,
      "step": 295
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9223540799930627,
      "learning_rate": 7.047619047619048e-06,
      "loss": 0.2504,
      "step": 296
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.033197568809861,
      "learning_rate": 7.0714285714285726e-06,
      "loss": 0.2549,
      "step": 297
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9447513722776977,
      "learning_rate": 7.095238095238096e-06,
      "loss": 0.2465,
      "step": 298
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.896405654316611,
      "learning_rate": 7.11904761904762e-06,
      "loss": 0.2532,
      "step": 299
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9867067163972432,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 0.2049,
      "step": 300
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.3317875248763205,
      "learning_rate": 7.166666666666667e-06,
      "loss": 0.5459,
      "step": 301
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9372191264235612,
      "learning_rate": 7.190476190476191e-06,
      "loss": 0.2256,
      "step": 302
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6900627862221096,
      "learning_rate": 7.2142857142857145e-06,
      "loss": 0.1987,
      "step": 303
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9262794348517647,
      "learning_rate": 7.238095238095239e-06,
      "loss": 0.2168,
      "step": 304
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7828153071973336,
      "learning_rate": 7.261904761904762e-06,
      "loss": 0.2578,
      "step": 305
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7914881338820046,
      "learning_rate": 7.285714285714286e-06,
      "loss": 0.2193,
      "step": 306
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0251201826842347,
      "learning_rate": 7.30952380952381e-06,
      "loss": 0.2115,
      "step": 307
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8815559512328361,
      "learning_rate": 7.333333333333333e-06,
      "loss": 0.2182,
      "step": 308
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9393022963495439,
      "learning_rate": 7.357142857142858e-06,
      "loss": 0.2492,
      "step": 309
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8774064763511955,
      "learning_rate": 7.380952380952382e-06,
      "loss": 0.2701,
      "step": 310
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7410433659647047,
      "learning_rate": 7.404761904761906e-06,
      "loss": 0.228,
      "step": 311
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8218936762193778,
      "learning_rate": 7.428571428571429e-06,
      "loss": 0.2339,
      "step": 312
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7743420188901728,
      "learning_rate": 7.4523809523809534e-06,
      "loss": 0.2082,
      "step": 313
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9498876741358502,
      "learning_rate": 7.476190476190477e-06,
      "loss": 0.2275,
      "step": 314
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8340673029230719,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.1732,
      "step": 315
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7993362037784466,
      "learning_rate": 7.523809523809524e-06,
      "loss": 0.2544,
      "step": 316
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.589264421487482,
      "learning_rate": 7.547619047619048e-06,
      "loss": 0.6711,
      "step": 317
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.918163110200814,
      "learning_rate": 7.571428571428572e-06,
      "loss": 0.253,
      "step": 318
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.794946759088491,
      "learning_rate": 7.595238095238095e-06,
      "loss": 0.2078,
      "step": 319
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8650702974456617,
      "learning_rate": 7.61904761904762e-06,
      "loss": 0.2106,
      "step": 320
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.074954415910446,
      "learning_rate": 7.642857142857143e-06,
      "loss": 0.2561,
      "step": 321
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7041227480686176,
      "learning_rate": 7.666666666666667e-06,
      "loss": 0.2088,
      "step": 322
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7710638920386534,
      "learning_rate": 7.690476190476191e-06,
      "loss": 0.2193,
      "step": 323
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.776195977604689,
      "learning_rate": 7.714285714285716e-06,
      "loss": 0.2069,
      "step": 324
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.333406501579369,
      "learning_rate": 7.738095238095238e-06,
      "loss": 0.6174,
      "step": 325
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7606158122625184,
      "learning_rate": 7.761904761904762e-06,
      "loss": 0.1681,
      "step": 326
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8052289128026775,
      "learning_rate": 7.785714285714287e-06,
      "loss": 0.2403,
      "step": 327
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7741816428620443,
      "learning_rate": 7.809523809523811e-06,
      "loss": 0.2368,
      "step": 328
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8610983125290597,
      "learning_rate": 7.833333333333333e-06,
      "loss": 0.2245,
      "step": 329
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.04955463527288,
      "learning_rate": 7.857142857142858e-06,
      "loss": 0.2186,
      "step": 330
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0167244128934745,
      "learning_rate": 7.880952380952382e-06,
      "loss": 0.297,
      "step": 331
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8044653628407887,
      "learning_rate": 7.904761904761904e-06,
      "loss": 0.2556,
      "step": 332
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.928061261332684,
      "learning_rate": 7.928571428571429e-06,
      "loss": 0.5937,
      "step": 333
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9778530731030248,
      "learning_rate": 7.952380952380953e-06,
      "loss": 0.2515,
      "step": 334
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.32571490301034,
      "learning_rate": 7.976190476190477e-06,
      "loss": 0.2326,
      "step": 335
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.001011449389012,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.2387,
      "step": 336
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8572135430582049,
      "learning_rate": 8.023809523809526e-06,
      "loss": 0.2426,
      "step": 337
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9183159668971745,
      "learning_rate": 8.047619047619048e-06,
      "loss": 0.2463,
      "step": 338
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.786747172402393,
      "learning_rate": 8.071428571428572e-06,
      "loss": 0.2279,
      "step": 339
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.803676540185769,
      "learning_rate": 8.095238095238097e-06,
      "loss": 0.2308,
      "step": 340
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7753751618746538,
      "learning_rate": 8.119047619047619e-06,
      "loss": 0.2113,
      "step": 341
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8864489687496189,
      "learning_rate": 8.142857142857143e-06,
      "loss": 0.2336,
      "step": 342
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9525613516439526,
      "learning_rate": 8.166666666666668e-06,
      "loss": 0.2354,
      "step": 343
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.584175122823716,
      "learning_rate": 8.190476190476192e-06,
      "loss": 0.2305,
      "step": 344
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5747505364884826,
      "learning_rate": 8.214285714285714e-06,
      "loss": 0.1818,
      "step": 345
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6505214720130668,
      "learning_rate": 8.238095238095239e-06,
      "loss": 0.1847,
      "step": 346
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7565908218840163,
      "learning_rate": 8.261904761904763e-06,
      "loss": 0.2054,
      "step": 347
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9977989317234695,
      "learning_rate": 8.285714285714287e-06,
      "loss": 0.2662,
      "step": 348
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0304709951751736,
      "learning_rate": 8.309523809523811e-06,
      "loss": 0.2218,
      "step": 349
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.003160810603304,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.2745,
      "step": 350
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7425848846524117,
      "learning_rate": 8.357142857142858e-06,
      "loss": 0.1893,
      "step": 351
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.838533165666302,
      "learning_rate": 8.380952380952382e-06,
      "loss": 0.2607,
      "step": 352
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7917766172163547,
      "learning_rate": 8.404761904761905e-06,
      "loss": 0.1962,
      "step": 353
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7148094981523583,
      "learning_rate": 8.428571428571429e-06,
      "loss": 0.1825,
      "step": 354
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.891884751231864,
      "learning_rate": 8.452380952380953e-06,
      "loss": 0.1998,
      "step": 355
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8160875367492477,
      "learning_rate": 8.476190476190477e-06,
      "loss": 0.1936,
      "step": 356
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.770843235882012,
      "learning_rate": 8.5e-06,
      "loss": 0.2507,
      "step": 357
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9045797791842085,
      "learning_rate": 8.523809523809524e-06,
      "loss": 0.2504,
      "step": 358
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9437448543812206,
      "learning_rate": 8.547619047619048e-06,
      "loss": 0.2439,
      "step": 359
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7558706974958636,
      "learning_rate": 8.571428571428571e-06,
      "loss": 0.2112,
      "step": 360
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8816222506716003,
      "learning_rate": 8.595238095238097e-06,
      "loss": 0.219,
      "step": 361
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8973290498259552,
      "learning_rate": 8.61904761904762e-06,
      "loss": 0.2357,
      "step": 362
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8179949121328882,
      "learning_rate": 8.642857142857144e-06,
      "loss": 0.2593,
      "step": 363
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9645533382791447,
      "learning_rate": 8.666666666666668e-06,
      "loss": 0.2791,
      "step": 364
    },
    {
      "epoch": 0.03,
      "grad_norm": 32.72426687805572,
      "learning_rate": 8.690476190476192e-06,
      "loss": 0.2297,
      "step": 365
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.745976022432055,
      "learning_rate": 8.714285714285715e-06,
      "loss": 0.6184,
      "step": 366
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.63107336379878,
      "learning_rate": 8.738095238095239e-06,
      "loss": 0.6297,
      "step": 367
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7141587525428186,
      "learning_rate": 8.761904761904763e-06,
      "loss": 0.2092,
      "step": 368
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7550000113092061,
      "learning_rate": 8.785714285714286e-06,
      "loss": 0.2486,
      "step": 369
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.737687192164475,
      "learning_rate": 8.80952380952381e-06,
      "loss": 0.2032,
      "step": 370
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7101112507842706,
      "learning_rate": 8.833333333333334e-06,
      "loss": 0.2235,
      "step": 371
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.7737299917561105,
      "learning_rate": 8.857142857142858e-06,
      "loss": 0.8163,
      "step": 372
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.846462983341253,
      "learning_rate": 8.88095238095238e-06,
      "loss": 0.2444,
      "step": 373
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8109692510286255,
      "learning_rate": 8.904761904761905e-06,
      "loss": 0.2336,
      "step": 374
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9405563274944506,
      "learning_rate": 8.92857142857143e-06,
      "loss": 0.2432,
      "step": 375
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9179483007281288,
      "learning_rate": 8.952380952380953e-06,
      "loss": 0.233,
      "step": 376
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.857294871080733,
      "learning_rate": 8.976190476190478e-06,
      "loss": 0.239,
      "step": 377
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.647348351590959,
      "learning_rate": 9e-06,
      "loss": 0.2399,
      "step": 378
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.094151279704169,
      "learning_rate": 9.023809523809524e-06,
      "loss": 0.2148,
      "step": 379
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.372210893251924,
      "learning_rate": 9.047619047619049e-06,
      "loss": 0.6163,
      "step": 380
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.775630431961017,
      "learning_rate": 9.071428571428573e-06,
      "loss": 0.2868,
      "step": 381
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9282720686813273,
      "learning_rate": 9.095238095238095e-06,
      "loss": 0.2492,
      "step": 382
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7765017512034897,
      "learning_rate": 9.11904761904762e-06,
      "loss": 0.2432,
      "step": 383
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.839053609571583,
      "learning_rate": 9.142857142857144e-06,
      "loss": 0.2602,
      "step": 384
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.630754028221605,
      "learning_rate": 9.166666666666666e-06,
      "loss": 0.1926,
      "step": 385
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9099510546191294,
      "learning_rate": 9.19047619047619e-06,
      "loss": 0.1862,
      "step": 386
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9361974835701143,
      "learning_rate": 9.214285714285715e-06,
      "loss": 0.2253,
      "step": 387
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9048612694275633,
      "learning_rate": 9.238095238095239e-06,
      "loss": 0.2587,
      "step": 388
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9553163714996091,
      "learning_rate": 9.261904761904763e-06,
      "loss": 0.2598,
      "step": 389
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.933166728189562,
      "learning_rate": 9.285714285714288e-06,
      "loss": 0.2263,
      "step": 390
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.031038888645553,
      "learning_rate": 9.30952380952381e-06,
      "loss": 0.2761,
      "step": 391
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9692502082507126,
      "learning_rate": 9.333333333333334e-06,
      "loss": 0.2266,
      "step": 392
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9290804306393872,
      "learning_rate": 9.357142857142859e-06,
      "loss": 0.1872,
      "step": 393
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.691436975844405,
      "learning_rate": 9.380952380952381e-06,
      "loss": 0.2216,
      "step": 394
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9862039722299842,
      "learning_rate": 9.404761904761905e-06,
      "loss": 0.2296,
      "step": 395
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6838464536412863,
      "learning_rate": 9.42857142857143e-06,
      "loss": 0.2287,
      "step": 396
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.664115811248073,
      "learning_rate": 9.452380952380952e-06,
      "loss": 0.2046,
      "step": 397
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7851447950299109,
      "learning_rate": 9.476190476190476e-06,
      "loss": 0.1923,
      "step": 398
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.267739813716185,
      "learning_rate": 9.5e-06,
      "loss": 0.6433,
      "step": 399
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7788691572222957,
      "learning_rate": 9.523809523809525e-06,
      "loss": 0.2051,
      "step": 400
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9296463534345818,
      "learning_rate": 9.547619047619049e-06,
      "loss": 0.2333,
      "step": 401
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8142126861811512,
      "learning_rate": 9.571428571428573e-06,
      "loss": 0.2505,
      "step": 402
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9692530937233412,
      "learning_rate": 9.595238095238096e-06,
      "loss": 0.2093,
      "step": 403
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.856043587537392,
      "learning_rate": 9.61904761904762e-06,
      "loss": 0.2635,
      "step": 404
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.94443034052922,
      "learning_rate": 9.642857142857144e-06,
      "loss": 0.861,
      "step": 405
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7327783967262806,
      "learning_rate": 9.666666666666667e-06,
      "loss": 0.2482,
      "step": 406
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8315765464465676,
      "learning_rate": 9.690476190476191e-06,
      "loss": 0.2289,
      "step": 407
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.112221522236389,
      "learning_rate": 9.714285714285715e-06,
      "loss": 0.2705,
      "step": 408
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9799258573051932,
      "learning_rate": 9.73809523809524e-06,
      "loss": 0.231,
      "step": 409
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7311808292751452,
      "learning_rate": 9.761904761904762e-06,
      "loss": 0.2559,
      "step": 410
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8502205815431971,
      "learning_rate": 9.785714285714286e-06,
      "loss": 0.212,
      "step": 411
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.738527229864594,
      "learning_rate": 9.80952380952381e-06,
      "loss": 0.2662,
      "step": 412
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6655245824942482,
      "learning_rate": 9.833333333333333e-06,
      "loss": 0.1926,
      "step": 413
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.552187474251555,
      "learning_rate": 9.857142857142859e-06,
      "loss": 0.209,
      "step": 414
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5768048771715486,
      "learning_rate": 9.880952380952381e-06,
      "loss": 0.2348,
      "step": 415
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6964878116119368,
      "learning_rate": 9.904761904761906e-06,
      "loss": 0.2021,
      "step": 416
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0526863748044817,
      "learning_rate": 9.92857142857143e-06,
      "loss": 0.2305,
      "step": 417
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.787547142632327,
      "learning_rate": 9.952380952380954e-06,
      "loss": 0.2533,
      "step": 418
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9419923735530755,
      "learning_rate": 9.976190476190477e-06,
      "loss": 0.2618,
      "step": 419
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1373191265477978,
      "learning_rate": 1e-05,
      "loss": 0.2711,
      "step": 420
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9310276169897496,
      "learning_rate": 9.999999865809933e-06,
      "loss": 0.2946,
      "step": 421
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7379387372744968,
      "learning_rate": 9.999999463239735e-06,
      "loss": 0.198,
      "step": 422
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8245445172166241,
      "learning_rate": 9.999998792289432e-06,
      "loss": 0.2517,
      "step": 423
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.822001276248104,
      "learning_rate": 9.999997852959055e-06,
      "loss": 0.2808,
      "step": 424
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9205509692763958,
      "learning_rate": 9.99999664524866e-06,
      "loss": 0.3423,
      "step": 425
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7888240575558378,
      "learning_rate": 9.999995169158308e-06,
      "loss": 0.2541,
      "step": 426
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7286963649374087,
      "learning_rate": 9.99999342468808e-06,
      "loss": 0.2232,
      "step": 427
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.780722086503442,
      "learning_rate": 9.999991411838067e-06,
      "loss": 0.1884,
      "step": 428
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.498913104958229,
      "learning_rate": 9.999989130608382e-06,
      "loss": 0.2023,
      "step": 429
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9628150553675676,
      "learning_rate": 9.99998658099914e-06,
      "loss": 0.213,
      "step": 430
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.94943086884824,
      "learning_rate": 9.999983763010485e-06,
      "loss": 0.6611,
      "step": 431
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8853688764627523,
      "learning_rate": 9.999980676642564e-06,
      "loss": 0.2968,
      "step": 432
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.231855994331173,
      "learning_rate": 9.999977321895547e-06,
      "loss": 0.5226,
      "step": 433
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.540912893938073,
      "learning_rate": 9.999973698769608e-06,
      "loss": 0.1505,
      "step": 434
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.971543961049707,
      "learning_rate": 9.999969807264946e-06,
      "loss": 0.758,
      "step": 435
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0221156004050185,
      "learning_rate": 9.99996564738177e-06,
      "loss": 0.2295,
      "step": 436
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8469466566406818,
      "learning_rate": 9.9999612191203e-06,
      "loss": 0.2348,
      "step": 437
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7898127914561228,
      "learning_rate": 9.999956522480776e-06,
      "loss": 0.2135,
      "step": 438
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.795620819692547,
      "learning_rate": 9.99995155746345e-06,
      "loss": 0.233,
      "step": 439
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7248384283309917,
      "learning_rate": 9.999946324068588e-06,
      "loss": 0.2332,
      "step": 440
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6883591048498006,
      "learning_rate": 9.99994082229647e-06,
      "loss": 0.2564,
      "step": 441
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7302591438345385,
      "learning_rate": 9.999935052147393e-06,
      "loss": 0.2302,
      "step": 442
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.544737991200359,
      "learning_rate": 9.999929013621667e-06,
      "loss": 0.6525,
      "step": 443
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.71751612570369,
      "learning_rate": 9.999922706719614e-06,
      "loss": 0.5309,
      "step": 444
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6018585595110737,
      "learning_rate": 9.999916131441574e-06,
      "loss": 0.1775,
      "step": 445
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6255304380107483,
      "learning_rate": 9.9999092877879e-06,
      "loss": 0.2079,
      "step": 446
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8395138268286515,
      "learning_rate": 9.99990217575896e-06,
      "loss": 0.2247,
      "step": 447
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.203220014861547,
      "learning_rate": 9.999894795355133e-06,
      "loss": 0.9359,
      "step": 448
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.091457747147549,
      "learning_rate": 9.999887146576817e-06,
      "loss": 0.5447,
      "step": 449
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9360642516162143,
      "learning_rate": 9.999879229424423e-06,
      "loss": 0.2391,
      "step": 450
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6517621913320264,
      "learning_rate": 9.999871043898377e-06,
      "loss": 0.2192,
      "step": 451
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7488625226819485,
      "learning_rate": 9.999862589999114e-06,
      "loss": 0.2259,
      "step": 452
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7034760495762693,
      "learning_rate": 9.999853867727092e-06,
      "loss": 0.3037,
      "step": 453
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.996622442113789,
      "learning_rate": 9.999844877082778e-06,
      "loss": 0.6067,
      "step": 454
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.845993104016002,
      "learning_rate": 9.999835618066654e-06,
      "loss": 0.2983,
      "step": 455
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4696482241127788,
      "learning_rate": 9.999826090679217e-06,
      "loss": 0.1791,
      "step": 456
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6490012695861709,
      "learning_rate": 9.99981629492098e-06,
      "loss": 0.2605,
      "step": 457
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.147175952765007,
      "learning_rate": 9.999806230792467e-06,
      "loss": 0.6311,
      "step": 458
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5997205298571935,
      "learning_rate": 9.999795898294217e-06,
      "loss": 0.2484,
      "step": 459
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6563332814372878,
      "learning_rate": 9.999785297426788e-06,
      "loss": 0.2163,
      "step": 460
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.604871893843957,
      "learning_rate": 9.999774428190748e-06,
      "loss": 0.2192,
      "step": 461
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8108955797125175,
      "learning_rate": 9.999763290586678e-06,
      "loss": 0.2634,
      "step": 462
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8090156663520274,
      "learning_rate": 9.999751884615179e-06,
      "loss": 0.2843,
      "step": 463
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.712980821493254,
      "learning_rate": 9.99974021027686e-06,
      "loss": 0.2689,
      "step": 464
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.802867706232354,
      "learning_rate": 9.99972826757235e-06,
      "loss": 0.2138,
      "step": 465
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7272535778282685,
      "learning_rate": 9.99971605650229e-06,
      "loss": 0.1555,
      "step": 466
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.809783852560339,
      "learning_rate": 9.999703577067335e-06,
      "loss": 0.2269,
      "step": 467
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.375224264584951,
      "learning_rate": 9.999690829268154e-06,
      "loss": 0.2804,
      "step": 468
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.741397257334788,
      "learning_rate": 9.999677813105433e-06,
      "loss": 0.2364,
      "step": 469
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7484815009682,
      "learning_rate": 9.999664528579869e-06,
      "loss": 0.203,
      "step": 470
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7966171894175609,
      "learning_rate": 9.999650975692177e-06,
      "loss": 0.2711,
      "step": 471
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8129529815924725,
      "learning_rate": 9.999637154443081e-06,
      "loss": 0.2353,
      "step": 472
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.758591110114017,
      "learning_rate": 9.999623064833326e-06,
      "loss": 0.2811,
      "step": 473
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5950412617261114,
      "learning_rate": 9.999608706863669e-06,
      "loss": 0.2259,
      "step": 474
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7918947169648147,
      "learning_rate": 9.999594080534876e-06,
      "loss": 0.2721,
      "step": 475
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5977561127411257,
      "learning_rate": 9.999579185847737e-06,
      "loss": 0.1673,
      "step": 476
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9241096196453091,
      "learning_rate": 9.999564022803048e-06,
      "loss": 0.3209,
      "step": 477
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5925113761025154,
      "learning_rate": 9.999548591401627e-06,
      "loss": 0.2161,
      "step": 478
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6444947157558782,
      "learning_rate": 9.999532891644298e-06,
      "loss": 0.2278,
      "step": 479
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6299073963546644,
      "learning_rate": 9.999516923531906e-06,
      "loss": 0.2053,
      "step": 480
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.766912757144763,
      "learning_rate": 9.999500687065306e-06,
      "loss": 0.2867,
      "step": 481
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6028071422876382,
      "learning_rate": 9.999484182245374e-06,
      "loss": 0.2154,
      "step": 482
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.341953397901124,
      "learning_rate": 9.999467409072991e-06,
      "loss": 0.5632,
      "step": 483
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5975418883221923,
      "learning_rate": 9.99945036754906e-06,
      "loss": 0.2249,
      "step": 484
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.725171549076565,
      "learning_rate": 9.999433057674495e-06,
      "loss": 0.2238,
      "step": 485
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7095499925260276,
      "learning_rate": 9.999415479450225e-06,
      "loss": 0.2416,
      "step": 486
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6737099249111227,
      "learning_rate": 9.999397632877193e-06,
      "loss": 0.1997,
      "step": 487
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.828428651517357,
      "learning_rate": 9.999379517956358e-06,
      "loss": 0.2025,
      "step": 488
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.672292412747221,
      "learning_rate": 9.999361134688694e-06,
      "loss": 0.2372,
      "step": 489
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8226958423104032,
      "learning_rate": 9.999342483075183e-06,
      "loss": 0.2134,
      "step": 490
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6337115996049818,
      "learning_rate": 9.99932356311683e-06,
      "loss": 0.2278,
      "step": 491
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8136645780795486,
      "learning_rate": 9.99930437481465e-06,
      "loss": 0.231,
      "step": 492
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6860539122480611,
      "learning_rate": 9.999284918169672e-06,
      "loss": 0.2384,
      "step": 493
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9109072690344266,
      "learning_rate": 9.999265193182938e-06,
      "loss": 0.2441,
      "step": 494
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6642202134097952,
      "learning_rate": 9.999245199855514e-06,
      "loss": 0.1918,
      "step": 495
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7254583544187543,
      "learning_rate": 9.999224938188466e-06,
      "loss": 0.2548,
      "step": 496
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.825079509601027,
      "learning_rate": 9.999204408182884e-06,
      "loss": 0.1956,
      "step": 497
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6797037764194405,
      "learning_rate": 9.99918360983987e-06,
      "loss": 0.1966,
      "step": 498
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8948119709599023,
      "learning_rate": 9.999162543160542e-06,
      "loss": 0.2585,
      "step": 499
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6886961037869472,
      "learning_rate": 9.999141208146029e-06,
      "loss": 0.2594,
      "step": 500
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7531131698865352,
      "learning_rate": 9.999119604797475e-06,
      "loss": 0.2305,
      "step": 501
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.785064566488959,
      "learning_rate": 9.999097733116042e-06,
      "loss": 0.1817,
      "step": 502
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7117784327730716,
      "learning_rate": 9.999075593102902e-06,
      "loss": 0.2505,
      "step": 503
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5942958873251942,
      "learning_rate": 9.999053184759247e-06,
      "loss": 0.2217,
      "step": 504
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6393439115054835,
      "learning_rate": 9.999030508086277e-06,
      "loss": 0.2063,
      "step": 505
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8830344378648498,
      "learning_rate": 9.999007563085205e-06,
      "loss": 0.2238,
      "step": 506
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6108881327048417,
      "learning_rate": 9.998984349757273e-06,
      "loss": 0.217,
      "step": 507
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.801066230572164,
      "learning_rate": 9.998960868103718e-06,
      "loss": 0.2626,
      "step": 508
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.199414809428491,
      "learning_rate": 9.998937118125802e-06,
      "loss": 0.5115,
      "step": 509
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5682770747611556,
      "learning_rate": 9.998913099824804e-06,
      "loss": 0.2126,
      "step": 510
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.875160179744944,
      "learning_rate": 9.998888813202008e-06,
      "loss": 0.2203,
      "step": 511
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.035463424478484,
      "learning_rate": 9.998864258258723e-06,
      "loss": 0.8033,
      "step": 512
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.568685121215722,
      "learning_rate": 9.998839434996262e-06,
      "loss": 0.2246,
      "step": 513
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5569014534674326,
      "learning_rate": 9.99881434341596e-06,
      "loss": 0.19,
      "step": 514
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9197100667105194,
      "learning_rate": 9.998788983519163e-06,
      "loss": 0.2458,
      "step": 515
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5768354593340916,
      "learning_rate": 9.998763355307232e-06,
      "loss": 0.2166,
      "step": 516
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6643234251109573,
      "learning_rate": 9.998737458781543e-06,
      "loss": 0.2073,
      "step": 517
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.685017082410479,
      "learning_rate": 9.998711293943487e-06,
      "loss": 0.5965,
      "step": 518
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7205352288266864,
      "learning_rate": 9.998684860794467e-06,
      "loss": 0.2376,
      "step": 519
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6077792074408113,
      "learning_rate": 9.998658159335903e-06,
      "loss": 0.2293,
      "step": 520
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8361190668365253,
      "learning_rate": 9.998631189569227e-06,
      "loss": 0.2702,
      "step": 521
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7618060906306372,
      "learning_rate": 9.998603951495889e-06,
      "loss": 0.2485,
      "step": 522
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4370266035532995,
      "learning_rate": 9.998576445117347e-06,
      "loss": 0.2043,
      "step": 523
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.729653395178178,
      "learning_rate": 9.998548670435083e-06,
      "loss": 0.7933,
      "step": 524
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.735281723583391,
      "learning_rate": 9.998520627450581e-06,
      "loss": 0.2206,
      "step": 525
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.4277089511375,
      "learning_rate": 9.998492316165352e-06,
      "loss": 0.7456,
      "step": 526
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7131942904930606,
      "learning_rate": 9.998463736580911e-06,
      "loss": 0.2047,
      "step": 527
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5834872257368824,
      "learning_rate": 9.998434888698796e-06,
      "loss": 0.2193,
      "step": 528
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9159050691937483,
      "learning_rate": 9.998405772520554e-06,
      "loss": 0.2162,
      "step": 529
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6620470306872097,
      "learning_rate": 9.998376388047748e-06,
      "loss": 0.2658,
      "step": 530
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8166883799793274,
      "learning_rate": 9.998346735281954e-06,
      "loss": 0.2654,
      "step": 531
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.765199078359851,
      "learning_rate": 9.998316814224765e-06,
      "loss": 0.6544,
      "step": 532
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7793716730033806,
      "learning_rate": 9.998286624877786e-06,
      "loss": 0.2085,
      "step": 533
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7968453636194168,
      "learning_rate": 9.99825616724264e-06,
      "loss": 0.2744,
      "step": 534
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.601402270695304,
      "learning_rate": 9.998225441320959e-06,
      "loss": 0.276,
      "step": 535
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7769052610542935,
      "learning_rate": 9.998194447114394e-06,
      "loss": 0.2315,
      "step": 536
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.635524613892197,
      "learning_rate": 9.998163184624606e-06,
      "loss": 0.2379,
      "step": 537
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6589381231401419,
      "learning_rate": 9.998131653853275e-06,
      "loss": 0.3002,
      "step": 538
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6155176067494363,
      "learning_rate": 9.998099854802095e-06,
      "loss": 0.2093,
      "step": 539
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7763240732316072,
      "learning_rate": 9.998067787472772e-06,
      "loss": 0.241,
      "step": 540
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8068331023253885,
      "learning_rate": 9.998035451867023e-06,
      "loss": 0.2371,
      "step": 541
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4999739203668576,
      "learning_rate": 9.99800284798659e-06,
      "loss": 0.2645,
      "step": 542
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7684112199228414,
      "learning_rate": 9.99796997583322e-06,
      "loss": 0.2248,
      "step": 543
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7109024922463631,
      "learning_rate": 9.997936835408675e-06,
      "loss": 0.2392,
      "step": 544
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5467644837186367,
      "learning_rate": 9.997903426714739e-06,
      "loss": 0.1684,
      "step": 545
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.545073681962836,
      "learning_rate": 9.997869749753202e-06,
      "loss": 0.2637,
      "step": 546
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.730639439800308,
      "learning_rate": 9.997835804525871e-06,
      "loss": 0.2764,
      "step": 547
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.620402046346925,
      "learning_rate": 9.99780159103457e-06,
      "loss": 0.2087,
      "step": 548
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7907960508042755,
      "learning_rate": 9.997767109281136e-06,
      "loss": 0.2523,
      "step": 549
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8005056652493097,
      "learning_rate": 9.997732359267419e-06,
      "loss": 0.3039,
      "step": 550
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6639212189483445,
      "learning_rate": 9.99769734099528e-06,
      "loss": 0.2458,
      "step": 551
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5738592148956088,
      "learning_rate": 9.997662054466607e-06,
      "loss": 0.2707,
      "step": 552
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6967595197664154,
      "learning_rate": 9.997626499683289e-06,
      "loss": 0.2427,
      "step": 553
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7570547044433409,
      "learning_rate": 9.997590676647233e-06,
      "loss": 0.2612,
      "step": 554
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.889790503370837,
      "learning_rate": 9.997554585360365e-06,
      "loss": 0.2475,
      "step": 555
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6774943962490119,
      "learning_rate": 9.997518225824621e-06,
      "loss": 0.2154,
      "step": 556
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8423758891610968,
      "learning_rate": 9.997481598041951e-06,
      "loss": 0.2919,
      "step": 557
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.377229010501095,
      "learning_rate": 9.997444702014326e-06,
      "loss": 0.6379,
      "step": 558
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6616479182827921,
      "learning_rate": 9.997407537743721e-06,
      "loss": 0.2243,
      "step": 559
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9841749170287015,
      "learning_rate": 9.997370105232134e-06,
      "loss": 0.2444,
      "step": 560
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9656193443421748,
      "learning_rate": 9.997332404481572e-06,
      "loss": 0.2968,
      "step": 561
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.471127362443591,
      "learning_rate": 9.99729443549406e-06,
      "loss": 0.2206,
      "step": 562
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5650716830935516,
      "learning_rate": 9.997256198271637e-06,
      "loss": 0.2013,
      "step": 563
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8484446399470826,
      "learning_rate": 9.997217692816355e-06,
      "loss": 0.2353,
      "step": 564
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5426262363806842,
      "learning_rate": 9.997178919130279e-06,
      "loss": 0.2006,
      "step": 565
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5623041578803294,
      "learning_rate": 9.997139877215492e-06,
      "loss": 0.2223,
      "step": 566
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.803744305102567,
      "learning_rate": 9.997100567074087e-06,
      "loss": 0.2532,
      "step": 567
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9311619338752148,
      "learning_rate": 9.997060988708178e-06,
      "loss": 0.2448,
      "step": 568
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8669588569798805,
      "learning_rate": 9.997021142119886e-06,
      "loss": 0.2512,
      "step": 569
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7416672147329488,
      "learning_rate": 9.996981027311352e-06,
      "loss": 0.2265,
      "step": 570
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4749202237317114,
      "learning_rate": 9.996940644284729e-06,
      "loss": 0.1873,
      "step": 571
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5578129264883946,
      "learning_rate": 9.996899993042182e-06,
      "loss": 0.2042,
      "step": 572
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6221564193193698,
      "learning_rate": 9.996859073585898e-06,
      "loss": 0.2117,
      "step": 573
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.678190107536808,
      "learning_rate": 9.996817885918069e-06,
      "loss": 0.2142,
      "step": 574
    },
    {
      "epoch": 0.04,
      "grad_norm": 11.29968171973042,
      "learning_rate": 9.996776430040906e-06,
      "loss": 0.6952,
      "step": 575
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4937877717830532,
      "learning_rate": 9.996734705956637e-06,
      "loss": 0.2218,
      "step": 576
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5677510808090842,
      "learning_rate": 9.996692713667498e-06,
      "loss": 0.2404,
      "step": 577
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6960950245606041,
      "learning_rate": 9.996650453175748e-06,
      "loss": 0.2073,
      "step": 578
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6029173822650022,
      "learning_rate": 9.99660792448365e-06,
      "loss": 0.2541,
      "step": 579
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6487313257896015,
      "learning_rate": 9.99656512759349e-06,
      "loss": 0.2577,
      "step": 580
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5680722873315158,
      "learning_rate": 9.996522062507564e-06,
      "loss": 0.204,
      "step": 581
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4884603105729501,
      "learning_rate": 9.996478729228182e-06,
      "loss": 0.228,
      "step": 582
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.585192741789178,
      "learning_rate": 9.996435127757673e-06,
      "loss": 0.1913,
      "step": 583
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6580058137613347,
      "learning_rate": 9.996391258098376e-06,
      "loss": 0.2408,
      "step": 584
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.2374068141697725,
      "learning_rate": 9.996347120252647e-06,
      "loss": 0.5011,
      "step": 585
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5613579859542268,
      "learning_rate": 9.996302714222853e-06,
      "loss": 0.2147,
      "step": 586
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6917879179396091,
      "learning_rate": 9.996258040011381e-06,
      "loss": 0.2214,
      "step": 587
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.74731716420905,
      "learning_rate": 9.996213097620623e-06,
      "loss": 0.2718,
      "step": 588
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5626750637003448,
      "learning_rate": 9.996167887052997e-06,
      "loss": 0.23,
      "step": 589
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6178635707855193,
      "learning_rate": 9.996122408310927e-06,
      "loss": 0.2523,
      "step": 590
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0111752792353403,
      "learning_rate": 9.996076661396854e-06,
      "loss": 0.2675,
      "step": 591
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.892157813613588,
      "learning_rate": 9.996030646313235e-06,
      "loss": 0.5959,
      "step": 592
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.80468775181741,
      "learning_rate": 9.995984363062538e-06,
      "loss": 0.2857,
      "step": 593
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7772367829113869,
      "learning_rate": 9.995937811647251e-06,
      "loss": 0.2688,
      "step": 594
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.581110070462405,
      "learning_rate": 9.995890992069868e-06,
      "loss": 0.2298,
      "step": 595
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5681972353664881,
      "learning_rate": 9.995843904332904e-06,
      "loss": 0.221,
      "step": 596
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6695363309663607,
      "learning_rate": 9.995796548438887e-06,
      "loss": 0.2676,
      "step": 597
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.665844879938811,
      "learning_rate": 9.995748924390358e-06,
      "loss": 0.2166,
      "step": 598
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.460510946143721,
      "learning_rate": 9.995701032189875e-06,
      "loss": 0.1801,
      "step": 599
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6493714929143417,
      "learning_rate": 9.995652871840006e-06,
      "loss": 0.2535,
      "step": 600
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7755574796736042,
      "learning_rate": 9.99560444334334e-06,
      "loss": 0.232,
      "step": 601
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.124709629697863,
      "learning_rate": 9.995555746702472e-06,
      "loss": 0.5456,
      "step": 602
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8831853454619656,
      "learning_rate": 9.995506781920018e-06,
      "loss": 0.2046,
      "step": 603
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7654722033428032,
      "learning_rate": 9.995457548998606e-06,
      "loss": 0.1914,
      "step": 604
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4581466027526737,
      "learning_rate": 9.99540804794088e-06,
      "loss": 0.233,
      "step": 605
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6869813631526638,
      "learning_rate": 9.995358278749494e-06,
      "loss": 0.243,
      "step": 606
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5826420172912294,
      "learning_rate": 9.995308241427122e-06,
      "loss": 0.267,
      "step": 607
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4567906792683805,
      "learning_rate": 9.995257935976448e-06,
      "loss": 0.1924,
      "step": 608
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8330309990573586,
      "learning_rate": 9.995207362400175e-06,
      "loss": 0.2692,
      "step": 609
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5430968381143486,
      "learning_rate": 9.995156520701012e-06,
      "loss": 0.1932,
      "step": 610
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9428066567116653,
      "learning_rate": 9.995105410881695e-06,
      "loss": 0.2585,
      "step": 611
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6630153503418232,
      "learning_rate": 9.995054032944963e-06,
      "loss": 0.2232,
      "step": 612
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.334326065352627,
      "learning_rate": 9.995002386893575e-06,
      "loss": 0.533,
      "step": 613
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.796800976591682,
      "learning_rate": 9.994950472730302e-06,
      "loss": 0.2185,
      "step": 614
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6366769835867134,
      "learning_rate": 9.994898290457932e-06,
      "loss": 0.243,
      "step": 615
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.828949240894937,
      "learning_rate": 9.994845840079267e-06,
      "loss": 0.2471,
      "step": 616
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6972468594596486,
      "learning_rate": 9.99479312159712e-06,
      "loss": 0.2647,
      "step": 617
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.490134572545603,
      "learning_rate": 9.99474013501432e-06,
      "loss": 0.6792,
      "step": 618
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5263151793306773,
      "learning_rate": 9.994686880333715e-06,
      "loss": 0.2045,
      "step": 619
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6258968809448344,
      "learning_rate": 9.994633357558158e-06,
      "loss": 0.2969,
      "step": 620
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6345802243864846,
      "learning_rate": 9.994579566690527e-06,
      "loss": 0.3127,
      "step": 621
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.617938165595077,
      "learning_rate": 9.994525507733708e-06,
      "loss": 0.2179,
      "step": 622
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5204451581647924,
      "learning_rate": 9.994471180690601e-06,
      "loss": 0.216,
      "step": 623
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5893303891887063,
      "learning_rate": 9.994416585564123e-06,
      "loss": 0.2505,
      "step": 624
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.665748344611308,
      "learning_rate": 9.994361722357204e-06,
      "loss": 0.2014,
      "step": 625
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.701428904090509,
      "learning_rate": 9.99430659107279e-06,
      "loss": 0.8376,
      "step": 626
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.569425936531288,
      "learning_rate": 9.99425119171384e-06,
      "loss": 0.6867,
      "step": 627
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7785769642846632,
      "learning_rate": 9.994195524283326e-06,
      "loss": 0.2215,
      "step": 628
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7001487048566615,
      "learning_rate": 9.994139588784238e-06,
      "loss": 0.2186,
      "step": 629
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8228283775986551,
      "learning_rate": 9.994083385219578e-06,
      "loss": 0.2557,
      "step": 630
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6706857871472391,
      "learning_rate": 9.99402691359236e-06,
      "loss": 0.2328,
      "step": 631
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5287312505615744,
      "learning_rate": 9.993970173905618e-06,
      "loss": 0.1659,
      "step": 632
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8023120839277955,
      "learning_rate": 9.993913166162399e-06,
      "loss": 0.2495,
      "step": 633
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8599250552416529,
      "learning_rate": 9.99385589036576e-06,
      "loss": 0.2374,
      "step": 634
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5766608007380987,
      "learning_rate": 9.993798346518777e-06,
      "loss": 0.2547,
      "step": 635
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.447632684960474,
      "learning_rate": 9.993740534624536e-06,
      "loss": 0.6147,
      "step": 636
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6379772903849688,
      "learning_rate": 9.993682454686144e-06,
      "loss": 0.2003,
      "step": 637
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9764683541501522,
      "learning_rate": 9.993624106706716e-06,
      "loss": 0.286,
      "step": 638
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6008619888219222,
      "learning_rate": 9.993565490689386e-06,
      "loss": 0.2571,
      "step": 639
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8045849937206637,
      "learning_rate": 9.993506606637297e-06,
      "loss": 0.2271,
      "step": 640
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.648400479548625,
      "learning_rate": 9.993447454553611e-06,
      "loss": 0.2765,
      "step": 641
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5451627165196342,
      "learning_rate": 9.993388034441505e-06,
      "loss": 0.2282,
      "step": 642
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6308379584056347,
      "learning_rate": 9.993328346304167e-06,
      "loss": 0.1996,
      "step": 643
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8580692733407562,
      "learning_rate": 9.9932683901448e-06,
      "loss": 0.2853,
      "step": 644
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7290502849348282,
      "learning_rate": 9.993208165966624e-06,
      "loss": 0.21,
      "step": 645
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4724880569233778,
      "learning_rate": 9.993147673772869e-06,
      "loss": 0.1773,
      "step": 646
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6977919819596643,
      "learning_rate": 9.993086913566785e-06,
      "loss": 0.1963,
      "step": 647
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6005949693978905,
      "learning_rate": 9.993025885351633e-06,
      "loss": 0.2643,
      "step": 648
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6977853432026133,
      "learning_rate": 9.992964589130685e-06,
      "loss": 0.2279,
      "step": 649
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6921275344161306,
      "learning_rate": 9.992903024907236e-06,
      "loss": 0.2101,
      "step": 650
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7501982248183987,
      "learning_rate": 9.992841192684589e-06,
      "loss": 0.253,
      "step": 651
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.517808439251331,
      "learning_rate": 9.99277909246606e-06,
      "loss": 0.594,
      "step": 652
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7139765120713024,
      "learning_rate": 9.992716724254985e-06,
      "loss": 0.2521,
      "step": 653
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.962550950443222,
      "learning_rate": 9.992654088054712e-06,
      "loss": 0.7427,
      "step": 654
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.773620290910626,
      "learning_rate": 9.992591183868602e-06,
      "loss": 0.2527,
      "step": 655
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6745668095849577,
      "learning_rate": 9.992528011700032e-06,
      "loss": 0.2472,
      "step": 656
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7151151980724444,
      "learning_rate": 9.99246457155239e-06,
      "loss": 0.2186,
      "step": 657
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7838451348370998,
      "learning_rate": 9.992400863429089e-06,
      "loss": 0.1938,
      "step": 658
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7456572374752544,
      "learning_rate": 9.992336887333541e-06,
      "loss": 0.2365,
      "step": 659
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8222414322056386,
      "learning_rate": 9.992272643269181e-06,
      "loss": 0.2185,
      "step": 660
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4609742655907418,
      "learning_rate": 9.99220813123946e-06,
      "loss": 0.176,
      "step": 661
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4997844934791071,
      "learning_rate": 9.992143351247839e-06,
      "loss": 0.2035,
      "step": 662
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.407079689251482,
      "learning_rate": 9.992078303297796e-06,
      "loss": 0.179,
      "step": 663
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6303614474181178,
      "learning_rate": 9.992012987392823e-06,
      "loss": 0.2523,
      "step": 664
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7768269793314064,
      "learning_rate": 9.991947403536424e-06,
      "loss": 0.2765,
      "step": 665
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7857421304152366,
      "learning_rate": 9.991881551732118e-06,
      "loss": 0.2966,
      "step": 666
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.79458873884545,
      "learning_rate": 9.991815431983446e-06,
      "loss": 0.6927,
      "step": 667
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.536120814698744,
      "learning_rate": 9.991749044293952e-06,
      "loss": 0.2026,
      "step": 668
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5791806604568266,
      "learning_rate": 9.991682388667199e-06,
      "loss": 0.2257,
      "step": 669
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.573932588629379,
      "learning_rate": 9.991615465106768e-06,
      "loss": 0.2142,
      "step": 670
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5744528996608034,
      "learning_rate": 9.991548273616249e-06,
      "loss": 0.2764,
      "step": 671
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6251326334329739,
      "learning_rate": 9.991480814199248e-06,
      "loss": 0.2144,
      "step": 672
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6384848773967653,
      "learning_rate": 9.991413086859388e-06,
      "loss": 0.212,
      "step": 673
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.707350982974174,
      "learning_rate": 9.991345091600303e-06,
      "loss": 0.2168,
      "step": 674
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6588311096027324,
      "learning_rate": 9.991276828425642e-06,
      "loss": 0.248,
      "step": 675
    },
    {
      "epoch": 0.05,
      "grad_norm": 17.06273966083748,
      "learning_rate": 9.991208297339072e-06,
      "loss": 0.6826,
      "step": 676
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6172737021030636,
      "learning_rate": 9.991139498344269e-06,
      "loss": 0.2693,
      "step": 677
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6406289137232704,
      "learning_rate": 9.991070431444927e-06,
      "loss": 0.2541,
      "step": 678
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4472254286293658,
      "learning_rate": 9.991001096644752e-06,
      "loss": 0.1982,
      "step": 679
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7086115832780626,
      "learning_rate": 9.990931493947467e-06,
      "loss": 0.2742,
      "step": 680
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.637502307686957,
      "learning_rate": 9.990861623356806e-06,
      "loss": 0.2779,
      "step": 681
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.935092125493366,
      "learning_rate": 9.990791484876521e-06,
      "loss": 0.2713,
      "step": 682
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5825049620749547,
      "learning_rate": 9.990721078510378e-06,
      "loss": 0.1946,
      "step": 683
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8895524878943704,
      "learning_rate": 9.990650404262152e-06,
      "loss": 0.2882,
      "step": 684
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7099982114694339,
      "learning_rate": 9.990579462135641e-06,
      "loss": 0.2186,
      "step": 685
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.766183409918529,
      "learning_rate": 9.990508252134651e-06,
      "loss": 0.1799,
      "step": 686
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5133854499847073,
      "learning_rate": 9.990436774263004e-06,
      "loss": 0.2567,
      "step": 687
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7683432324267765,
      "learning_rate": 9.990365028524535e-06,
      "loss": 0.2489,
      "step": 688
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4618691871535132,
      "learning_rate": 9.9902930149231e-06,
      "loss": 0.1862,
      "step": 689
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5048617126919146,
      "learning_rate": 9.990220733462559e-06,
      "loss": 0.2025,
      "step": 690
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6063300636267117,
      "learning_rate": 9.990148184146795e-06,
      "loss": 0.1922,
      "step": 691
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5346977645706932,
      "learning_rate": 9.990075366979702e-06,
      "loss": 0.2207,
      "step": 692
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.570164765297416,
      "learning_rate": 9.990002281965187e-06,
      "loss": 0.7797,
      "step": 693
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6102851763440187,
      "learning_rate": 9.989928929107174e-06,
      "loss": 0.2384,
      "step": 694
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5115832870752945,
      "learning_rate": 9.989855308409601e-06,
      "loss": 0.197,
      "step": 695
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.3942936973553355,
      "learning_rate": 9.989781419876416e-06,
      "loss": 0.6406,
      "step": 696
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3234101586603415,
      "learning_rate": 9.98970726351159e-06,
      "loss": 0.1828,
      "step": 697
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6838225515556275,
      "learning_rate": 9.9896328393191e-06,
      "loss": 0.2348,
      "step": 698
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7439114767871668,
      "learning_rate": 9.989558147302943e-06,
      "loss": 0.2174,
      "step": 699
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.54608858355113,
      "learning_rate": 9.989483187467128e-06,
      "loss": 0.2556,
      "step": 700
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6389923853748325,
      "learning_rate": 9.989407959815676e-06,
      "loss": 0.2099,
      "step": 701
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6648925661720466,
      "learning_rate": 9.989332464352625e-06,
      "loss": 0.2089,
      "step": 702
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.438464988623418,
      "learning_rate": 9.989256701082032e-06,
      "loss": 0.6251,
      "step": 703
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7124364911399221,
      "learning_rate": 9.989180670007958e-06,
      "loss": 0.2474,
      "step": 704
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.777433049900891,
      "learning_rate": 9.989104371134489e-06,
      "loss": 0.2134,
      "step": 705
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.571749507479074,
      "learning_rate": 9.989027804465716e-06,
      "loss": 0.2628,
      "step": 706
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7634914746136903,
      "learning_rate": 9.988950970005753e-06,
      "loss": 0.2337,
      "step": 707
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.88274237620404,
      "learning_rate": 9.98887386775872e-06,
      "loss": 0.6121,
      "step": 708
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5456658966513586,
      "learning_rate": 9.98879649772876e-06,
      "loss": 0.193,
      "step": 709
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5696296058011945,
      "learning_rate": 9.98871885992002e-06,
      "loss": 0.224,
      "step": 710
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8070220959594163,
      "learning_rate": 9.988640954336672e-06,
      "loss": 0.249,
      "step": 711
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.223432795464063,
      "learning_rate": 9.988562780982898e-06,
      "loss": 0.2749,
      "step": 712
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6042503033623325,
      "learning_rate": 9.98848433986289e-06,
      "loss": 0.2133,
      "step": 713
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9881422444036783,
      "learning_rate": 9.988405630980863e-06,
      "loss": 0.2854,
      "step": 714
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6171597287041666,
      "learning_rate": 9.98832665434104e-06,
      "loss": 0.2663,
      "step": 715
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5171023362890572,
      "learning_rate": 9.988247409947657e-06,
      "loss": 0.2471,
      "step": 716
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6010564818425659,
      "learning_rate": 9.988167897804971e-06,
      "loss": 0.23,
      "step": 717
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6810768730751677,
      "learning_rate": 9.98808811791725e-06,
      "loss": 0.2866,
      "step": 718
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4668506914852013,
      "learning_rate": 9.988008070288774e-06,
      "loss": 0.2071,
      "step": 719
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7820393576147726,
      "learning_rate": 9.987927754923844e-06,
      "loss": 0.2329,
      "step": 720
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6135470769069782,
      "learning_rate": 9.987847171826767e-06,
      "loss": 0.2276,
      "step": 721
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5251872921699774,
      "learning_rate": 9.987766321001868e-06,
      "loss": 0.211,
      "step": 722
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5780578362093696,
      "learning_rate": 9.98768520245349e-06,
      "loss": 0.2466,
      "step": 723
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4645401624370016,
      "learning_rate": 9.987603816185985e-06,
      "loss": 0.2459,
      "step": 724
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6685975345119328,
      "learning_rate": 9.98752216220372e-06,
      "loss": 0.2286,
      "step": 725
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4899719769996609,
      "learning_rate": 9.987440240511082e-06,
      "loss": 0.2323,
      "step": 726
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5610374154455227,
      "learning_rate": 9.987358051112465e-06,
      "loss": 0.2087,
      "step": 727
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7222604084697974,
      "learning_rate": 9.987275594012283e-06,
      "loss": 0.2576,
      "step": 728
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5177727573154718,
      "learning_rate": 9.987192869214958e-06,
      "loss": 0.1773,
      "step": 729
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6254157786608658,
      "learning_rate": 9.987109876724934e-06,
      "loss": 0.2268,
      "step": 730
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8479037625509187,
      "learning_rate": 9.987026616546665e-06,
      "loss": 0.2641,
      "step": 731
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5998119854604769,
      "learning_rate": 9.986943088684619e-06,
      "loss": 0.2244,
      "step": 732
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7630506131543133,
      "learning_rate": 9.986859293143279e-06,
      "loss": 0.2645,
      "step": 733
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.486017086475076,
      "learning_rate": 9.986775229927147e-06,
      "loss": 0.5149,
      "step": 734
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6643683239221327,
      "learning_rate": 9.98669089904073e-06,
      "loss": 0.2519,
      "step": 735
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6954390164325874,
      "learning_rate": 9.986606300488555e-06,
      "loss": 0.2187,
      "step": 736
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6818453456134628,
      "learning_rate": 9.986521434275167e-06,
      "loss": 0.2722,
      "step": 737
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6359009320457223,
      "learning_rate": 9.986436300405119e-06,
      "loss": 0.2121,
      "step": 738
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5059629148004405,
      "learning_rate": 9.98635089888298e-06,
      "loss": 0.2181,
      "step": 739
    },
    {
      "epoch": 0.05,
      "grad_norm": 21.529067341605003,
      "learning_rate": 9.986265229713332e-06,
      "loss": 0.7089,
      "step": 740
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.49801488114809,
      "learning_rate": 9.98617929290078e-06,
      "loss": 0.2113,
      "step": 741
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6178266851261276,
      "learning_rate": 9.986093088449928e-06,
      "loss": 0.1937,
      "step": 742
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5300996288740902,
      "learning_rate": 9.986006616365412e-06,
      "loss": 0.2046,
      "step": 743
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.793540317769164,
      "learning_rate": 9.985919876651866e-06,
      "loss": 0.2715,
      "step": 744
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.490481328209008,
      "learning_rate": 9.985832869313952e-06,
      "loss": 0.202,
      "step": 745
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8506943491030012,
      "learning_rate": 9.985745594356336e-06,
      "loss": 0.2809,
      "step": 746
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6763576622228822,
      "learning_rate": 9.985658051783701e-06,
      "loss": 0.1876,
      "step": 747
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7361491326028007,
      "learning_rate": 9.985570241600754e-06,
      "loss": 0.2731,
      "step": 748
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6595152747163822,
      "learning_rate": 9.9854821638122e-06,
      "loss": 0.2224,
      "step": 749
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8806129727918472,
      "learning_rate": 9.985393818422768e-06,
      "loss": 0.2431,
      "step": 750
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6534792721881997,
      "learning_rate": 9.985305205437205e-06,
      "loss": 0.2484,
      "step": 751
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.600172218807455,
      "learning_rate": 9.985216324860264e-06,
      "loss": 0.7512,
      "step": 752
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.768667856574418,
      "learning_rate": 9.985127176696713e-06,
      "loss": 0.2521,
      "step": 753
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.590104236346727,
      "learning_rate": 9.985037760951342e-06,
      "loss": 0.1812,
      "step": 754
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5420818366462943,
      "learning_rate": 9.984948077628948e-06,
      "loss": 0.2207,
      "step": 755
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5362526188179724,
      "learning_rate": 9.984858126734345e-06,
      "loss": 0.2865,
      "step": 756
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5819938234825788,
      "learning_rate": 9.984767908272363e-06,
      "loss": 0.1727,
      "step": 757
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5092683006036836,
      "learning_rate": 9.984677422247842e-06,
      "loss": 0.2446,
      "step": 758
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5961083910100657,
      "learning_rate": 9.984586668665641e-06,
      "loss": 0.2195,
      "step": 759
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.731370683117459,
      "learning_rate": 9.98449564753063e-06,
      "loss": 0.2654,
      "step": 760
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5923246566796907,
      "learning_rate": 9.984404358847695e-06,
      "loss": 0.2687,
      "step": 761
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6208198861905956,
      "learning_rate": 9.984312802621734e-06,
      "loss": 0.2096,
      "step": 762
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.62726720911841,
      "learning_rate": 9.984220978857665e-06,
      "loss": 0.2667,
      "step": 763
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6641816454125784,
      "learning_rate": 9.984128887560416e-06,
      "loss": 0.2185,
      "step": 764
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.498015379718007,
      "learning_rate": 9.984036528734928e-06,
      "loss": 0.6812,
      "step": 765
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.641383835517193,
      "learning_rate": 9.98394390238616e-06,
      "loss": 0.2267,
      "step": 766
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7406510567344653,
      "learning_rate": 9.983851008519083e-06,
      "loss": 0.2502,
      "step": 767
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6869258010557757,
      "learning_rate": 9.983757847138684e-06,
      "loss": 0.2197,
      "step": 768
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6790302875768113,
      "learning_rate": 9.983664418249964e-06,
      "loss": 0.2105,
      "step": 769
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.523423356938257,
      "learning_rate": 9.983570721857937e-06,
      "loss": 0.2151,
      "step": 770
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6781003019700231,
      "learning_rate": 9.983476757967632e-06,
      "loss": 0.2467,
      "step": 771
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5297832789332957,
      "learning_rate": 9.983382526584092e-06,
      "loss": 0.2244,
      "step": 772
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4803313421752438,
      "learning_rate": 9.983288027712377e-06,
      "loss": 0.2111,
      "step": 773
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5239553020266614,
      "learning_rate": 9.983193261357558e-06,
      "loss": 0.218,
      "step": 774
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5803720717550547,
      "learning_rate": 9.98309822752472e-06,
      "loss": 0.2012,
      "step": 775
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5977916552001226,
      "learning_rate": 9.983002926218969e-06,
      "loss": 0.2225,
      "step": 776
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.015422428924549,
      "learning_rate": 9.982907357445417e-06,
      "loss": 0.8359,
      "step": 777
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6134212217568487,
      "learning_rate": 9.982811521209192e-06,
      "loss": 0.2542,
      "step": 778
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4308179305800175,
      "learning_rate": 9.982715417515441e-06,
      "loss": 0.1969,
      "step": 779
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9238535652058442,
      "learning_rate": 9.982619046369321e-06,
      "loss": 0.2356,
      "step": 780
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4308037964444096,
      "learning_rate": 9.982522407776008e-06,
      "loss": 0.2318,
      "step": 781
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9275258585130022,
      "learning_rate": 9.982425501740684e-06,
      "loss": 0.209,
      "step": 782
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.090426309102259,
      "learning_rate": 9.982328328268556e-06,
      "loss": 0.7767,
      "step": 783
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7494064359877008,
      "learning_rate": 9.982230887364834e-06,
      "loss": 0.2402,
      "step": 784
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5801026043613933,
      "learning_rate": 9.982133179034753e-06,
      "loss": 0.2233,
      "step": 785
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7092706711367565,
      "learning_rate": 9.982035203283555e-06,
      "loss": 0.2323,
      "step": 786
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6849510886361898,
      "learning_rate": 9.981936960116501e-06,
      "loss": 0.2675,
      "step": 787
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4579634476147931,
      "learning_rate": 9.981838449538863e-06,
      "loss": 0.2056,
      "step": 788
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5941772188675964,
      "learning_rate": 9.981739671555928e-06,
      "loss": 0.221,
      "step": 789
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4654145850205578,
      "learning_rate": 9.981640626173e-06,
      "loss": 0.2026,
      "step": 790
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.5623872044185765,
      "learning_rate": 9.981541313395394e-06,
      "loss": 0.5875,
      "step": 791
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8404440212328006,
      "learning_rate": 9.98144173322844e-06,
      "loss": 0.2459,
      "step": 792
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.533952107562485,
      "learning_rate": 9.981341885677485e-06,
      "loss": 0.2468,
      "step": 793
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.754967270623259,
      "learning_rate": 9.981241770747885e-06,
      "loss": 0.3142,
      "step": 794
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.947918992862247,
      "learning_rate": 9.981141388445019e-06,
      "loss": 0.6977,
      "step": 795
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.115271351973183,
      "learning_rate": 9.981040738774272e-06,
      "loss": 0.8959,
      "step": 796
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5278633938690163,
      "learning_rate": 9.980939821741045e-06,
      "loss": 0.2186,
      "step": 797
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4439536237648316,
      "learning_rate": 9.980838637350758e-06,
      "loss": 0.227,
      "step": 798
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6124461572495545,
      "learning_rate": 9.98073718560884e-06,
      "loss": 0.2297,
      "step": 799
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5324257672166732,
      "learning_rate": 9.980635466520738e-06,
      "loss": 0.2657,
      "step": 800
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.655572465187251,
      "learning_rate": 9.98053348009191e-06,
      "loss": 0.2349,
      "step": 801
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.505102462473888,
      "learning_rate": 9.980431226327834e-06,
      "loss": 0.2116,
      "step": 802
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5851032602851012,
      "learning_rate": 9.980328705233992e-06,
      "loss": 0.2418,
      "step": 803
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5383959820494633,
      "learning_rate": 9.980225916815894e-06,
      "loss": 0.2207,
      "step": 804
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.66980817554449,
      "learning_rate": 9.980122861079054e-06,
      "loss": 0.2316,
      "step": 805
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7250264187671456,
      "learning_rate": 9.980019538029002e-06,
      "loss": 0.2288,
      "step": 806
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5515955076857662,
      "learning_rate": 9.979915947671288e-06,
      "loss": 0.2665,
      "step": 807
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.807998162350851,
      "learning_rate": 9.979812090011469e-06,
      "loss": 0.2638,
      "step": 808
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7707263240485176,
      "learning_rate": 9.979707965055121e-06,
      "loss": 0.2766,
      "step": 809
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7275708639041143,
      "learning_rate": 9.979603572807831e-06,
      "loss": 0.2208,
      "step": 810
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0457549708009677,
      "learning_rate": 9.979498913275206e-06,
      "loss": 0.2354,
      "step": 811
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7686453384057466,
      "learning_rate": 9.979393986462862e-06,
      "loss": 0.2213,
      "step": 812
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9739061175518136,
      "learning_rate": 9.979288792376429e-06,
      "loss": 0.2766,
      "step": 813
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4667542400930431,
      "learning_rate": 9.979183331021558e-06,
      "loss": 0.2769,
      "step": 814
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6148132786265412,
      "learning_rate": 9.979077602403906e-06,
      "loss": 0.2285,
      "step": 815
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.748139672996976,
      "learning_rate": 9.97897160652915e-06,
      "loss": 0.2853,
      "step": 816
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5530650282822267,
      "learning_rate": 9.978865343402976e-06,
      "loss": 0.235,
      "step": 817
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.501294170254973,
      "learning_rate": 9.978758813031092e-06,
      "loss": 0.2096,
      "step": 818
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4880783019869301,
      "learning_rate": 9.978652015419212e-06,
      "loss": 0.2337,
      "step": 819
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5775724956647952,
      "learning_rate": 9.978544950573075e-06,
      "loss": 0.2087,
      "step": 820
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7167820050080607,
      "learning_rate": 9.97843761849842e-06,
      "loss": 0.2853,
      "step": 821
    },
    {
      "epoch": 0.06,
      "grad_norm": 35.2705906826178,
      "learning_rate": 9.978330019201015e-06,
      "loss": 0.5936,
      "step": 822
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4571735422392338,
      "learning_rate": 9.978222152686632e-06,
      "loss": 0.1913,
      "step": 823
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.602937585278564,
      "learning_rate": 9.97811401896106e-06,
      "loss": 0.2362,
      "step": 824
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.603369690247898,
      "learning_rate": 9.978005618030104e-06,
      "loss": 0.2277,
      "step": 825
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7138086039905924,
      "learning_rate": 9.977896949899585e-06,
      "loss": 0.2277,
      "step": 826
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.650828522769254,
      "learning_rate": 9.977788014575333e-06,
      "loss": 0.2831,
      "step": 827
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5335644767490908,
      "learning_rate": 9.977678812063195e-06,
      "loss": 0.2538,
      "step": 828
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8352130086396239,
      "learning_rate": 9.977569342369034e-06,
      "loss": 0.2622,
      "step": 829
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.690886564264814,
      "learning_rate": 9.977459605498728e-06,
      "loss": 0.2822,
      "step": 830
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4422358251298906,
      "learning_rate": 9.977349601458161e-06,
      "loss": 0.1812,
      "step": 831
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0856979401922504,
      "learning_rate": 9.977239330253243e-06,
      "loss": 0.2573,
      "step": 832
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5934180670082112,
      "learning_rate": 9.977128791889892e-06,
      "loss": 0.1998,
      "step": 833
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.542659392723648,
      "learning_rate": 9.97701798637404e-06,
      "loss": 0.178,
      "step": 834
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5389935197646,
      "learning_rate": 9.976906913711634e-06,
      "loss": 0.2522,
      "step": 835
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4725179304387765,
      "learning_rate": 9.97679557390864e-06,
      "loss": 0.2277,
      "step": 836
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5704049070910084,
      "learning_rate": 9.97668396697103e-06,
      "loss": 0.2438,
      "step": 837
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.678906995398683,
      "learning_rate": 9.976572092904795e-06,
      "loss": 0.2614,
      "step": 838
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7149728778845337,
      "learning_rate": 9.976459951715941e-06,
      "loss": 0.2191,
      "step": 839
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6278354665289996,
      "learning_rate": 9.976347543410487e-06,
      "loss": 0.2412,
      "step": 840
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7365054099376105,
      "learning_rate": 9.976234867994467e-06,
      "loss": 0.241,
      "step": 841
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.427659833130422,
      "learning_rate": 9.976121925473931e-06,
      "loss": 0.2037,
      "step": 842
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5940463531031388,
      "learning_rate": 9.976008715854936e-06,
      "loss": 0.207,
      "step": 843
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5354101692087971,
      "learning_rate": 9.975895239143564e-06,
      "loss": 0.2284,
      "step": 844
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6081952704100781,
      "learning_rate": 9.975781495345903e-06,
      "loss": 0.2188,
      "step": 845
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5790256490995627,
      "learning_rate": 9.97566748446806e-06,
      "loss": 0.1857,
      "step": 846
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5844230644950563,
      "learning_rate": 9.975553206516153e-06,
      "loss": 0.2174,
      "step": 847
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6421126751219843,
      "learning_rate": 9.975438661496314e-06,
      "loss": 0.1999,
      "step": 848
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6206840404127683,
      "learning_rate": 9.975323849414698e-06,
      "loss": 0.2054,
      "step": 849
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2904874287566863,
      "learning_rate": 9.975208770277462e-06,
      "loss": 0.1837,
      "step": 850
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.70264692972742,
      "learning_rate": 9.975093424090785e-06,
      "loss": 0.2282,
      "step": 851
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5271252620187525,
      "learning_rate": 9.974977810860858e-06,
      "loss": 0.195,
      "step": 852
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6219489092879902,
      "learning_rate": 9.974861930593886e-06,
      "loss": 0.2042,
      "step": 853
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6191719672569294,
      "learning_rate": 9.97474578329609e-06,
      "loss": 0.2469,
      "step": 854
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4206726258229492,
      "learning_rate": 9.974629368973705e-06,
      "loss": 0.1726,
      "step": 855
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.584993069062993,
      "learning_rate": 9.974512687632978e-06,
      "loss": 0.6527,
      "step": 856
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.8990817437797625,
      "learning_rate": 9.974395739280172e-06,
      "loss": 0.2153,
      "step": 857
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4712668098786195,
      "learning_rate": 9.974278523921564e-06,
      "loss": 0.2145,
      "step": 858
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8510417067707465,
      "learning_rate": 9.974161041563448e-06,
      "loss": 0.2717,
      "step": 859
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5179264655749278,
      "learning_rate": 9.974043292212129e-06,
      "loss": 0.1958,
      "step": 860
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6123037604890762,
      "learning_rate": 9.973925275873926e-06,
      "loss": 0.2787,
      "step": 861
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.721121822032707,
      "learning_rate": 9.973806992555175e-06,
      "loss": 0.2423,
      "step": 862
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6309141528509064,
      "learning_rate": 9.973688442262224e-06,
      "loss": 0.1932,
      "step": 863
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8142396829400371,
      "learning_rate": 9.973569625001438e-06,
      "loss": 0.2959,
      "step": 864
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5920608255860877,
      "learning_rate": 9.973450540779193e-06,
      "loss": 0.2274,
      "step": 865
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5123799268103486,
      "learning_rate": 9.97333118960188e-06,
      "loss": 0.2155,
      "step": 866
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.553935628963694,
      "learning_rate": 9.973211571475909e-06,
      "loss": 0.2587,
      "step": 867
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5455077890832842,
      "learning_rate": 9.973091686407697e-06,
      "loss": 0.2001,
      "step": 868
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5358338304583472,
      "learning_rate": 9.972971534403681e-06,
      "loss": 0.2575,
      "step": 869
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5166559131058521,
      "learning_rate": 9.972851115470308e-06,
      "loss": 0.2039,
      "step": 870
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4559514271139282,
      "learning_rate": 9.972730429614044e-06,
      "loss": 0.1968,
      "step": 871
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.700957756057884,
      "learning_rate": 9.972609476841368e-06,
      "loss": 0.2518,
      "step": 872
    },
    {
      "epoch": 0.06,
      "grad_norm": 9.901233465877464,
      "learning_rate": 9.972488257158768e-06,
      "loss": 0.7071,
      "step": 873
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7669273349950874,
      "learning_rate": 9.972366770572755e-06,
      "loss": 0.2409,
      "step": 874
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3752067980995801,
      "learning_rate": 9.972245017089846e-06,
      "loss": 0.2118,
      "step": 875
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6180475222911945,
      "learning_rate": 9.97212299671658e-06,
      "loss": 0.2294,
      "step": 876
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.592938504733884,
      "learning_rate": 9.972000709459503e-06,
      "loss": 0.1946,
      "step": 877
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6366563942855288,
      "learning_rate": 9.971878155325182e-06,
      "loss": 0.2414,
      "step": 878
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.733104902323832,
      "learning_rate": 9.971755334320193e-06,
      "loss": 0.2162,
      "step": 879
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5688544180670083,
      "learning_rate": 9.97163224645113e-06,
      "loss": 0.2221,
      "step": 880
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6534573236758896,
      "learning_rate": 9.971508891724599e-06,
      "loss": 0.3192,
      "step": 881
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6426065087780641,
      "learning_rate": 9.97138527014722e-06,
      "loss": 0.2384,
      "step": 882
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.613366042446708,
      "learning_rate": 9.971261381725632e-06,
      "loss": 0.2227,
      "step": 883
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6749161156738825,
      "learning_rate": 9.971137226466482e-06,
      "loss": 0.2383,
      "step": 884
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7151584706333372,
      "learning_rate": 9.971012804376433e-06,
      "loss": 0.2421,
      "step": 885
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5387726601970686,
      "learning_rate": 9.97088811546217e-06,
      "loss": 0.2294,
      "step": 886
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.559544878333592,
      "learning_rate": 9.970763159730378e-06,
      "loss": 0.246,
      "step": 887
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6184225893374276,
      "learning_rate": 9.970637937187768e-06,
      "loss": 0.2472,
      "step": 888
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.661850326024661,
      "learning_rate": 9.970512447841061e-06,
      "loss": 0.2128,
      "step": 889
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5605615897481977,
      "learning_rate": 9.970386691696993e-06,
      "loss": 0.2223,
      "step": 890
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6576224210433015,
      "learning_rate": 9.970260668762313e-06,
      "loss": 0.2524,
      "step": 891
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6005518622094375,
      "learning_rate": 9.970134379043787e-06,
      "loss": 0.2471,
      "step": 892
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.532801506372282,
      "learning_rate": 9.970007822548191e-06,
      "loss": 0.2304,
      "step": 893
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5284133665120598,
      "learning_rate": 9.969880999282321e-06,
      "loss": 0.2142,
      "step": 894
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.421626529821479,
      "learning_rate": 9.969753909252984e-06,
      "loss": 0.2215,
      "step": 895
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6157719282384164,
      "learning_rate": 9.969626552467e-06,
      "loss": 0.2975,
      "step": 896
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.781303909704055,
      "learning_rate": 9.969498928931207e-06,
      "loss": 0.8017,
      "step": 897
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5426468687715291,
      "learning_rate": 9.969371038652453e-06,
      "loss": 0.2634,
      "step": 898
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.463209584547742,
      "learning_rate": 9.969242881637603e-06,
      "loss": 0.1993,
      "step": 899
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6062066599823315,
      "learning_rate": 9.96911445789354e-06,
      "loss": 0.2116,
      "step": 900
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.54423099328606,
      "learning_rate": 9.968985767427152e-06,
      "loss": 0.179,
      "step": 901
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8176125815329367,
      "learning_rate": 9.96885681024535e-06,
      "loss": 0.3119,
      "step": 902
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5222956348615937,
      "learning_rate": 9.968727586355053e-06,
      "loss": 0.2296,
      "step": 903
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6368882644654417,
      "learning_rate": 9.9685980957632e-06,
      "loss": 0.2471,
      "step": 904
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5935706288550355,
      "learning_rate": 9.96846833847674e-06,
      "loss": 0.1829,
      "step": 905
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4885986494261274,
      "learning_rate": 9.968338314502638e-06,
      "loss": 0.1977,
      "step": 906
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4588710763738337,
      "learning_rate": 9.968208023847875e-06,
      "loss": 0.2238,
      "step": 907
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.329017768769023,
      "learning_rate": 9.968077466519441e-06,
      "loss": 0.7373,
      "step": 908
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6607736213558415,
      "learning_rate": 9.967946642524346e-06,
      "loss": 0.252,
      "step": 909
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6420196087204497,
      "learning_rate": 9.967815551869613e-06,
      "loss": 0.2392,
      "step": 910
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6506488118836737,
      "learning_rate": 9.967684194562276e-06,
      "loss": 0.2504,
      "step": 911
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5358795510934171,
      "learning_rate": 9.967552570609387e-06,
      "loss": 0.2398,
      "step": 912
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4994411465727906,
      "learning_rate": 9.967420680018011e-06,
      "loss": 0.2305,
      "step": 913
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.358979407106109,
      "learning_rate": 9.96728852279523e-06,
      "loss": 0.6195,
      "step": 914
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6041590173560905,
      "learning_rate": 9.967156098948132e-06,
      "loss": 0.1652,
      "step": 915
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6934992184388324,
      "learning_rate": 9.967023408483828e-06,
      "loss": 0.3271,
      "step": 916
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5223110269152327,
      "learning_rate": 9.966890451409442e-06,
      "loss": 0.227,
      "step": 917
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5352648963577378,
      "learning_rate": 9.96675722773211e-06,
      "loss": 0.2099,
      "step": 918
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.524685601829434,
      "learning_rate": 9.96662373745898e-06,
      "loss": 0.238,
      "step": 919
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.832712105040142,
      "learning_rate": 9.966489980597217e-06,
      "loss": 0.2435,
      "step": 920
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.651158576144617,
      "learning_rate": 9.966355957154007e-06,
      "loss": 0.262,
      "step": 921
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5475918496449772,
      "learning_rate": 9.966221667136538e-06,
      "loss": 0.2493,
      "step": 922
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.55299882685511,
      "learning_rate": 9.966087110552018e-06,
      "loss": 0.1902,
      "step": 923
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.235272797053675,
      "learning_rate": 9.965952287407674e-06,
      "loss": 0.7592,
      "step": 924
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3831406711799605,
      "learning_rate": 9.965817197710738e-06,
      "loss": 0.2513,
      "step": 925
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4741567312307593,
      "learning_rate": 9.965681841468462e-06,
      "loss": 0.2555,
      "step": 926
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4048510468745647,
      "learning_rate": 9.965546218688115e-06,
      "loss": 0.1918,
      "step": 927
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.46812150736536,
      "learning_rate": 9.965410329376973e-06,
      "loss": 0.2391,
      "step": 928
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.755434499237654,
      "learning_rate": 9.96527417354233e-06,
      "loss": 0.3131,
      "step": 929
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6937683571286128,
      "learning_rate": 9.965137751191496e-06,
      "loss": 0.2236,
      "step": 930
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7512725369067013,
      "learning_rate": 9.965001062331794e-06,
      "loss": 0.2345,
      "step": 931
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7423063881768497,
      "learning_rate": 9.964864106970558e-06,
      "loss": 0.2368,
      "step": 932
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5079932650361105,
      "learning_rate": 9.964726885115143e-06,
      "loss": 0.1771,
      "step": 933
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.728006394200748,
      "learning_rate": 9.96458939677291e-06,
      "loss": 0.2762,
      "step": 934
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6514740941051578,
      "learning_rate": 9.964451641951245e-06,
      "loss": 0.2104,
      "step": 935
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.691657934625792,
      "learning_rate": 9.964313620657535e-06,
      "loss": 0.5863,
      "step": 936
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.226453049608885,
      "learning_rate": 9.964175332899195e-06,
      "loss": 0.6499,
      "step": 937
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.630677747583257,
      "learning_rate": 9.964036778683643e-06,
      "loss": 0.4104,
      "step": 938
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5788880678360995,
      "learning_rate": 9.963897958018319e-06,
      "loss": 0.2342,
      "step": 939
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8291733111366622,
      "learning_rate": 9.963758870910672e-06,
      "loss": 0.287,
      "step": 940
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5325230123014268,
      "learning_rate": 9.963619517368169e-06,
      "loss": 0.2369,
      "step": 941
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6504989571594537,
      "learning_rate": 9.96347989739829e-06,
      "loss": 0.2524,
      "step": 942
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.61217167711618,
      "learning_rate": 9.963340011008529e-06,
      "loss": 0.2426,
      "step": 943
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4962188746734713,
      "learning_rate": 9.963199858206393e-06,
      "loss": 0.2309,
      "step": 944
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.48074521140785,
      "learning_rate": 9.963059438999408e-06,
      "loss": 0.2741,
      "step": 945
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.449598767367919,
      "learning_rate": 9.96291875339511e-06,
      "loss": 0.6946,
      "step": 946
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.795464179187031,
      "learning_rate": 9.962777801401049e-06,
      "loss": 0.2927,
      "step": 947
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4815926665375998,
      "learning_rate": 9.962636583024792e-06,
      "loss": 0.2389,
      "step": 948
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.614062984684148,
      "learning_rate": 9.962495098273919e-06,
      "loss": 0.2371,
      "step": 949
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4228904690752817,
      "learning_rate": 9.962353347156023e-06,
      "loss": 0.1964,
      "step": 950
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.63019977509266,
      "learning_rate": 9.962211329678716e-06,
      "loss": 0.2447,
      "step": 951
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6166119929566223,
      "learning_rate": 9.962069045849617e-06,
      "loss": 0.2843,
      "step": 952
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3885502152335687,
      "learning_rate": 9.961926495676364e-06,
      "loss": 0.1813,
      "step": 953
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.94825947513103,
      "learning_rate": 9.96178367916661e-06,
      "loss": 0.7012,
      "step": 954
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6483847798299498,
      "learning_rate": 9.96164059632802e-06,
      "loss": 0.2631,
      "step": 955
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.638739558848523,
      "learning_rate": 9.961497247168274e-06,
      "loss": 0.2459,
      "step": 956
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6003613799432537,
      "learning_rate": 9.961353631695069e-06,
      "loss": 0.2088,
      "step": 957
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.623685768222847,
      "learning_rate": 9.96120974991611e-06,
      "loss": 0.2662,
      "step": 958
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.930013283164487,
      "learning_rate": 9.961065601839121e-06,
      "loss": 0.632,
      "step": 959
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.381462313226097,
      "learning_rate": 9.960921187471841e-06,
      "loss": 0.5662,
      "step": 960
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7552007258908067,
      "learning_rate": 9.960776506822019e-06,
      "loss": 0.2474,
      "step": 961
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.487902375137377,
      "learning_rate": 9.960631559897422e-06,
      "loss": 0.2098,
      "step": 962
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5128555437128612,
      "learning_rate": 9.960486346705833e-06,
      "loss": 0.2478,
      "step": 963
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6140132617818768,
      "learning_rate": 9.960340867255042e-06,
      "loss": 0.2555,
      "step": 964
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5401110662117066,
      "learning_rate": 9.96019512155286e-06,
      "loss": 0.2711,
      "step": 965
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6425389288891483,
      "learning_rate": 9.960049109607109e-06,
      "loss": 0.2617,
      "step": 966
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.319944009309532,
      "learning_rate": 9.959902831425628e-06,
      "loss": 0.6344,
      "step": 967
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6161669054356949,
      "learning_rate": 9.959756287016268e-06,
      "loss": 0.2004,
      "step": 968
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.602691630624776,
      "learning_rate": 9.959609476386893e-06,
      "loss": 0.689,
      "step": 969
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5953130640977635,
      "learning_rate": 9.959462399545385e-06,
      "loss": 0.213,
      "step": 970
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5036475372123885,
      "learning_rate": 9.95931505649964e-06,
      "loss": 0.2317,
      "step": 971
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7071886112631878,
      "learning_rate": 9.959167447257563e-06,
      "loss": 0.2716,
      "step": 972
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6212299792475897,
      "learning_rate": 9.95901957182708e-06,
      "loss": 0.2366,
      "step": 973
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6200587720853887,
      "learning_rate": 9.958871430216128e-06,
      "loss": 0.2662,
      "step": 974
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.841970417836384,
      "learning_rate": 9.958723022432658e-06,
      "loss": 0.252,
      "step": 975
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7644818196634862,
      "learning_rate": 9.958574348484635e-06,
      "loss": 0.2734,
      "step": 976
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5990626523788047,
      "learning_rate": 9.958425408380042e-06,
      "loss": 0.2069,
      "step": 977
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4292401560271797,
      "learning_rate": 9.958276202126871e-06,
      "loss": 0.2159,
      "step": 978
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3679376732221504,
      "learning_rate": 9.958126729733131e-06,
      "loss": 0.1647,
      "step": 979
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5217900723874238,
      "learning_rate": 9.957976991206847e-06,
      "loss": 0.2181,
      "step": 980
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6747889880308333,
      "learning_rate": 9.957826986556054e-06,
      "loss": 0.2234,
      "step": 981
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9021529860399558,
      "learning_rate": 9.957676715788806e-06,
      "loss": 0.2603,
      "step": 982
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6240680076233796,
      "learning_rate": 9.957526178913166e-06,
      "loss": 0.2359,
      "step": 983
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6213087323842932,
      "learning_rate": 9.957375375937217e-06,
      "loss": 0.2686,
      "step": 984
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.530914533968772,
      "learning_rate": 9.957224306869053e-06,
      "loss": 0.2526,
      "step": 985
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.962147982402997,
      "learning_rate": 9.957072971716782e-06,
      "loss": 0.8784,
      "step": 986
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7154213061160235,
      "learning_rate": 9.956921370488525e-06,
      "loss": 0.2762,
      "step": 987
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5005354505371913,
      "learning_rate": 9.956769503192426e-06,
      "loss": 0.2111,
      "step": 988
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.452198476061592,
      "learning_rate": 9.956617369836628e-06,
      "loss": 0.1901,
      "step": 989
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6766577721974762,
      "learning_rate": 9.956464970429303e-06,
      "loss": 0.2254,
      "step": 990
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.594729048086781,
      "learning_rate": 9.95631230497863e-06,
      "loss": 0.2235,
      "step": 991
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.592530275426905,
      "learning_rate": 9.956159373492802e-06,
      "loss": 0.2261,
      "step": 992
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4149798986491637,
      "learning_rate": 9.956006175980028e-06,
      "loss": 0.253,
      "step": 993
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5927402174211944,
      "learning_rate": 9.955852712448534e-06,
      "loss": 0.2576,
      "step": 994
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.457494320290799,
      "learning_rate": 9.955698982906552e-06,
      "loss": 0.2489,
      "step": 995
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4527550820359607,
      "learning_rate": 9.955544987362339e-06,
      "loss": 0.294,
      "step": 996
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5491164128132342,
      "learning_rate": 9.955390725824158e-06,
      "loss": 0.2554,
      "step": 997
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5021718477638957,
      "learning_rate": 9.955236198300289e-06,
      "loss": 0.246,
      "step": 998
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6303270064818196,
      "learning_rate": 9.955081404799026e-06,
      "loss": 0.2078,
      "step": 999
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7228813204533746,
      "learning_rate": 9.95492634532868e-06,
      "loss": 0.2227,
      "step": 1000
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7308317923845082,
      "learning_rate": 9.954771019897572e-06,
      "loss": 0.2805,
      "step": 1001
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.56778598008499,
      "learning_rate": 9.954615428514041e-06,
      "loss": 0.2276,
      "step": 1002
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6151375579070484,
      "learning_rate": 9.954459571186436e-06,
      "loss": 0.2747,
      "step": 1003
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5868432427891044,
      "learning_rate": 9.954303447923125e-06,
      "loss": 0.2059,
      "step": 1004
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4889428936652886,
      "learning_rate": 9.954147058732487e-06,
      "loss": 0.2156,
      "step": 1005
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4971643014065947,
      "learning_rate": 9.953990403622918e-06,
      "loss": 0.2236,
      "step": 1006
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6076050935851762,
      "learning_rate": 9.953833482602824e-06,
      "loss": 0.238,
      "step": 1007
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.22936853818382,
      "learning_rate": 9.953676295680628e-06,
      "loss": 0.6448,
      "step": 1008
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.439165206743901,
      "learning_rate": 9.95351884286477e-06,
      "loss": 0.2322,
      "step": 1009
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.073318128977774,
      "learning_rate": 9.9533611241637e-06,
      "loss": 0.2042,
      "step": 1010
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6915242133156365,
      "learning_rate": 9.953203139585883e-06,
      "loss": 0.284,
      "step": 1011
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4657506249351238,
      "learning_rate": 9.9530448891398e-06,
      "loss": 0.1963,
      "step": 1012
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.345719020870333,
      "learning_rate": 9.952886372833943e-06,
      "loss": 0.672,
      "step": 1013
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.602264922686953,
      "learning_rate": 9.952727590676824e-06,
      "loss": 0.6901,
      "step": 1014
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.685921053439328,
      "learning_rate": 9.952568542676963e-06,
      "loss": 0.2481,
      "step": 1015
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6795568931682097,
      "learning_rate": 9.9524092288429e-06,
      "loss": 0.2044,
      "step": 1016
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3857739558708633,
      "learning_rate": 9.952249649183183e-06,
      "loss": 0.2124,
      "step": 1017
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3608516768501657,
      "learning_rate": 9.95208980370638e-06,
      "loss": 0.2583,
      "step": 1018
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4809319433741186,
      "learning_rate": 9.951929692421069e-06,
      "loss": 0.2468,
      "step": 1019
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4419709553370685,
      "learning_rate": 9.951769315335843e-06,
      "loss": 0.2132,
      "step": 1020
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.650704989013261,
      "learning_rate": 9.951608672459316e-06,
      "loss": 0.2676,
      "step": 1021
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5398038052938297,
      "learning_rate": 9.951447763800106e-06,
      "loss": 0.2401,
      "step": 1022
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4932313993994064,
      "learning_rate": 9.951286589366852e-06,
      "loss": 0.2363,
      "step": 1023
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.449952888601099,
      "learning_rate": 9.951125149168203e-06,
      "loss": 0.2131,
      "step": 1024
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6093305970158536,
      "learning_rate": 9.950963443212826e-06,
      "loss": 0.2342,
      "step": 1025
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.546821521047033,
      "learning_rate": 9.950801471509402e-06,
      "loss": 0.264,
      "step": 1026
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.401297575007682,
      "learning_rate": 9.95063923406662e-06,
      "loss": 0.2469,
      "step": 1027
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5341210641132266,
      "learning_rate": 9.950476730893195e-06,
      "loss": 0.2182,
      "step": 1028
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6354599001862569,
      "learning_rate": 9.950313961997845e-06,
      "loss": 0.253,
      "step": 1029
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6302879145004092,
      "learning_rate": 9.950150927389308e-06,
      "loss": 0.2092,
      "step": 1030
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5392831763325525,
      "learning_rate": 9.949987627076337e-06,
      "loss": 0.2205,
      "step": 1031
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7121577887554154,
      "learning_rate": 9.949824061067692e-06,
      "loss": 0.2983,
      "step": 1032
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7152318786831497,
      "learning_rate": 9.94966022937216e-06,
      "loss": 0.2419,
      "step": 1033
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6075780577184156,
      "learning_rate": 9.949496131998527e-06,
      "loss": 0.2481,
      "step": 1034
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5406417578218947,
      "learning_rate": 9.949331768955606e-06,
      "loss": 0.192,
      "step": 1035
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5666289529594795,
      "learning_rate": 9.949167140252221e-06,
      "loss": 0.2438,
      "step": 1036
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.537364271359149,
      "learning_rate": 9.949002245897203e-06,
      "loss": 0.6747,
      "step": 1037
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4485754810341074,
      "learning_rate": 9.948837085899407e-06,
      "loss": 0.2323,
      "step": 1038
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8458906881058217,
      "learning_rate": 9.948671660267696e-06,
      "loss": 0.2931,
      "step": 1039
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6363330382828778,
      "learning_rate": 9.94850596901095e-06,
      "loss": 0.2517,
      "step": 1040
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.929636429784305,
      "learning_rate": 9.948340012138064e-06,
      "loss": 0.5907,
      "step": 1041
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.736192737380444,
      "learning_rate": 9.948173789657946e-06,
      "loss": 0.2119,
      "step": 1042
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7634488955715786,
      "learning_rate": 9.948007301579516e-06,
      "loss": 0.1989,
      "step": 1043
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3714381198643377,
      "learning_rate": 9.94784054791171e-06,
      "loss": 0.21,
      "step": 1044
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8850691204885872,
      "learning_rate": 9.947673528663482e-06,
      "loss": 0.1957,
      "step": 1045
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5800922825057861,
      "learning_rate": 9.947506243843795e-06,
      "loss": 0.2511,
      "step": 1046
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5351815574859178,
      "learning_rate": 9.947338693461626e-06,
      "loss": 0.254,
      "step": 1047
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5414123343060968,
      "learning_rate": 9.947170877525972e-06,
      "loss": 0.1957,
      "step": 1048
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5314756562783467,
      "learning_rate": 9.94700279604584e-06,
      "loss": 0.2126,
      "step": 1049
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.15792517419071,
      "learning_rate": 9.946834449030252e-06,
      "loss": 0.2546,
      "step": 1050
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4474075591114994,
      "learning_rate": 9.946665836488241e-06,
      "loss": 0.2088,
      "step": 1051
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6796718151856187,
      "learning_rate": 9.946496958428862e-06,
      "loss": 0.2568,
      "step": 1052
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4911973841778483,
      "learning_rate": 9.946327814861176e-06,
      "loss": 0.2314,
      "step": 1053
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4264998384092136,
      "learning_rate": 9.946158405794266e-06,
      "loss": 0.2221,
      "step": 1054
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.205927728205507,
      "learning_rate": 9.945988731237221e-06,
      "loss": 0.4087,
      "step": 1055
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.650392206599657,
      "learning_rate": 9.94581879119915e-06,
      "loss": 0.2368,
      "step": 1056
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.302826269422322,
      "learning_rate": 9.945648585689176e-06,
      "loss": 0.7123,
      "step": 1057
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.533800732905702,
      "learning_rate": 9.945478114716433e-06,
      "loss": 0.2274,
      "step": 1058
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.608307598501455,
      "learning_rate": 9.945307378290073e-06,
      "loss": 0.2735,
      "step": 1059
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7982908724076274,
      "learning_rate": 9.94513637641926e-06,
      "loss": 0.2438,
      "step": 1060
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6234499350765255,
      "learning_rate": 9.944965109113171e-06,
      "loss": 0.1964,
      "step": 1061
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6008586635131068,
      "learning_rate": 9.944793576381002e-06,
      "loss": 0.2253,
      "step": 1062
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.294362153837577,
      "learning_rate": 9.944621778231956e-06,
      "loss": 0.6718,
      "step": 1063
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6364230117572207,
      "learning_rate": 9.944449714675259e-06,
      "loss": 0.1984,
      "step": 1064
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.618832407664547,
      "learning_rate": 9.944277385720143e-06,
      "loss": 0.2123,
      "step": 1065
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4838019647981555,
      "learning_rate": 9.944104791375862e-06,
      "loss": 0.212,
      "step": 1066
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3781034204679117,
      "learning_rate": 9.943931931651675e-06,
      "loss": 0.2266,
      "step": 1067
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4310720361956148,
      "learning_rate": 9.943758806556864e-06,
      "loss": 0.2203,
      "step": 1068
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.362197559705591,
      "learning_rate": 9.94358541610072e-06,
      "loss": 0.2338,
      "step": 1069
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5611010576678022,
      "learning_rate": 9.943411760292552e-06,
      "loss": 0.2457,
      "step": 1070
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.518704795287663,
      "learning_rate": 9.94323783914168e-06,
      "loss": 0.2159,
      "step": 1071
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5303631008057625,
      "learning_rate": 9.94306365265744e-06,
      "loss": 0.2508,
      "step": 1072
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4062407195004938,
      "learning_rate": 9.942889200849178e-06,
      "loss": 0.2226,
      "step": 1073
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6175547562475912,
      "learning_rate": 9.942714483726263e-06,
      "loss": 0.2283,
      "step": 1074
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3525813966884446,
      "learning_rate": 9.94253950129807e-06,
      "loss": 0.2262,
      "step": 1075
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4714837178548161,
      "learning_rate": 9.942364253573993e-06,
      "loss": 0.1844,
      "step": 1076
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.010952712421839,
      "learning_rate": 9.942188740563437e-06,
      "loss": 0.437,
      "step": 1077
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.473048192083719,
      "learning_rate": 9.942012962275824e-06,
      "loss": 0.1895,
      "step": 1078
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4909759589197886,
      "learning_rate": 9.941836918720588e-06,
      "loss": 0.2048,
      "step": 1079
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.604877836246275,
      "learning_rate": 9.94166060990718e-06,
      "loss": 0.2546,
      "step": 1080
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6048371961017147,
      "learning_rate": 9.941484035845061e-06,
      "loss": 0.2848,
      "step": 1081
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5320663252210094,
      "learning_rate": 9.941307196543712e-06,
      "loss": 0.2647,
      "step": 1082
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5490125158760615,
      "learning_rate": 9.941130092012624e-06,
      "loss": 0.2358,
      "step": 1083
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5042218924932877,
      "learning_rate": 9.940952722261302e-06,
      "loss": 0.2078,
      "step": 1084
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0980929109302493,
      "learning_rate": 9.940775087299267e-06,
      "loss": 0.2855,
      "step": 1085
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.533576415958193,
      "learning_rate": 9.940597187136054e-06,
      "loss": 0.2005,
      "step": 1086
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.476112180082261,
      "learning_rate": 9.940419021781214e-06,
      "loss": 0.2065,
      "step": 1087
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4267578886258836,
      "learning_rate": 9.940240591244306e-06,
      "loss": 0.2217,
      "step": 1088
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6035014287713436,
      "learning_rate": 9.94006189553491e-06,
      "loss": 0.2159,
      "step": 1089
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5143518062808157,
      "learning_rate": 9.939882934662617e-06,
      "loss": 0.2006,
      "step": 1090
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5330403345386265,
      "learning_rate": 9.939703708637033e-06,
      "loss": 0.2251,
      "step": 1091
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8029975676302326,
      "learning_rate": 9.939524217467779e-06,
      "loss": 0.3352,
      "step": 1092
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6536029638125382,
      "learning_rate": 9.939344461164488e-06,
      "loss": 0.2141,
      "step": 1093
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8072385104890933,
      "learning_rate": 9.93916443973681e-06,
      "loss": 0.2432,
      "step": 1094
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5992702637251062,
      "learning_rate": 9.938984153194406e-06,
      "loss": 0.2186,
      "step": 1095
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6006705193402297,
      "learning_rate": 9.938803601546957e-06,
      "loss": 0.2712,
      "step": 1096
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.443343958671591,
      "learning_rate": 9.938622784804149e-06,
      "loss": 0.2242,
      "step": 1097
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6050528431399598,
      "learning_rate": 9.938441702975689e-06,
      "loss": 0.1861,
      "step": 1098
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.103841139360225,
      "learning_rate": 9.938260356071299e-06,
      "loss": 0.7147,
      "step": 1099
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7027121410950583,
      "learning_rate": 9.938078744100713e-06,
      "loss": 0.2155,
      "step": 1100
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6848330906460747,
      "learning_rate": 9.937896867073678e-06,
      "loss": 0.2064,
      "step": 1101
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4678293835639407,
      "learning_rate": 9.937714724999953e-06,
      "loss": 0.2026,
      "step": 1102
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2260575484814509,
      "learning_rate": 9.937532317889322e-06,
      "loss": 0.2115,
      "step": 1103
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3860522733887144,
      "learning_rate": 9.93734964575157e-06,
      "loss": 0.2065,
      "step": 1104
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.631746959209183,
      "learning_rate": 9.937166708596504e-06,
      "loss": 0.7735,
      "step": 1105
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5163915010165294,
      "learning_rate": 9.936983506433945e-06,
      "loss": 0.2593,
      "step": 1106
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.090629076623357,
      "learning_rate": 9.936800039273724e-06,
      "loss": 0.6032,
      "step": 1107
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.12088079571952,
      "learning_rate": 9.93661630712569e-06,
      "loss": 0.6263,
      "step": 1108
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5926099572905967,
      "learning_rate": 9.936432309999704e-06,
      "loss": 0.2661,
      "step": 1109
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3855609528300077,
      "learning_rate": 9.936248047905644e-06,
      "loss": 0.1831,
      "step": 1110
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5507617689660618,
      "learning_rate": 9.936063520853399e-06,
      "loss": 0.2646,
      "step": 1111
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5036292226530938,
      "learning_rate": 9.935878728852873e-06,
      "loss": 0.244,
      "step": 1112
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5683808728634534,
      "learning_rate": 9.935693671913986e-06,
      "loss": 0.2226,
      "step": 1113
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5142295296673947,
      "learning_rate": 9.935508350046672e-06,
      "loss": 0.2167,
      "step": 1114
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7190014522576007,
      "learning_rate": 9.93532276326088e-06,
      "loss": 0.2242,
      "step": 1115
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6279209691401062,
      "learning_rate": 9.935136911566566e-06,
      "loss": 0.2144,
      "step": 1116
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5955580635135345,
      "learning_rate": 9.934950794973708e-06,
      "loss": 0.2377,
      "step": 1117
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.42747071263188,
      "learning_rate": 9.9347644134923e-06,
      "loss": 0.7366,
      "step": 1118
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5054239783526073,
      "learning_rate": 9.934577767132341e-06,
      "loss": 0.2454,
      "step": 1119
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.706862813092047,
      "learning_rate": 9.934390855903852e-06,
      "loss": 0.2594,
      "step": 1120
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.515741794429253,
      "learning_rate": 9.934203679816865e-06,
      "loss": 0.2021,
      "step": 1121
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4469333636053692,
      "learning_rate": 9.934016238881428e-06,
      "loss": 0.2444,
      "step": 1122
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3676366169245366,
      "learning_rate": 9.9338285331076e-06,
      "loss": 0.2038,
      "step": 1123
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.100441246083172,
      "learning_rate": 9.933640562505458e-06,
      "loss": 0.7188,
      "step": 1124
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5527255683800822,
      "learning_rate": 9.93345232708509e-06,
      "loss": 0.2976,
      "step": 1125
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5974038948754141,
      "learning_rate": 9.933263826856601e-06,
      "loss": 0.2052,
      "step": 1126
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7007809308034525,
      "learning_rate": 9.93307506183011e-06,
      "loss": 0.21,
      "step": 1127
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6665260086696043,
      "learning_rate": 9.932886032015746e-06,
      "loss": 0.2197,
      "step": 1128
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7421161507008942,
      "learning_rate": 9.932696737423656e-06,
      "loss": 0.2739,
      "step": 1129
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.67225587235492,
      "learning_rate": 9.932507178064006e-06,
      "loss": 0.2428,
      "step": 1130
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4742943367921002,
      "learning_rate": 9.932317353946964e-06,
      "loss": 0.2378,
      "step": 1131
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.390701134572608,
      "learning_rate": 9.93212726508272e-06,
      "loss": 0.6425,
      "step": 1132
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6743390097737465,
      "learning_rate": 9.931936911481481e-06,
      "loss": 0.228,
      "step": 1133
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5685427760985085,
      "learning_rate": 9.93174629315346e-06,
      "loss": 0.2662,
      "step": 1134
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4979379074089965,
      "learning_rate": 9.931555410108893e-06,
      "loss": 0.27,
      "step": 1135
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.1305112799585135,
      "learning_rate": 9.931364262358023e-06,
      "loss": 0.6212,
      "step": 1136
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4919989447400548,
      "learning_rate": 9.93117284991111e-06,
      "loss": 0.242,
      "step": 1137
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6726226969839952,
      "learning_rate": 9.93098117277843e-06,
      "loss": 0.2618,
      "step": 1138
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7796480013875138,
      "learning_rate": 9.930789230970269e-06,
      "loss": 0.3036,
      "step": 1139
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6059924919551873,
      "learning_rate": 9.930597024496933e-06,
      "loss": 0.2369,
      "step": 1140
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4118458684547648,
      "learning_rate": 9.930404553368733e-06,
      "loss": 0.2756,
      "step": 1141
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4364800363949266,
      "learning_rate": 9.930211817596008e-06,
      "loss": 0.227,
      "step": 1142
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4990861665393787,
      "learning_rate": 9.930018817189097e-06,
      "loss": 0.2508,
      "step": 1143
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4195372228809104,
      "learning_rate": 9.929825552158364e-06,
      "loss": 0.2081,
      "step": 1144
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5782544500786064,
      "learning_rate": 9.92963202251418e-06,
      "loss": 0.25,
      "step": 1145
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.384373391912052,
      "learning_rate": 9.929438228266932e-06,
      "loss": 0.193,
      "step": 1146
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.649296242057296,
      "learning_rate": 9.929244169427023e-06,
      "loss": 0.2512,
      "step": 1147
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.541559488811707,
      "learning_rate": 9.929049846004873e-06,
      "loss": 0.2051,
      "step": 1148
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3813331356300795,
      "learning_rate": 9.928855258010907e-06,
      "loss": 0.1937,
      "step": 1149
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4689029334594381,
      "learning_rate": 9.928660405455572e-06,
      "loss": 0.1947,
      "step": 1150
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4539609865114502,
      "learning_rate": 9.928465288349328e-06,
      "loss": 0.1889,
      "step": 1151
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5608175162923896,
      "learning_rate": 9.928269906702647e-06,
      "loss": 0.2299,
      "step": 1152
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.550219914255769,
      "learning_rate": 9.928074260526014e-06,
      "loss": 0.2385,
      "step": 1153
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4124121742478943,
      "learning_rate": 9.927878349829935e-06,
      "loss": 0.2203,
      "step": 1154
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6128076753656464,
      "learning_rate": 9.927682174624925e-06,
      "loss": 0.2533,
      "step": 1155
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5212727843845337,
      "learning_rate": 9.927485734921511e-06,
      "loss": 0.2397,
      "step": 1156
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5044143232968135,
      "learning_rate": 9.927289030730239e-06,
      "loss": 0.2378,
      "step": 1157
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5732720262508046,
      "learning_rate": 9.927092062061667e-06,
      "loss": 0.2392,
      "step": 1158
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.41769202569861,
      "learning_rate": 9.926894828926366e-06,
      "loss": 0.1861,
      "step": 1159
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4346003944295718,
      "learning_rate": 9.926697331334924e-06,
      "loss": 0.2158,
      "step": 1160
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.576747012069802,
      "learning_rate": 9.926499569297943e-06,
      "loss": 0.207,
      "step": 1161
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.656663878987908,
      "learning_rate": 9.926301542826038e-06,
      "loss": 0.2502,
      "step": 1162
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.541954005611327,
      "learning_rate": 9.926103251929836e-06,
      "loss": 0.2507,
      "step": 1163
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.530091662294955,
      "learning_rate": 9.925904696619983e-06,
      "loss": 0.2362,
      "step": 1164
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7139948753263308,
      "learning_rate": 9.925705876907134e-06,
      "loss": 0.2616,
      "step": 1165
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4635097188487525,
      "learning_rate": 9.925506792801962e-06,
      "loss": 0.2328,
      "step": 1166
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3361479247209194,
      "learning_rate": 9.925307444315155e-06,
      "loss": 0.1916,
      "step": 1167
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.399460259709606,
      "learning_rate": 9.92510783145741e-06,
      "loss": 0.5664,
      "step": 1168
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1518466693233513,
      "learning_rate": 9.924907954239446e-06,
      "loss": 0.2602,
      "step": 1169
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5263425809631828,
      "learning_rate": 9.924707812671986e-06,
      "loss": 0.2519,
      "step": 1170
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6937103464510361,
      "learning_rate": 9.924507406765776e-06,
      "loss": 0.2889,
      "step": 1171
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6133497529694538,
      "learning_rate": 9.924306736531572e-06,
      "loss": 0.2677,
      "step": 1172
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6704593822957057,
      "learning_rate": 9.924105801980146e-06,
      "loss": 0.2491,
      "step": 1173
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.614246328017673,
      "learning_rate": 9.923904603122283e-06,
      "loss": 0.2925,
      "step": 1174
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4067812460657783,
      "learning_rate": 9.923703139968784e-06,
      "loss": 0.2025,
      "step": 1175
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6920495664016721,
      "learning_rate": 9.923501412530458e-06,
      "loss": 0.2604,
      "step": 1176
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4881384936536035,
      "learning_rate": 9.92329942081814e-06,
      "loss": 0.2449,
      "step": 1177
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4971607178529065,
      "learning_rate": 9.923097164842666e-06,
      "loss": 0.2001,
      "step": 1178
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4818095392721269,
      "learning_rate": 9.922894644614895e-06,
      "loss": 0.2229,
      "step": 1179
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7579447390164427,
      "learning_rate": 9.922691860145696e-06,
      "loss": 0.2918,
      "step": 1180
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6318483684604876,
      "learning_rate": 9.922488811445957e-06,
      "loss": 0.2466,
      "step": 1181
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.397677123564768,
      "learning_rate": 9.922285498526575e-06,
      "loss": 0.206,
      "step": 1182
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3998255162937556,
      "learning_rate": 9.922081921398458e-06,
      "loss": 0.2154,
      "step": 1183
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5175236094470617,
      "learning_rate": 9.921878080072542e-06,
      "loss": 0.2611,
      "step": 1184
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6542587296439337,
      "learning_rate": 9.921673974559763e-06,
      "loss": 0.2527,
      "step": 1185
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6569505658546817,
      "learning_rate": 9.92146960487108e-06,
      "loss": 0.2777,
      "step": 1186
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5769165582412397,
      "learning_rate": 9.921264971017457e-06,
      "loss": 0.2121,
      "step": 1187
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5301618160383677,
      "learning_rate": 9.921060073009884e-06,
      "loss": 0.2063,
      "step": 1188
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4483243795898726,
      "learning_rate": 9.920854910859356e-06,
      "loss": 0.1978,
      "step": 1189
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5470267121396963,
      "learning_rate": 9.920649484576885e-06,
      "loss": 0.2588,
      "step": 1190
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6008871512704161,
      "learning_rate": 9.9204437941735e-06,
      "loss": 0.2213,
      "step": 1191
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6873839191317468,
      "learning_rate": 9.920237839660239e-06,
      "loss": 0.2338,
      "step": 1192
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6556225732153522,
      "learning_rate": 9.920031621048157e-06,
      "loss": 0.2025,
      "step": 1193
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4842168332252705,
      "learning_rate": 9.919825138348325e-06,
      "loss": 0.212,
      "step": 1194
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.470374524714982,
      "learning_rate": 9.919618391571825e-06,
      "loss": 0.19,
      "step": 1195
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.582239383420492,
      "learning_rate": 9.919411380729753e-06,
      "loss": 0.2331,
      "step": 1196
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7227821288652811,
      "learning_rate": 9.919204105833224e-06,
      "loss": 0.3193,
      "step": 1197
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.889018099274441,
      "learning_rate": 9.91899656689336e-06,
      "loss": 0.5393,
      "step": 1198
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6496164398115643,
      "learning_rate": 9.918788763921303e-06,
      "loss": 0.2534,
      "step": 1199
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6618716608891957,
      "learning_rate": 9.918580696928206e-06,
      "loss": 0.2561,
      "step": 1200
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6148497683512029,
      "learning_rate": 9.918372365925238e-06,
      "loss": 0.2274,
      "step": 1201
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.394778092833336,
      "learning_rate": 9.91816377092358e-06,
      "loss": 0.1879,
      "step": 1202
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5494693348575201,
      "learning_rate": 9.917954911934433e-06,
      "loss": 0.2691,
      "step": 1203
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8110328118956778,
      "learning_rate": 9.917745788969003e-06,
      "loss": 0.2544,
      "step": 1204
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.1062923464027055,
      "learning_rate": 9.917536402038513e-06,
      "loss": 0.709,
      "step": 1205
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3807016297769146,
      "learning_rate": 9.917326751154208e-06,
      "loss": 0.2091,
      "step": 1206
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5357985916483932,
      "learning_rate": 9.91711683632734e-06,
      "loss": 0.2486,
      "step": 1207
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5483645553279626,
      "learning_rate": 9.916906657569171e-06,
      "loss": 0.2655,
      "step": 1208
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.961588269399571,
      "learning_rate": 9.916696214890988e-06,
      "loss": 0.6549,
      "step": 1209
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3710636795800009,
      "learning_rate": 9.916485508304085e-06,
      "loss": 0.1995,
      "step": 1210
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3715187586041144,
      "learning_rate": 9.916274537819774e-06,
      "loss": 0.2305,
      "step": 1211
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5184926761534008,
      "learning_rate": 9.916063303449374e-06,
      "loss": 0.2377,
      "step": 1212
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4819978853399634,
      "learning_rate": 9.91585180520423e-06,
      "loss": 0.2488,
      "step": 1213
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5176154221603166,
      "learning_rate": 9.91564004309569e-06,
      "loss": 0.225,
      "step": 1214
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3686378811279407,
      "learning_rate": 9.915428017135117e-06,
      "loss": 0.2255,
      "step": 1215
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5532582572507878,
      "learning_rate": 9.9152157273339e-06,
      "loss": 0.1976,
      "step": 1216
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5731653038139306,
      "learning_rate": 9.91500317370343e-06,
      "loss": 0.2305,
      "step": 1217
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4695741876759432,
      "learning_rate": 9.914790356255115e-06,
      "loss": 0.2105,
      "step": 1218
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6214055641875886,
      "learning_rate": 9.914577275000378e-06,
      "loss": 0.2508,
      "step": 1219
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.342472617228648,
      "learning_rate": 9.91436392995066e-06,
      "loss": 0.1919,
      "step": 1220
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.422136636302543,
      "learning_rate": 9.914150321117408e-06,
      "loss": 0.2324,
      "step": 1221
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8441553307439726,
      "learning_rate": 9.91393644851209e-06,
      "loss": 0.2887,
      "step": 1222
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3825142476097159,
      "learning_rate": 9.913722312146186e-06,
      "loss": 0.1851,
      "step": 1223
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5899096399160513,
      "learning_rate": 9.913507912031188e-06,
      "loss": 0.3012,
      "step": 1224
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.469712012626539,
      "learning_rate": 9.913293248178608e-06,
      "loss": 0.1877,
      "step": 1225
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7902861632241287,
      "learning_rate": 9.913078320599963e-06,
      "loss": 0.2418,
      "step": 1226
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.647543320833529,
      "learning_rate": 9.912863129306796e-06,
      "loss": 0.2806,
      "step": 1227
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.11614893267168,
      "learning_rate": 9.912647674310652e-06,
      "loss": 0.5523,
      "step": 1228
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.62136549114214,
      "learning_rate": 9.912431955623096e-06,
      "loss": 0.227,
      "step": 1229
    },
    {
      "epoch": 0.09,
      "grad_norm": 12.017234390944381,
      "learning_rate": 9.91221597325571e-06,
      "loss": 0.511,
      "step": 1230
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.762855647248683,
      "learning_rate": 9.911999727220088e-06,
      "loss": 0.2332,
      "step": 1231
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.427478503346927,
      "learning_rate": 9.911783217527833e-06,
      "loss": 0.2173,
      "step": 1232
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3197486313148674,
      "learning_rate": 9.911566444190568e-06,
      "loss": 0.1993,
      "step": 1233
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4720266891056886,
      "learning_rate": 9.91134940721993e-06,
      "loss": 0.1967,
      "step": 1234
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5855416613511666,
      "learning_rate": 9.911132106627565e-06,
      "loss": 0.2397,
      "step": 1235
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5034429604427793,
      "learning_rate": 9.910914542425143e-06,
      "loss": 0.2126,
      "step": 1236
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7639865151303904,
      "learning_rate": 9.910696714624336e-06,
      "loss": 0.2739,
      "step": 1237
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4596367840619353,
      "learning_rate": 9.910478623236838e-06,
      "loss": 0.2565,
      "step": 1238
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5653254401290901,
      "learning_rate": 9.910260268274358e-06,
      "loss": 0.2415,
      "step": 1239
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6400025579810504,
      "learning_rate": 9.910041649748613e-06,
      "loss": 0.2033,
      "step": 1240
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5388289233467425,
      "learning_rate": 9.90982276767134e-06,
      "loss": 0.2312,
      "step": 1241
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7193470979175998,
      "learning_rate": 9.909603622054285e-06,
      "loss": 0.2522,
      "step": 1242
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6148638184867419,
      "learning_rate": 9.909384212909214e-06,
      "loss": 0.2485,
      "step": 1243
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.36300648709624,
      "learning_rate": 9.909164540247902e-06,
      "loss": 0.1926,
      "step": 1244
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4856544469731123,
      "learning_rate": 9.908944604082139e-06,
      "loss": 0.226,
      "step": 1245
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3508106269052123,
      "learning_rate": 9.908724404423732e-06,
      "loss": 0.2034,
      "step": 1246
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.538132258303328,
      "learning_rate": 9.908503941284502e-06,
      "loss": 0.2195,
      "step": 1247
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6295258370561332,
      "learning_rate": 9.90828321467628e-06,
      "loss": 0.2147,
      "step": 1248
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.726687679437476,
      "learning_rate": 9.908062224610915e-06,
      "loss": 0.3342,
      "step": 1249
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5460182083651854,
      "learning_rate": 9.907840971100268e-06,
      "loss": 0.236,
      "step": 1250
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5197901052458545,
      "learning_rate": 9.907619454156215e-06,
      "loss": 0.2125,
      "step": 1251
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.518320084139661,
      "learning_rate": 9.907397673790648e-06,
      "loss": 0.2032,
      "step": 1252
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6049641941003987,
      "learning_rate": 9.907175630015468e-06,
      "loss": 0.1844,
      "step": 1253
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.546122120994793,
      "learning_rate": 9.906953322842598e-06,
      "loss": 0.1985,
      "step": 1254
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3832051092119177,
      "learning_rate": 9.906730752283965e-06,
      "loss": 0.2494,
      "step": 1255
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.640238718358799,
      "learning_rate": 9.90650791835152e-06,
      "loss": 0.2726,
      "step": 1256
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6405238650307712,
      "learning_rate": 9.906284821057223e-06,
      "loss": 0.2289,
      "step": 1257
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.563050941351312,
      "learning_rate": 9.906061460413047e-06,
      "loss": 0.2771,
      "step": 1258
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.756122356907297,
      "learning_rate": 9.905837836430985e-06,
      "loss": 0.425,
      "step": 1259
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5651730236004022,
      "learning_rate": 9.905613949123036e-06,
      "loss": 0.2445,
      "step": 1260
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5033551117000115,
      "learning_rate": 9.905389798501219e-06,
      "loss": 0.243,
      "step": 1261
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.264898963025065,
      "learning_rate": 9.905165384577567e-06,
      "loss": 0.6549,
      "step": 1262
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3901725308933657,
      "learning_rate": 9.904940707364123e-06,
      "loss": 0.2207,
      "step": 1263
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5141064897345597,
      "learning_rate": 9.90471576687295e-06,
      "loss": 0.2277,
      "step": 1264
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3814064661575949,
      "learning_rate": 9.904490563116118e-06,
      "loss": 0.1814,
      "step": 1265
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6340115808253446,
      "learning_rate": 9.904265096105719e-06,
      "loss": 0.2645,
      "step": 1266
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3998514593752909,
      "learning_rate": 9.90403936585385e-06,
      "loss": 0.2031,
      "step": 1267
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.146258705481285,
      "learning_rate": 9.903813372372633e-06,
      "loss": 0.8096,
      "step": 1268
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.70104468800173,
      "learning_rate": 9.903587115674196e-06,
      "loss": 0.3099,
      "step": 1269
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6901948751544316,
      "learning_rate": 9.903360595770684e-06,
      "loss": 0.2236,
      "step": 1270
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4854755572626794,
      "learning_rate": 9.903133812674252e-06,
      "loss": 0.2529,
      "step": 1271
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.478866278812421,
      "learning_rate": 9.90290676639708e-06,
      "loss": 0.2555,
      "step": 1272
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4216083076053283,
      "learning_rate": 9.90267945695135e-06,
      "loss": 0.2227,
      "step": 1273
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5423743169132416,
      "learning_rate": 9.902451884349263e-06,
      "loss": 0.2002,
      "step": 1274
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3479660981400268,
      "learning_rate": 9.902224048603036e-06,
      "loss": 0.1932,
      "step": 1275
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3940957398424716,
      "learning_rate": 9.901995949724897e-06,
      "loss": 0.2187,
      "step": 1276
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5846144536629947,
      "learning_rate": 9.90176758772709e-06,
      "loss": 0.2196,
      "step": 1277
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5294697189331377,
      "learning_rate": 9.901538962621873e-06,
      "loss": 0.2626,
      "step": 1278
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.550598250091117,
      "learning_rate": 9.901310074421518e-06,
      "loss": 0.2681,
      "step": 1279
    },
    {
      "epoch": 0.09,
      "grad_norm": 8.967554515038206,
      "learning_rate": 9.901080923138308e-06,
      "loss": 0.6241,
      "step": 1280
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.490726447050626,
      "learning_rate": 9.900851508784548e-06,
      "loss": 0.2159,
      "step": 1281
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6237615341273124,
      "learning_rate": 9.900621831372545e-06,
      "loss": 0.2517,
      "step": 1282
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6117545928584855,
      "learning_rate": 9.900391890914634e-06,
      "loss": 0.2551,
      "step": 1283
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2333949243894047,
      "learning_rate": 9.900161687423155e-06,
      "loss": 0.1823,
      "step": 1284
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4832588808142355,
      "learning_rate": 9.899931220910463e-06,
      "loss": 0.2169,
      "step": 1285
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3608787648012268,
      "learning_rate": 9.89970049138893e-06,
      "loss": 0.2012,
      "step": 1286
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4953636941675037,
      "learning_rate": 9.89946949887094e-06,
      "loss": 0.2311,
      "step": 1287
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4113748872192358,
      "learning_rate": 9.899238243368889e-06,
      "loss": 0.2047,
      "step": 1288
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.466324016834527,
      "learning_rate": 9.899006724895197e-06,
      "loss": 0.217,
      "step": 1289
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5682297216273149,
      "learning_rate": 9.898774943462285e-06,
      "loss": 0.2479,
      "step": 1290
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5234740176823738,
      "learning_rate": 9.898542899082595e-06,
      "loss": 0.2564,
      "step": 1291
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4887896434159156,
      "learning_rate": 9.898310591768583e-06,
      "loss": 0.241,
      "step": 1292
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5827242336135758,
      "learning_rate": 9.89807802153272e-06,
      "loss": 0.256,
      "step": 1293
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4631264116159273,
      "learning_rate": 9.897845188387487e-06,
      "loss": 0.1931,
      "step": 1294
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5003118170591392,
      "learning_rate": 9.897612092345383e-06,
      "loss": 0.2029,
      "step": 1295
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7666410515699988,
      "learning_rate": 9.897378733418918e-06,
      "loss": 0.2885,
      "step": 1296
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4839784911516873,
      "learning_rate": 9.897145111620618e-06,
      "loss": 0.2389,
      "step": 1297
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6910140011172734,
      "learning_rate": 9.896911226963026e-06,
      "loss": 0.2213,
      "step": 1298
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3231164481107967,
      "learning_rate": 9.896677079458692e-06,
      "loss": 0.205,
      "step": 1299
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5126383231950964,
      "learning_rate": 9.896442669120188e-06,
      "loss": 0.2261,
      "step": 1300
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4454015904020094,
      "learning_rate": 9.896207995960092e-06,
      "loss": 0.2145,
      "step": 1301
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5877457980205394,
      "learning_rate": 9.895973059991003e-06,
      "loss": 0.2209,
      "step": 1302
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6136708411679674,
      "learning_rate": 9.89573786122553e-06,
      "loss": 0.2335,
      "step": 1303
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4076390352518964,
      "learning_rate": 9.8955023996763e-06,
      "loss": 0.2312,
      "step": 1304
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.863147246066385,
      "learning_rate": 9.895266675355949e-06,
      "loss": 0.3263,
      "step": 1305
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4040915020753781,
      "learning_rate": 9.895030688277131e-06,
      "loss": 0.1963,
      "step": 1306
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.870829493177399,
      "learning_rate": 9.894794438452513e-06,
      "loss": 0.6985,
      "step": 1307
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6166373408190315,
      "learning_rate": 9.894557925894775e-06,
      "loss": 0.2609,
      "step": 1308
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5293210424360062,
      "learning_rate": 9.894321150616612e-06,
      "loss": 0.2113,
      "step": 1309
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.536872763552488,
      "learning_rate": 9.894084112630735e-06,
      "loss": 0.7996,
      "step": 1310
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4916100794669982,
      "learning_rate": 9.893846811949865e-06,
      "loss": 0.2438,
      "step": 1311
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.394152381409212,
      "learning_rate": 9.89360924858674e-06,
      "loss": 0.2331,
      "step": 1312
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4745935563236365,
      "learning_rate": 9.893371422554114e-06,
      "loss": 0.2359,
      "step": 1313
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6634350163496008,
      "learning_rate": 9.893133333864749e-06,
      "loss": 0.2474,
      "step": 1314
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6554119110665917,
      "learning_rate": 9.892894982531425e-06,
      "loss": 0.2473,
      "step": 1315
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5294445931095408,
      "learning_rate": 9.892656368566938e-06,
      "loss": 0.2349,
      "step": 1316
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4601902084755956,
      "learning_rate": 9.892417491984095e-06,
      "loss": 0.2039,
      "step": 1317
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.597125024458567,
      "learning_rate": 9.892178352795716e-06,
      "loss": 0.278,
      "step": 1318
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.511022778737241,
      "learning_rate": 9.891938951014639e-06,
      "loss": 0.2567,
      "step": 1319
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5964420404067765,
      "learning_rate": 9.891699286653714e-06,
      "loss": 0.2598,
      "step": 1320
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.44226348445606,
      "learning_rate": 9.891459359725805e-06,
      "loss": 0.2188,
      "step": 1321
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5032803246630735,
      "learning_rate": 9.891219170243791e-06,
      "loss": 0.2266,
      "step": 1322
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.588228800107201,
      "learning_rate": 9.890978718220563e-06,
      "loss": 0.2722,
      "step": 1323
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.422475275426151,
      "learning_rate": 9.890738003669029e-06,
      "loss": 0.6185,
      "step": 1324
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4254569142519447,
      "learning_rate": 9.890497026602108e-06,
      "loss": 0.2258,
      "step": 1325
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.392841461002796,
      "learning_rate": 9.890255787032736e-06,
      "loss": 0.2142,
      "step": 1326
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.368277498764144,
      "learning_rate": 9.89001428497386e-06,
      "loss": 0.2266,
      "step": 1327
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4457252361224269,
      "learning_rate": 9.889772520438447e-06,
      "loss": 0.1947,
      "step": 1328
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.586152564556905,
      "learning_rate": 9.88953049343947e-06,
      "loss": 0.7718,
      "step": 1329
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3388920790034404,
      "learning_rate": 9.889288203989919e-06,
      "loss": 0.2241,
      "step": 1330
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.478649561798433,
      "learning_rate": 9.889045652102805e-06,
      "loss": 0.2186,
      "step": 1331
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4237016509765614,
      "learning_rate": 9.88880283779114e-06,
      "loss": 0.2196,
      "step": 1332
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.7298903836119,
      "learning_rate": 9.888559761067963e-06,
      "loss": 0.6355,
      "step": 1333
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6566928382950308,
      "learning_rate": 9.888316421946317e-06,
      "loss": 0.1872,
      "step": 1334
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.826340890527177,
      "learning_rate": 9.88807282043927e-06,
      "loss": 0.2396,
      "step": 1335
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6141911071296187,
      "learning_rate": 9.88782895655989e-06,
      "loss": 0.2201,
      "step": 1336
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4698602469849074,
      "learning_rate": 9.887584830321271e-06,
      "loss": 0.1795,
      "step": 1337
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6341767551365083,
      "learning_rate": 9.887340441736515e-06,
      "loss": 0.2546,
      "step": 1338
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4045203198486547,
      "learning_rate": 9.887095790818739e-06,
      "loss": 0.1918,
      "step": 1339
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4944385918927947,
      "learning_rate": 9.886850877581079e-06,
      "loss": 0.2423,
      "step": 1340
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4789529967825175,
      "learning_rate": 9.886605702036678e-06,
      "loss": 0.2613,
      "step": 1341
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.388234486678817,
      "learning_rate": 9.886360264198695e-06,
      "loss": 0.2352,
      "step": 1342
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3978771715379719,
      "learning_rate": 9.886114564080307e-06,
      "loss": 0.1892,
      "step": 1343
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7467107363801577,
      "learning_rate": 9.8858686016947e-06,
      "loss": 0.2613,
      "step": 1344
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4715879407005288,
      "learning_rate": 9.885622377055076e-06,
      "loss": 0.2097,
      "step": 1345
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4092776507249072,
      "learning_rate": 9.885375890174653e-06,
      "loss": 0.2013,
      "step": 1346
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5703790976069962,
      "learning_rate": 9.88512914106666e-06,
      "loss": 0.209,
      "step": 1347
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7074739303008952,
      "learning_rate": 9.884882129744344e-06,
      "loss": 0.2355,
      "step": 1348
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.050644198503285,
      "learning_rate": 9.88463485622096e-06,
      "loss": 0.6451,
      "step": 1349
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6793638226009862,
      "learning_rate": 9.884387320509785e-06,
      "loss": 0.2722,
      "step": 1350
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.661645961215535,
      "learning_rate": 9.884139522624103e-06,
      "loss": 0.7158,
      "step": 1351
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4789662739595457,
      "learning_rate": 9.883891462577212e-06,
      "loss": 0.2387,
      "step": 1352
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3949955057286314,
      "learning_rate": 9.883643140382434e-06,
      "loss": 0.2468,
      "step": 1353
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3808342402643738,
      "learning_rate": 9.883394556053092e-06,
      "loss": 0.2098,
      "step": 1354
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.538101242370762,
      "learning_rate": 9.883145709602531e-06,
      "loss": 0.2584,
      "step": 1355
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.86290822172693,
      "learning_rate": 9.882896601044107e-06,
      "loss": 0.316,
      "step": 1356
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.713126617537648,
      "learning_rate": 9.882647230391194e-06,
      "loss": 0.6937,
      "step": 1357
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.635197733997727,
      "learning_rate": 9.882397597657173e-06,
      "loss": 0.1901,
      "step": 1358
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6553254263807922,
      "learning_rate": 9.88214770285545e-06,
      "loss": 0.2587,
      "step": 1359
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5840421647850982,
      "learning_rate": 9.88189754599943e-06,
      "loss": 0.2402,
      "step": 1360
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.314996874350391,
      "learning_rate": 9.881647127102544e-06,
      "loss": 0.1854,
      "step": 1361
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7767371321777545,
      "learning_rate": 9.881396446178238e-06,
      "loss": 0.2666,
      "step": 1362
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5712443977101482,
      "learning_rate": 9.881145503239959e-06,
      "loss": 0.2217,
      "step": 1363
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5370653074684109,
      "learning_rate": 9.880894298301185e-06,
      "loss": 0.2662,
      "step": 1364
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.782971856972029,
      "learning_rate": 9.880642831375394e-06,
      "loss": 0.2392,
      "step": 1365
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5590543363252756,
      "learning_rate": 9.880391102476085e-06,
      "loss": 0.2323,
      "step": 1366
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4294490476660207,
      "learning_rate": 9.880139111616771e-06,
      "loss": 0.2432,
      "step": 1367
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5086601115971956,
      "learning_rate": 9.87988685881098e-06,
      "loss": 0.2377,
      "step": 1368
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.188232703548946,
      "learning_rate": 9.879634344072245e-06,
      "loss": 0.5783,
      "step": 1369
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5366564476829037,
      "learning_rate": 9.879381567414126e-06,
      "loss": 0.2209,
      "step": 1370
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.810069839209789,
      "learning_rate": 9.879128528850187e-06,
      "loss": 0.2811,
      "step": 1371
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.445279820439034,
      "learning_rate": 9.878875228394016e-06,
      "loss": 0.2288,
      "step": 1372
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5009254399163685,
      "learning_rate": 9.878621666059204e-06,
      "loss": 0.2398,
      "step": 1373
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5208096591918494,
      "learning_rate": 9.878367841859361e-06,
      "loss": 0.267,
      "step": 1374
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5387355246904635,
      "learning_rate": 9.878113755808116e-06,
      "loss": 0.2193,
      "step": 1375
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4428283899541279,
      "learning_rate": 9.877859407919102e-06,
      "loss": 0.2227,
      "step": 1376
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.375032257385307,
      "learning_rate": 9.877604798205975e-06,
      "loss": 0.1773,
      "step": 1377
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4172625933970604,
      "learning_rate": 9.8773499266824e-06,
      "loss": 0.2059,
      "step": 1378
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.598015126783416,
      "learning_rate": 9.877094793362056e-06,
      "loss": 0.2244,
      "step": 1379
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5089052939344718,
      "learning_rate": 9.87683939825864e-06,
      "loss": 0.2684,
      "step": 1380
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4463561033010983,
      "learning_rate": 9.876583741385861e-06,
      "loss": 0.2302,
      "step": 1381
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.378942209157623,
      "learning_rate": 9.87632782275744e-06,
      "loss": 0.2216,
      "step": 1382
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5367213748371558,
      "learning_rate": 9.876071642387113e-06,
      "loss": 0.2238,
      "step": 1383
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3655811912055094,
      "learning_rate": 9.875815200288632e-06,
      "loss": 0.2173,
      "step": 1384
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4302463571759334,
      "learning_rate": 9.875558496475764e-06,
      "loss": 0.2303,
      "step": 1385
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4661578307794918,
      "learning_rate": 9.875301530962282e-06,
      "loss": 0.2152,
      "step": 1386
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4123416509132967,
      "learning_rate": 9.875044303761985e-06,
      "loss": 0.2466,
      "step": 1387
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5824875667225462,
      "learning_rate": 9.874786814888674e-06,
      "loss": 0.2409,
      "step": 1388
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6213635274561233,
      "learning_rate": 9.874529064356177e-06,
      "loss": 0.2641,
      "step": 1389
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.344568638193726,
      "learning_rate": 9.874271052178324e-06,
      "loss": 0.1976,
      "step": 1390
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3848614477717307,
      "learning_rate": 9.874012778368965e-06,
      "loss": 0.21,
      "step": 1391
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4648344774919115,
      "learning_rate": 9.873754242941964e-06,
      "loss": 0.2489,
      "step": 1392
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5944472896231816,
      "learning_rate": 9.873495445911197e-06,
      "loss": 0.2464,
      "step": 1393
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4393610623013384,
      "learning_rate": 9.873236387290557e-06,
      "loss": 0.2328,
      "step": 1394
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.468170863362504,
      "learning_rate": 9.872977067093948e-06,
      "loss": 0.2172,
      "step": 1395
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4834676281640755,
      "learning_rate": 9.872717485335287e-06,
      "loss": 0.2533,
      "step": 1396
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4175652988328173,
      "learning_rate": 9.87245764202851e-06,
      "loss": 0.2021,
      "step": 1397
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7699221445436606,
      "learning_rate": 9.872197537187567e-06,
      "loss": 0.2551,
      "step": 1398
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.739787879716992,
      "learning_rate": 9.871937170826414e-06,
      "loss": 0.6319,
      "step": 1399
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.454609823514992,
      "learning_rate": 9.87167654295903e-06,
      "loss": 0.229,
      "step": 1400
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3478402265288982,
      "learning_rate": 9.871415653599404e-06,
      "loss": 0.2381,
      "step": 1401
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.801904732038938,
      "learning_rate": 9.871154502761536e-06,
      "loss": 0.5994,
      "step": 1402
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3981770010013133,
      "learning_rate": 9.870893090459448e-06,
      "loss": 0.2066,
      "step": 1403
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.359653918232644,
      "learning_rate": 9.87063141670717e-06,
      "loss": 0.1611,
      "step": 1404
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4374560294241465,
      "learning_rate": 9.870369481518749e-06,
      "loss": 0.2492,
      "step": 1405
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5193541537658246,
      "learning_rate": 9.87010728490824e-06,
      "loss": 0.214,
      "step": 1406
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.809640693434221,
      "learning_rate": 9.869844826889723e-06,
      "loss": 0.8256,
      "step": 1407
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3111602627172836,
      "learning_rate": 9.86958210747728e-06,
      "loss": 0.1878,
      "step": 1408
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5615759105051201,
      "learning_rate": 9.869319126685017e-06,
      "loss": 0.2577,
      "step": 1409
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5287341708843745,
      "learning_rate": 9.869055884527047e-06,
      "loss": 0.2202,
      "step": 1410
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6369503175082545,
      "learning_rate": 9.868792381017501e-06,
      "loss": 0.2813,
      "step": 1411
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5117116157951451,
      "learning_rate": 9.868528616170523e-06,
      "loss": 0.2508,
      "step": 1412
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.639172446193039,
      "learning_rate": 9.868264590000272e-06,
      "loss": 0.2187,
      "step": 1413
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7574758403881272,
      "learning_rate": 9.868000302520917e-06,
      "loss": 0.2415,
      "step": 1414
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4144429675892014,
      "learning_rate": 9.867735753746643e-06,
      "loss": 0.1933,
      "step": 1415
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4666066165361122,
      "learning_rate": 9.867470943691654e-06,
      "loss": 0.1935,
      "step": 1416
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6444728998872147,
      "learning_rate": 9.867205872370166e-06,
      "loss": 0.2166,
      "step": 1417
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5836241537079765,
      "learning_rate": 9.866940539796397e-06,
      "loss": 0.2153,
      "step": 1418
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.577980713045395,
      "learning_rate": 9.866674945984599e-06,
      "loss": 0.2176,
      "step": 1419
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5466560768089,
      "learning_rate": 9.866409090949023e-06,
      "loss": 0.2418,
      "step": 1420
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.59109360630513,
      "learning_rate": 9.86614297470394e-06,
      "loss": 0.2586,
      "step": 1421
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.497974793847826,
      "learning_rate": 9.865876597263635e-06,
      "loss": 0.2182,
      "step": 1422
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6881123943416616,
      "learning_rate": 9.865609958642405e-06,
      "loss": 0.2316,
      "step": 1423
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4818849219579886,
      "learning_rate": 9.865343058854563e-06,
      "loss": 0.2174,
      "step": 1424
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5061861040664406,
      "learning_rate": 9.865075897914433e-06,
      "loss": 0.194,
      "step": 1425
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4974686259819203,
      "learning_rate": 9.864808475836358e-06,
      "loss": 0.2447,
      "step": 1426
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5112938625335517,
      "learning_rate": 9.86454079263469e-06,
      "loss": 0.2066,
      "step": 1427
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9084662965686543,
      "learning_rate": 9.8642728483238e-06,
      "loss": 0.2695,
      "step": 1428
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4207834195531535,
      "learning_rate": 9.864004642918066e-06,
      "loss": 0.2163,
      "step": 1429
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.3265899699538,
      "learning_rate": 9.863736176431888e-06,
      "loss": 0.6267,
      "step": 1430
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.317240798229465,
      "learning_rate": 9.863467448879674e-06,
      "loss": 0.8054,
      "step": 1431
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4032825742106048,
      "learning_rate": 9.863198460275849e-06,
      "loss": 0.2002,
      "step": 1432
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5301223322063986,
      "learning_rate": 9.86292921063485e-06,
      "loss": 0.2366,
      "step": 1433
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5369791845645449,
      "learning_rate": 9.862659699971132e-06,
      "loss": 0.234,
      "step": 1434
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5355685919818973,
      "learning_rate": 9.862389928299159e-06,
      "loss": 0.2725,
      "step": 1435
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4940004911644824,
      "learning_rate": 9.862119895633411e-06,
      "loss": 0.2084,
      "step": 1436
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.644860330097462,
      "learning_rate": 9.861849601988384e-06,
      "loss": 0.2277,
      "step": 1437
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6286894781024046,
      "learning_rate": 9.861579047378586e-06,
      "loss": 0.2452,
      "step": 1438
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.454955867530021,
      "learning_rate": 9.861308231818539e-06,
      "loss": 0.2156,
      "step": 1439
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.690383889166342,
      "learning_rate": 9.861037155322777e-06,
      "loss": 0.2032,
      "step": 1440
    },
    {
      "epoch": 0.1,
      "grad_norm": 7.819750828463083,
      "learning_rate": 9.860765817905852e-06,
      "loss": 0.7353,
      "step": 1441
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.421978227146228,
      "learning_rate": 9.860494219582331e-06,
      "loss": 0.2123,
      "step": 1442
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3351254143219158,
      "learning_rate": 9.86022236036679e-06,
      "loss": 0.1875,
      "step": 1443
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7989610752793848,
      "learning_rate": 9.859950240273822e-06,
      "loss": 0.2518,
      "step": 1444
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3956003870777745,
      "learning_rate": 9.85967785931803e-06,
      "loss": 0.2649,
      "step": 1445
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6201214330558396,
      "learning_rate": 9.859405217514039e-06,
      "loss": 0.2396,
      "step": 1446
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4973721987809194,
      "learning_rate": 9.859132314876481e-06,
      "loss": 0.2138,
      "step": 1447
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5078236735255501,
      "learning_rate": 9.858859151420003e-06,
      "loss": 0.2748,
      "step": 1448
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6282415931031449,
      "learning_rate": 9.85858572715927e-06,
      "loss": 0.2332,
      "step": 1449
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.454121187430639,
      "learning_rate": 9.85831204210896e-06,
      "loss": 0.2139,
      "step": 1450
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6401966087079514,
      "learning_rate": 9.858038096283758e-06,
      "loss": 0.2529,
      "step": 1451
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3924547392692403,
      "learning_rate": 9.85776388969837e-06,
      "loss": 0.2163,
      "step": 1452
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.530921255774927,
      "learning_rate": 9.857489422367516e-06,
      "loss": 0.2236,
      "step": 1453
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4774001958307623,
      "learning_rate": 9.857214694305928e-06,
      "loss": 0.2093,
      "step": 1454
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5306239890154436,
      "learning_rate": 9.856939705528352e-06,
      "loss": 0.2222,
      "step": 1455
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4937244683186444,
      "learning_rate": 9.856664456049549e-06,
      "loss": 0.2622,
      "step": 1456
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5147168265999564,
      "learning_rate": 9.856388945884292e-06,
      "loss": 0.2377,
      "step": 1457
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5064246101230958,
      "learning_rate": 9.856113175047369e-06,
      "loss": 0.2114,
      "step": 1458
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.410979062646319,
      "learning_rate": 9.855837143553585e-06,
      "loss": 0.2166,
      "step": 1459
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7487139270100134,
      "learning_rate": 9.855560851417752e-06,
      "loss": 0.2657,
      "step": 1460
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5749822438510075,
      "learning_rate": 9.855284298654703e-06,
      "loss": 0.254,
      "step": 1461
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3581809626850272,
      "learning_rate": 9.855007485279282e-06,
      "loss": 0.1948,
      "step": 1462
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6604290188908999,
      "learning_rate": 9.854730411306347e-06,
      "loss": 0.2504,
      "step": 1463
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4272296751612261,
      "learning_rate": 9.854453076750772e-06,
      "loss": 0.2377,
      "step": 1464
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.351974537197615,
      "learning_rate": 9.85417548162744e-06,
      "loss": 0.6479,
      "step": 1465
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3645267990815164,
      "learning_rate": 9.853897625951254e-06,
      "loss": 0.19,
      "step": 1466
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5737788045126675,
      "learning_rate": 9.853619509737125e-06,
      "loss": 0.2397,
      "step": 1467
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.378486674344173,
      "learning_rate": 9.853341132999985e-06,
      "loss": 0.1968,
      "step": 1468
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.346890235286319,
      "learning_rate": 9.853062495754772e-06,
      "loss": 0.7932,
      "step": 1469
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.408982958566723,
      "learning_rate": 9.852783598016447e-06,
      "loss": 0.2296,
      "step": 1470
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.488275838404948,
      "learning_rate": 9.852504439799978e-06,
      "loss": 0.6584,
      "step": 1471
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5125708137449658,
      "learning_rate": 9.852225021120345e-06,
      "loss": 0.2032,
      "step": 1472
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7434891700545965,
      "learning_rate": 9.851945341992552e-06,
      "loss": 0.2442,
      "step": 1473
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.576583034324645,
      "learning_rate": 9.85166540243161e-06,
      "loss": 0.2309,
      "step": 1474
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3906884559484975,
      "learning_rate": 9.85138520245254e-06,
      "loss": 0.2111,
      "step": 1475
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.526006313461169,
      "learning_rate": 9.85110474207039e-06,
      "loss": 0.1961,
      "step": 1476
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.551682966981367,
      "learning_rate": 9.850824021300208e-06,
      "loss": 0.1879,
      "step": 1477
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4366966787017894,
      "learning_rate": 9.850543040157064e-06,
      "loss": 0.2292,
      "step": 1478
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.482174174932269,
      "learning_rate": 9.85026179865604e-06,
      "loss": 0.2234,
      "step": 1479
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3565803059264792,
      "learning_rate": 9.849980296812231e-06,
      "loss": 0.1874,
      "step": 1480
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.547091982246654,
      "learning_rate": 9.849698534640748e-06,
      "loss": 0.2555,
      "step": 1481
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.468160034672372,
      "learning_rate": 9.849416512156716e-06,
      "loss": 0.2528,
      "step": 1482
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4671801075807596,
      "learning_rate": 9.84913422937527e-06,
      "loss": 0.213,
      "step": 1483
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.719235216958124,
      "learning_rate": 9.848851686311562e-06,
      "loss": 0.4794,
      "step": 1484
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3899382537601066,
      "learning_rate": 9.84856888298076e-06,
      "loss": 0.1926,
      "step": 1485
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4726749722108121,
      "learning_rate": 9.848285819398044e-06,
      "loss": 0.2437,
      "step": 1486
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4163095491147437,
      "learning_rate": 9.848002495578606e-06,
      "loss": 0.197,
      "step": 1487
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6454671654988682,
      "learning_rate": 9.847718911537652e-06,
      "loss": 0.2429,
      "step": 1488
    },
    {
      "epoch": 0.11,
      "grad_norm": 30.914938053365645,
      "learning_rate": 9.847435067290407e-06,
      "loss": 0.6408,
      "step": 1489
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4760425009869969,
      "learning_rate": 9.847150962852105e-06,
      "loss": 0.2325,
      "step": 1490
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4737565855294696,
      "learning_rate": 9.846866598238e-06,
      "loss": 0.2088,
      "step": 1491
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.493551648435666,
      "learning_rate": 9.846581973463346e-06,
      "loss": 0.2349,
      "step": 1492
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.534414687676963,
      "learning_rate": 9.846297088543428e-06,
      "loss": 0.2201,
      "step": 1493
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4602197898074398,
      "learning_rate": 9.846011943493536e-06,
      "loss": 0.2247,
      "step": 1494
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5233278091529827,
      "learning_rate": 9.845726538328975e-06,
      "loss": 0.2489,
      "step": 1495
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.482065377638207,
      "learning_rate": 9.845440873065063e-06,
      "loss": 0.2263,
      "step": 1496
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3919579904472246,
      "learning_rate": 9.845154947717137e-06,
      "loss": 0.1904,
      "step": 1497
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4439690021659912,
      "learning_rate": 9.844868762300542e-06,
      "loss": 0.2261,
      "step": 1498
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5338843773573032,
      "learning_rate": 9.844582316830639e-06,
      "loss": 0.2335,
      "step": 1499
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.416272077405108,
      "learning_rate": 9.844295611322804e-06,
      "loss": 0.2265,
      "step": 1500
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3908467277223857,
      "learning_rate": 9.844008645792425e-06,
      "loss": 0.2211,
      "step": 1501
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.545979264940256,
      "learning_rate": 9.843721420254908e-06,
      "loss": 0.2421,
      "step": 1502
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6487964807360322,
      "learning_rate": 9.843433934725665e-06,
      "loss": 0.2122,
      "step": 1503
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3254313451364343,
      "learning_rate": 9.843146189220133e-06,
      "loss": 0.2237,
      "step": 1504
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6098553246456513,
      "learning_rate": 9.842858183753754e-06,
      "loss": 0.2389,
      "step": 1505
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.484733036630676,
      "learning_rate": 9.842569918341988e-06,
      "loss": 0.2302,
      "step": 1506
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6277556106239954,
      "learning_rate": 9.842281393000306e-06,
      "loss": 0.2237,
      "step": 1507
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.63480555826644,
      "learning_rate": 9.841992607744198e-06,
      "loss": 0.5224,
      "step": 1508
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6298117657719016,
      "learning_rate": 9.84170356258916e-06,
      "loss": 0.3044,
      "step": 1509
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.475429486196651,
      "learning_rate": 9.841414257550712e-06,
      "loss": 0.2142,
      "step": 1510
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5859147463782726,
      "learning_rate": 9.84112469264438e-06,
      "loss": 0.2087,
      "step": 1511
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4264245655539785,
      "learning_rate": 9.840834867885706e-06,
      "loss": 0.1594,
      "step": 1512
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6378256740072248,
      "learning_rate": 9.84054478329025e-06,
      "loss": 0.2605,
      "step": 1513
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.438318811935071,
      "learning_rate": 9.84025443887358e-06,
      "loss": 0.171,
      "step": 1514
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5217683214874616,
      "learning_rate": 9.83996383465128e-06,
      "loss": 0.2561,
      "step": 1515
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.532853255708979,
      "learning_rate": 9.83967297063895e-06,
      "loss": 0.2118,
      "step": 1516
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5956952054895646,
      "learning_rate": 9.839381846852203e-06,
      "loss": 0.2602,
      "step": 1517
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4898384063066419,
      "learning_rate": 9.839090463306663e-06,
      "loss": 0.1995,
      "step": 1518
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6585037479244904,
      "learning_rate": 9.838798820017972e-06,
      "loss": 0.2107,
      "step": 1519
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5929778340254797,
      "learning_rate": 9.838506917001784e-06,
      "loss": 0.2116,
      "step": 1520
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4636014751855189,
      "learning_rate": 9.838214754273768e-06,
      "loss": 0.2512,
      "step": 1521
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5910332107395206,
      "learning_rate": 9.837922331849604e-06,
      "loss": 0.2659,
      "step": 1522
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.626336220017116,
      "learning_rate": 9.837629649744989e-06,
      "loss": 0.6054,
      "step": 1523
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4785843323445005,
      "learning_rate": 9.837336707975633e-06,
      "loss": 0.2705,
      "step": 1524
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4834704203724265,
      "learning_rate": 9.837043506557261e-06,
      "loss": 0.2256,
      "step": 1525
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.489207975590961,
      "learning_rate": 9.83675004550561e-06,
      "loss": 0.7058,
      "step": 1526
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6301629330971392,
      "learning_rate": 9.836456324836432e-06,
      "loss": 0.166,
      "step": 1527
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.345588468704059,
      "learning_rate": 9.836162344565493e-06,
      "loss": 0.6935,
      "step": 1528
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.669816651205717,
      "learning_rate": 9.835868104708572e-06,
      "loss": 0.231,
      "step": 1529
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4432564199624374,
      "learning_rate": 9.835573605281462e-06,
      "loss": 0.2332,
      "step": 1530
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.145547683477056,
      "learning_rate": 9.835278846299974e-06,
      "loss": 0.5424,
      "step": 1531
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.39277031606631,
      "learning_rate": 9.834983827779926e-06,
      "loss": 0.2107,
      "step": 1532
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4512375794572607,
      "learning_rate": 9.834688549737154e-06,
      "loss": 0.2153,
      "step": 1533
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.347347554996215,
      "learning_rate": 9.83439301218751e-06,
      "loss": 0.2253,
      "step": 1534
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3406302593803636,
      "learning_rate": 9.834097215146851e-06,
      "loss": 0.2351,
      "step": 1535
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4771306942863593,
      "learning_rate": 9.833801158631062e-06,
      "loss": 0.1979,
      "step": 1536
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5263956381612158,
      "learning_rate": 9.833504842656029e-06,
      "loss": 0.2476,
      "step": 1537
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3429885859612332,
      "learning_rate": 9.833208267237658e-06,
      "loss": 0.2288,
      "step": 1538
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8363183773471008,
      "learning_rate": 9.83291143239187e-06,
      "loss": 0.295,
      "step": 1539
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4662700259047743,
      "learning_rate": 9.832614338134595e-06,
      "loss": 0.2282,
      "step": 1540
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.585094546474449,
      "learning_rate": 9.832316984481783e-06,
      "loss": 0.2283,
      "step": 1541
    },
    {
      "epoch": 0.11,
      "grad_norm": 12.286584560215076,
      "learning_rate": 9.832019371449391e-06,
      "loss": 0.6859,
      "step": 1542
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7799698311643788,
      "learning_rate": 9.831721499053397e-06,
      "loss": 0.2964,
      "step": 1543
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.431979054902501,
      "learning_rate": 9.831423367309788e-06,
      "loss": 0.2644,
      "step": 1544
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.424940258021316,
      "learning_rate": 9.831124976234568e-06,
      "loss": 0.2053,
      "step": 1545
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5666413112239748,
      "learning_rate": 9.83082632584375e-06,
      "loss": 0.2366,
      "step": 1546
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4622931146485236,
      "learning_rate": 9.830527416153368e-06,
      "loss": 0.2482,
      "step": 1547
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6568808251615852,
      "learning_rate": 9.830228247179465e-06,
      "loss": 0.2481,
      "step": 1548
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5658953099891781,
      "learning_rate": 9.8299288189381e-06,
      "loss": 0.2155,
      "step": 1549
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4056570897983416,
      "learning_rate": 9.829629131445342e-06,
      "loss": 0.2109,
      "step": 1550
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5857742306507916,
      "learning_rate": 9.829329184717281e-06,
      "loss": 0.2299,
      "step": 1551
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.676225632467976,
      "learning_rate": 9.829028978770015e-06,
      "loss": 0.2943,
      "step": 1552
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5838331474659668,
      "learning_rate": 9.828728513619659e-06,
      "loss": 0.2818,
      "step": 1553
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6099143879898696,
      "learning_rate": 9.828427789282337e-06,
      "loss": 0.2546,
      "step": 1554
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5127258143455455,
      "learning_rate": 9.828126805774197e-06,
      "loss": 0.2379,
      "step": 1555
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4977762823766838,
      "learning_rate": 9.827825563111389e-06,
      "loss": 0.2343,
      "step": 1556
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.466637582859912,
      "learning_rate": 9.827524061310085e-06,
      "loss": 0.2714,
      "step": 1557
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3445130615270007,
      "learning_rate": 9.82722230038647e-06,
      "loss": 0.1946,
      "step": 1558
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4335509391352084,
      "learning_rate": 9.826920280356737e-06,
      "loss": 0.2348,
      "step": 1559
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.73324219308033,
      "learning_rate": 9.826618001237101e-06,
      "loss": 0.653,
      "step": 1560
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4494109190769389,
      "learning_rate": 9.826315463043785e-06,
      "loss": 0.2031,
      "step": 1561
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5854440510035475,
      "learning_rate": 9.82601266579303e-06,
      "loss": 0.2499,
      "step": 1562
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.7349652784427265,
      "learning_rate": 9.825709609501088e-06,
      "loss": 0.64,
      "step": 1563
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3863898822031804,
      "learning_rate": 9.825406294184225e-06,
      "loss": 0.2296,
      "step": 1564
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5376796384950608,
      "learning_rate": 9.825102719858722e-06,
      "loss": 0.2541,
      "step": 1565
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.511888999228989,
      "learning_rate": 9.824798886540874e-06,
      "loss": 0.2072,
      "step": 1566
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.371233568847546,
      "learning_rate": 9.82449479424699e-06,
      "loss": 0.7218,
      "step": 1567
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6560407943327873,
      "learning_rate": 9.824190442993392e-06,
      "loss": 0.2115,
      "step": 1568
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4589841234716807,
      "learning_rate": 9.823885832796419e-06,
      "loss": 0.2473,
      "step": 1569
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4226470373261548,
      "learning_rate": 9.823580963672415e-06,
      "loss": 0.2578,
      "step": 1570
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6943826821235448,
      "learning_rate": 9.82327583563775e-06,
      "loss": 0.2338,
      "step": 1571
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.369131334580333,
      "learning_rate": 9.822970448708799e-06,
      "loss": 0.1939,
      "step": 1572
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5648149620275735,
      "learning_rate": 9.822664802901956e-06,
      "loss": 0.2377,
      "step": 1573
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2842799330598096,
      "learning_rate": 9.822358898233625e-06,
      "loss": 0.2021,
      "step": 1574
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4681206066600627,
      "learning_rate": 9.822052734720228e-06,
      "loss": 0.2358,
      "step": 1575
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3934825890155422,
      "learning_rate": 9.821746312378196e-06,
      "loss": 0.2305,
      "step": 1576
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5973548165869802,
      "learning_rate": 9.821439631223978e-06,
      "loss": 0.2542,
      "step": 1577
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.5638008040552105,
      "learning_rate": 9.821132691274035e-06,
      "loss": 0.5038,
      "step": 1578
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3709727651556627,
      "learning_rate": 9.820825492544842e-06,
      "loss": 0.1994,
      "step": 1579
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5316623562308376,
      "learning_rate": 9.82051803505289e-06,
      "loss": 0.2622,
      "step": 1580
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5586460884475626,
      "learning_rate": 9.820210318814679e-06,
      "loss": 0.2422,
      "step": 1581
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4980698420031409,
      "learning_rate": 9.81990234384673e-06,
      "loss": 0.2199,
      "step": 1582
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.471700998427309,
      "learning_rate": 9.81959411016557e-06,
      "loss": 0.5387,
      "step": 1583
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7200792458866005,
      "learning_rate": 9.819285617787746e-06,
      "loss": 0.236,
      "step": 1584
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5252434679303168,
      "learning_rate": 9.818976866729814e-06,
      "loss": 0.1757,
      "step": 1585
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3117876732376015,
      "learning_rate": 9.818667857008352e-06,
      "loss": 0.2159,
      "step": 1586
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6017841907018093,
      "learning_rate": 9.81835858863994e-06,
      "loss": 0.2623,
      "step": 1587
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.623047665308878,
      "learning_rate": 9.818049061641183e-06,
      "loss": 0.2483,
      "step": 1588
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5876520427908178,
      "learning_rate": 9.817739276028694e-06,
      "loss": 0.253,
      "step": 1589
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5858352347151505,
      "learning_rate": 9.8174292318191e-06,
      "loss": 0.2391,
      "step": 1590
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.441814385223122,
      "learning_rate": 9.817118929029042e-06,
      "loss": 0.2205,
      "step": 1591
    },
    {
      "epoch": 0.11,
      "grad_norm": 10.696672970972553,
      "learning_rate": 9.816808367675179e-06,
      "loss": 0.6145,
      "step": 1592
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4618408635629998,
      "learning_rate": 9.816497547774178e-06,
      "loss": 0.2008,
      "step": 1593
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4357087994743714,
      "learning_rate": 9.816186469342724e-06,
      "loss": 0.2358,
      "step": 1594
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3465233183511685,
      "learning_rate": 9.815875132397515e-06,
      "loss": 0.2013,
      "step": 1595
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8198660558733726,
      "learning_rate": 9.81556353695526e-06,
      "loss": 0.2367,
      "step": 1596
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.568972485691452,
      "learning_rate": 9.815251683032684e-06,
      "loss": 0.2194,
      "step": 1597
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5877705447355996,
      "learning_rate": 9.814939570646532e-06,
      "loss": 0.218,
      "step": 1598
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4760635293787112,
      "learning_rate": 9.814627199813549e-06,
      "loss": 0.2302,
      "step": 1599
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4058410956326552,
      "learning_rate": 9.814314570550506e-06,
      "loss": 0.2242,
      "step": 1600
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.374485548166533,
      "learning_rate": 9.814001682874183e-06,
      "loss": 0.1985,
      "step": 1601
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2933922310576191,
      "learning_rate": 9.813688536801375e-06,
      "loss": 0.1892,
      "step": 1602
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.425693142424464,
      "learning_rate": 9.813375132348893e-06,
      "loss": 0.189,
      "step": 1603
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5217491709222601,
      "learning_rate": 9.813061469533552e-06,
      "loss": 0.2037,
      "step": 1604
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.403142426244644,
      "learning_rate": 9.812747548372195e-06,
      "loss": 0.7243,
      "step": 1605
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3099323250937103,
      "learning_rate": 9.81243336888167e-06,
      "loss": 0.1915,
      "step": 1606
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6004250497925583,
      "learning_rate": 9.812118931078837e-06,
      "loss": 0.2336,
      "step": 1607
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.208773259980145,
      "learning_rate": 9.811804234980583e-06,
      "loss": 0.6985,
      "step": 1608
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5875724798666768,
      "learning_rate": 9.81148928060379e-06,
      "loss": 0.2079,
      "step": 1609
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5450388873752503,
      "learning_rate": 9.811174067965367e-06,
      "loss": 0.2655,
      "step": 1610
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.700370463813713,
      "learning_rate": 9.810858597082235e-06,
      "loss": 0.2399,
      "step": 1611
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.44645809705951,
      "learning_rate": 9.810542867971326e-06,
      "loss": 0.2399,
      "step": 1612
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.620041395369137,
      "learning_rate": 9.810226880649589e-06,
      "loss": 0.7007,
      "step": 1613
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4216453626938492,
      "learning_rate": 9.80991063513398e-06,
      "loss": 0.1686,
      "step": 1614
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.112719232669365,
      "learning_rate": 9.809594131441479e-06,
      "loss": 0.2593,
      "step": 1615
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5865792604852162,
      "learning_rate": 9.809277369589072e-06,
      "loss": 0.2665,
      "step": 1616
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.438636177630681,
      "learning_rate": 9.808960349593761e-06,
      "loss": 0.2438,
      "step": 1617
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4194285898096355,
      "learning_rate": 9.808643071472564e-06,
      "loss": 0.2208,
      "step": 1618
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.365753865598493,
      "learning_rate": 9.80832553524251e-06,
      "loss": 0.1963,
      "step": 1619
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7057091545871734,
      "learning_rate": 9.808007740920647e-06,
      "loss": 0.2424,
      "step": 1620
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4806090639600953,
      "learning_rate": 9.807689688524027e-06,
      "loss": 0.2119,
      "step": 1621
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3747608366564992,
      "learning_rate": 9.807371378069723e-06,
      "loss": 0.1633,
      "step": 1622
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5289310003563972,
      "learning_rate": 9.807052809574824e-06,
      "loss": 0.2821,
      "step": 1623
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.94711764436591,
      "learning_rate": 9.806733983056429e-06,
      "loss": 0.4993,
      "step": 1624
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.458270531451846,
      "learning_rate": 9.806414898531647e-06,
      "loss": 0.247,
      "step": 1625
    },
    {
      "epoch": 0.12,
      "grad_norm": 7.4652526167248725,
      "learning_rate": 9.80609555601761e-06,
      "loss": 0.7849,
      "step": 1626
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4756719514712213,
      "learning_rate": 9.805775955531456e-06,
      "loss": 0.2282,
      "step": 1627
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.396563422642613,
      "learning_rate": 9.805456097090343e-06,
      "loss": 0.213,
      "step": 1628
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.488272823795172,
      "learning_rate": 9.805135980711436e-06,
      "loss": 0.2653,
      "step": 1629
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.999818599273434,
      "learning_rate": 9.80481560641192e-06,
      "loss": 0.5572,
      "step": 1630
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4526772851541578,
      "learning_rate": 9.80449497420899e-06,
      "loss": 0.1932,
      "step": 1631
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4891417212197875,
      "learning_rate": 9.804174084119857e-06,
      "loss": 0.222,
      "step": 1632
    },
    {
      "epoch": 0.12,
      "grad_norm": 38.74248963793999,
      "learning_rate": 9.803852936161747e-06,
      "loss": 0.6564,
      "step": 1633
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5068881197059105,
      "learning_rate": 9.803531530351894e-06,
      "loss": 0.2604,
      "step": 1634
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.521933240760097,
      "learning_rate": 9.803209866707553e-06,
      "loss": 0.2484,
      "step": 1635
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3377806412565296,
      "learning_rate": 9.802887945245987e-06,
      "loss": 0.1848,
      "step": 1636
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.569442325202885,
      "learning_rate": 9.80256576598448e-06,
      "loss": 0.2314,
      "step": 1637
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5517290957281664,
      "learning_rate": 9.802243328940319e-06,
      "loss": 0.2716,
      "step": 1638
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3700517009054844,
      "learning_rate": 9.801920634130816e-06,
      "loss": 0.1853,
      "step": 1639
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5748658447719495,
      "learning_rate": 9.80159768157329e-06,
      "loss": 0.2476,
      "step": 1640
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4859273474333772,
      "learning_rate": 9.801274471285076e-06,
      "loss": 0.1984,
      "step": 1641
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6768396259375484,
      "learning_rate": 9.800951003283522e-06,
      "loss": 0.2931,
      "step": 1642
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3055000353275865,
      "learning_rate": 9.800627277585995e-06,
      "loss": 0.2,
      "step": 1643
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5995618887871297,
      "learning_rate": 9.800303294209865e-06,
      "loss": 0.2768,
      "step": 1644
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4494311430660118,
      "learning_rate": 9.799979053172524e-06,
      "loss": 0.2049,
      "step": 1645
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.371483479323944,
      "learning_rate": 9.799654554491379e-06,
      "loss": 0.2096,
      "step": 1646
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4368264268889892,
      "learning_rate": 9.799329798183843e-06,
      "loss": 0.2214,
      "step": 1647
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3510505877201295,
      "learning_rate": 9.799004784267352e-06,
      "loss": 0.2271,
      "step": 1648
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4399895824051956,
      "learning_rate": 9.798679512759347e-06,
      "loss": 0.1996,
      "step": 1649
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3674449587836637,
      "learning_rate": 9.798353983677294e-06,
      "loss": 0.207,
      "step": 1650
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.537947248078371,
      "learning_rate": 9.798028197038657e-06,
      "loss": 0.2456,
      "step": 1651
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.57858660425834,
      "learning_rate": 9.797702152860932e-06,
      "loss": 0.2545,
      "step": 1652
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3461232070921907,
      "learning_rate": 9.797375851161615e-06,
      "loss": 0.2155,
      "step": 1653
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3409813556752181,
      "learning_rate": 9.79704929195822e-06,
      "loss": 0.1998,
      "step": 1654
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3936042232648944,
      "learning_rate": 9.796722475268278e-06,
      "loss": 0.1869,
      "step": 1655
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4930864115753177,
      "learning_rate": 9.79639540110933e-06,
      "loss": 0.1855,
      "step": 1656
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.618483161995016,
      "learning_rate": 9.796068069498931e-06,
      "loss": 0.2222,
      "step": 1657
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.552176239190835,
      "learning_rate": 9.795740480454651e-06,
      "loss": 0.205,
      "step": 1658
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5505038348518316,
      "learning_rate": 9.795412633994077e-06,
      "loss": 0.2486,
      "step": 1659
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.326956273328569,
      "learning_rate": 9.795084530134801e-06,
      "loss": 0.1742,
      "step": 1660
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2964390822908187,
      "learning_rate": 9.794756168894439e-06,
      "loss": 0.1951,
      "step": 1661
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.495084984643208,
      "learning_rate": 9.794427550290615e-06,
      "loss": 0.2617,
      "step": 1662
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3584011835260723,
      "learning_rate": 9.794098674340966e-06,
      "loss": 0.1905,
      "step": 1663
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3101838433835598,
      "learning_rate": 9.793769541063147e-06,
      "loss": 0.1766,
      "step": 1664
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4821449750998308,
      "learning_rate": 9.793440150474822e-06,
      "loss": 0.1994,
      "step": 1665
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5526190114019054,
      "learning_rate": 9.793110502593674e-06,
      "loss": 0.255,
      "step": 1666
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2332512322109075,
      "learning_rate": 9.792780597437397e-06,
      "loss": 0.2156,
      "step": 1667
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5703369628987611,
      "learning_rate": 9.792450435023699e-06,
      "loss": 0.2306,
      "step": 1668
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4000080992387918,
      "learning_rate": 9.7921200153703e-06,
      "loss": 0.2222,
      "step": 1669
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4235960362370275,
      "learning_rate": 9.791789338494937e-06,
      "loss": 0.2029,
      "step": 1670
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.562190153953838,
      "learning_rate": 9.791458404415357e-06,
      "loss": 0.6406,
      "step": 1671
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4883614268854848,
      "learning_rate": 9.791127213149328e-06,
      "loss": 0.2228,
      "step": 1672
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3960957996252932,
      "learning_rate": 9.790795764714622e-06,
      "loss": 0.1991,
      "step": 1673
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.653576076591192,
      "learning_rate": 9.790464059129034e-06,
      "loss": 0.6018,
      "step": 1674
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.537778401628917,
      "learning_rate": 9.790132096410366e-06,
      "loss": 0.2336,
      "step": 1675
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3946258826829696,
      "learning_rate": 9.789799876576438e-06,
      "loss": 0.2562,
      "step": 1676
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.293227697243076,
      "learning_rate": 9.789467399645081e-06,
      "loss": 0.6905,
      "step": 1677
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.673971303387585,
      "learning_rate": 9.789134665634142e-06,
      "loss": 0.2008,
      "step": 1678
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5293938763424837,
      "learning_rate": 9.788801674561479e-06,
      "loss": 0.1998,
      "step": 1679
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5169792737748706,
      "learning_rate": 9.788468426444968e-06,
      "loss": 0.2187,
      "step": 1680
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4733994479348527,
      "learning_rate": 9.788134921302495e-06,
      "loss": 0.2478,
      "step": 1681
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7553052184536297,
      "learning_rate": 9.787801159151962e-06,
      "loss": 0.2492,
      "step": 1682
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3746876644012884,
      "learning_rate": 9.787467140011284e-06,
      "loss": 0.2197,
      "step": 1683
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.605912095347559,
      "learning_rate": 9.78713286389839e-06,
      "loss": 0.2589,
      "step": 1684
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3788070834718391,
      "learning_rate": 9.78679833083122e-06,
      "loss": 0.2414,
      "step": 1685
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.420012774090065,
      "learning_rate": 9.786463540827734e-06,
      "loss": 0.1933,
      "step": 1686
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5492844611671754,
      "learning_rate": 9.7861284939059e-06,
      "loss": 0.2465,
      "step": 1687
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5474295324375043,
      "learning_rate": 9.785793190083705e-06,
      "loss": 0.2368,
      "step": 1688
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5570560163743425,
      "learning_rate": 9.785457629379143e-06,
      "loss": 0.2175,
      "step": 1689
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3188894238129203,
      "learning_rate": 9.785121811810225e-06,
      "loss": 0.1809,
      "step": 1690
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5685988685450034,
      "learning_rate": 9.784785737394982e-06,
      "loss": 0.2384,
      "step": 1691
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.277813732749272,
      "learning_rate": 9.784449406151448e-06,
      "loss": 0.5431,
      "step": 1692
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5102070798738463,
      "learning_rate": 9.784112818097678e-06,
      "loss": 0.2111,
      "step": 1693
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5958130357216775,
      "learning_rate": 9.783775973251737e-06,
      "loss": 0.226,
      "step": 1694
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5472157649871132,
      "learning_rate": 9.783438871631708e-06,
      "loss": 0.2413,
      "step": 1695
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.574898315776022,
      "learning_rate": 9.783101513255681e-06,
      "loss": 0.2564,
      "step": 1696
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6031253781044434,
      "learning_rate": 9.78276389814177e-06,
      "loss": 0.2559,
      "step": 1697
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.331173793948848,
      "learning_rate": 9.782426026308094e-06,
      "loss": 0.4703,
      "step": 1698
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.506905336500835,
      "learning_rate": 9.782087897772787e-06,
      "loss": 0.2385,
      "step": 1699
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6377424364641004,
      "learning_rate": 9.781749512554e-06,
      "loss": 0.2384,
      "step": 1700
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5610477285967144,
      "learning_rate": 9.781410870669896e-06,
      "loss": 0.2117,
      "step": 1701
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.530807762671212,
      "learning_rate": 9.781071972138654e-06,
      "loss": 0.2559,
      "step": 1702
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5709572317569678,
      "learning_rate": 9.780732816978462e-06,
      "loss": 0.2066,
      "step": 1703
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6710030826487932,
      "learning_rate": 9.780393405207524e-06,
      "loss": 0.2208,
      "step": 1704
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5839199379682274,
      "learning_rate": 9.78005373684406e-06,
      "loss": 0.2244,
      "step": 1705
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6121870410905914,
      "learning_rate": 9.7797138119063e-06,
      "loss": 0.2607,
      "step": 1706
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3095778207377542,
      "learning_rate": 9.779373630412493e-06,
      "loss": 0.1767,
      "step": 1707
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5897047503971178,
      "learning_rate": 9.779033192380897e-06,
      "loss": 0.2321,
      "step": 1708
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.180408482820496,
      "learning_rate": 9.778692497829786e-06,
      "loss": 0.5975,
      "step": 1709
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5784143261414403,
      "learning_rate": 9.778351546777445e-06,
      "loss": 0.2675,
      "step": 1710
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.463473005063729,
      "learning_rate": 9.778010339242178e-06,
      "loss": 0.2393,
      "step": 1711
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5795846447636275,
      "learning_rate": 9.777668875242296e-06,
      "loss": 0.2475,
      "step": 1712
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.160087427735706,
      "learning_rate": 9.777327154796131e-06,
      "loss": 0.599,
      "step": 1713
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7361205806389284,
      "learning_rate": 9.776985177922022e-06,
      "loss": 0.2808,
      "step": 1714
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6872275639199157,
      "learning_rate": 9.776642944638328e-06,
      "loss": 0.2479,
      "step": 1715
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.398505760704376,
      "learning_rate": 9.776300454963417e-06,
      "loss": 0.2149,
      "step": 1716
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4211355219914745,
      "learning_rate": 9.775957708915674e-06,
      "loss": 0.1917,
      "step": 1717
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.763358612769277,
      "learning_rate": 9.775614706513495e-06,
      "loss": 0.5881,
      "step": 1718
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4396613813075212,
      "learning_rate": 9.775271447775289e-06,
      "loss": 0.2025,
      "step": 1719
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.0827953130407915,
      "learning_rate": 9.774927932719484e-06,
      "loss": 0.7363,
      "step": 1720
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5086118597374356,
      "learning_rate": 9.774584161364517e-06,
      "loss": 0.2313,
      "step": 1721
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6470142925591267,
      "learning_rate": 9.77424013372884e-06,
      "loss": 0.2387,
      "step": 1722
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2810149240468711,
      "learning_rate": 9.773895849830922e-06,
      "loss": 0.1739,
      "step": 1723
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5627294806312633,
      "learning_rate": 9.773551309689237e-06,
      "loss": 0.2443,
      "step": 1724
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.211473621619358,
      "learning_rate": 9.773206513322284e-06,
      "loss": 0.5538,
      "step": 1725
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7302147571694546,
      "learning_rate": 9.772861460748568e-06,
      "loss": 0.2359,
      "step": 1726
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.515940136314209,
      "learning_rate": 9.772516151986612e-06,
      "loss": 0.2424,
      "step": 1727
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4052992604018248,
      "learning_rate": 9.772170587054948e-06,
      "loss": 0.1985,
      "step": 1728
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4020074968576497,
      "learning_rate": 9.771824765972124e-06,
      "loss": 0.1953,
      "step": 1729
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2730540696667596,
      "learning_rate": 9.771478688756706e-06,
      "loss": 0.211,
      "step": 1730
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6253760509498694,
      "learning_rate": 9.771132355427269e-06,
      "loss": 0.2608,
      "step": 1731
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5770357867311013,
      "learning_rate": 9.7707857660024e-06,
      "loss": 0.2575,
      "step": 1732
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5051016396234373,
      "learning_rate": 9.770438920500705e-06,
      "loss": 0.21,
      "step": 1733
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3931984271292237,
      "learning_rate": 9.7700918189408e-06,
      "loss": 0.2131,
      "step": 1734
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5452160295887971,
      "learning_rate": 9.769744461341317e-06,
      "loss": 0.2604,
      "step": 1735
    },
    {
      "epoch": 0.12,
      "grad_norm": 8.711680271083882,
      "learning_rate": 9.769396847720901e-06,
      "loss": 0.5922,
      "step": 1736
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6180297131926675,
      "learning_rate": 9.76904897809821e-06,
      "loss": 0.2631,
      "step": 1737
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5166668938850616,
      "learning_rate": 9.768700852491916e-06,
      "loss": 0.2239,
      "step": 1738
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3382487003321437,
      "learning_rate": 9.768352470920706e-06,
      "loss": 0.2356,
      "step": 1739
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7296301780916024,
      "learning_rate": 9.768003833403278e-06,
      "loss": 0.2786,
      "step": 1740
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6428518041136655,
      "learning_rate": 9.767654939958347e-06,
      "loss": 0.2622,
      "step": 1741
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.807508333347318,
      "learning_rate": 9.76730579060464e-06,
      "loss": 0.2384,
      "step": 1742
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5213725388838946,
      "learning_rate": 9.766956385360897e-06,
      "loss": 0.221,
      "step": 1743
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4421635541538649,
      "learning_rate": 9.766606724245873e-06,
      "loss": 0.2172,
      "step": 1744
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7554754573381086,
      "learning_rate": 9.766256807278336e-06,
      "loss": 0.255,
      "step": 1745
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5144631931216825,
      "learning_rate": 9.765906634477072e-06,
      "loss": 0.2123,
      "step": 1746
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4189427324171482,
      "learning_rate": 9.765556205860871e-06,
      "loss": 0.2559,
      "step": 1747
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.312330677864779,
      "learning_rate": 9.765205521448546e-06,
      "loss": 0.1968,
      "step": 1748
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4840485751257777,
      "learning_rate": 9.76485458125892e-06,
      "loss": 0.2316,
      "step": 1749
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4689633521815892,
      "learning_rate": 9.764503385310832e-06,
      "loss": 0.2059,
      "step": 1750
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5109896077523253,
      "learning_rate": 9.764151933623128e-06,
      "loss": 0.2694,
      "step": 1751
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3950757427748874,
      "learning_rate": 9.763800226214676e-06,
      "loss": 0.1811,
      "step": 1752
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5540312326284627,
      "learning_rate": 9.763448263104354e-06,
      "loss": 0.2009,
      "step": 1753
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7146612018399212,
      "learning_rate": 9.763096044311052e-06,
      "loss": 0.2365,
      "step": 1754
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5064354944989968,
      "learning_rate": 9.762743569853678e-06,
      "loss": 0.2076,
      "step": 1755
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.425400050179596,
      "learning_rate": 9.76239083975115e-06,
      "loss": 0.2521,
      "step": 1756
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.558954480252376,
      "learning_rate": 9.762037854022404e-06,
      "loss": 0.2807,
      "step": 1757
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5583352573535572,
      "learning_rate": 9.761684612686381e-06,
      "loss": 0.2358,
      "step": 1758
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.014469495657771,
      "learning_rate": 9.761331115762047e-06,
      "loss": 0.6238,
      "step": 1759
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.508560217232071,
      "learning_rate": 9.760977363268374e-06,
      "loss": 0.2143,
      "step": 1760
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3482809216255482,
      "learning_rate": 9.760623355224353e-06,
      "loss": 0.1849,
      "step": 1761
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6786324406044018,
      "learning_rate": 9.760269091648979e-06,
      "loss": 0.2445,
      "step": 1762
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.830279810178743,
      "learning_rate": 9.759914572561275e-06,
      "loss": 0.7206,
      "step": 1763
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.550231933125463,
      "learning_rate": 9.759559797980265e-06,
      "loss": 0.2323,
      "step": 1764
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5302449602768893,
      "learning_rate": 9.759204767924995e-06,
      "loss": 0.181,
      "step": 1765
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6621683153472693,
      "learning_rate": 9.75884948241452e-06,
      "loss": 0.2191,
      "step": 1766
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.551382313097393,
      "learning_rate": 9.758493941467911e-06,
      "loss": 0.5484,
      "step": 1767
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4561481588967966,
      "learning_rate": 9.75813814510425e-06,
      "loss": 0.2128,
      "step": 1768
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.394196728667399,
      "learning_rate": 9.757782093342639e-06,
      "loss": 0.6855,
      "step": 1769
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.588281939706975,
      "learning_rate": 9.757425786202184e-06,
      "loss": 0.2358,
      "step": 1770
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3867498330884303,
      "learning_rate": 9.757069223702017e-06,
      "loss": 0.2448,
      "step": 1771
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7777796306460585,
      "learning_rate": 9.756712405861268e-06,
      "loss": 0.1957,
      "step": 1772
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5269974671345954,
      "learning_rate": 9.756355332699097e-06,
      "loss": 0.2596,
      "step": 1773
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.320295742760158,
      "learning_rate": 9.755998004234667e-06,
      "loss": 0.1685,
      "step": 1774
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.418358661214108,
      "learning_rate": 9.75564042048716e-06,
      "loss": 0.2043,
      "step": 1775
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.587408036171723,
      "learning_rate": 9.755282581475769e-06,
      "loss": 0.2129,
      "step": 1776
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3738416867046819,
      "learning_rate": 9.754924487219698e-06,
      "loss": 0.2174,
      "step": 1777
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4149776247821395,
      "learning_rate": 9.754566137738174e-06,
      "loss": 0.2301,
      "step": 1778
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5187928825303296,
      "learning_rate": 9.754207533050425e-06,
      "loss": 0.2258,
      "step": 1779
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3851972909800576,
      "learning_rate": 9.753848673175707e-06,
      "loss": 0.2059,
      "step": 1780
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5088714909427967,
      "learning_rate": 9.753489558133276e-06,
      "loss": 0.2103,
      "step": 1781
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4958459168108815,
      "learning_rate": 9.75313018794241e-06,
      "loss": 0.2323,
      "step": 1782
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5086159100190941,
      "learning_rate": 9.7527705626224e-06,
      "loss": 0.2025,
      "step": 1783
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3492698858552774,
      "learning_rate": 9.752410682192547e-06,
      "loss": 0.1612,
      "step": 1784
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3816693681545629,
      "learning_rate": 9.75205054667217e-06,
      "loss": 0.1852,
      "step": 1785
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6267463663412634,
      "learning_rate": 9.751690156080597e-06,
      "loss": 0.2008,
      "step": 1786
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3506938109702742,
      "learning_rate": 9.751329510437176e-06,
      "loss": 0.2189,
      "step": 1787
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5270215059072123,
      "learning_rate": 9.75096860976126e-06,
      "loss": 0.2276,
      "step": 1788
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4478679000628312,
      "learning_rate": 9.750607454072225e-06,
      "loss": 0.2643,
      "step": 1789
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3239777373607575,
      "learning_rate": 9.750246043389455e-06,
      "loss": 0.1986,
      "step": 1790
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5057977821757051,
      "learning_rate": 9.74988437773235e-06,
      "loss": 0.1868,
      "step": 1791
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.578865105189045,
      "learning_rate": 9.74952245712032e-06,
      "loss": 0.2054,
      "step": 1792
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6020558265755471,
      "learning_rate": 9.749160281572795e-06,
      "loss": 0.2263,
      "step": 1793
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4617795501102437,
      "learning_rate": 9.748797851109213e-06,
      "loss": 0.178,
      "step": 1794
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3554280430765748,
      "learning_rate": 9.748435165749029e-06,
      "loss": 0.2187,
      "step": 1795
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4977533848883124,
      "learning_rate": 9.748072225511708e-06,
      "loss": 0.2151,
      "step": 1796
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4777549976267563,
      "learning_rate": 9.747709030416734e-06,
      "loss": 0.2201,
      "step": 1797
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5003483822933346,
      "learning_rate": 9.747345580483602e-06,
      "loss": 0.2658,
      "step": 1798
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5124434011049583,
      "learning_rate": 9.746981875731817e-06,
      "loss": 0.2185,
      "step": 1799
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4908411378947168,
      "learning_rate": 9.746617916180906e-06,
      "loss": 0.2204,
      "step": 1800
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5142043770873934,
      "learning_rate": 9.7462537018504e-06,
      "loss": 0.2221,
      "step": 1801
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.554203407901688,
      "learning_rate": 9.745889232759854e-06,
      "loss": 0.25,
      "step": 1802
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4186244329769964,
      "learning_rate": 9.745524508928827e-06,
      "loss": 0.2357,
      "step": 1803
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.470528685888625,
      "learning_rate": 9.745159530376897e-06,
      "loss": 0.2284,
      "step": 1804
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.538893527051276,
      "learning_rate": 9.744794297123656e-06,
      "loss": 0.1968,
      "step": 1805
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3570080127073203,
      "learning_rate": 9.744428809188707e-06,
      "loss": 0.1709,
      "step": 1806
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.716985202567522,
      "learning_rate": 9.74406306659167e-06,
      "loss": 0.227,
      "step": 1807
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5930294146323556,
      "learning_rate": 9.74369706935217e-06,
      "loss": 0.2538,
      "step": 1808
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.477987442929875,
      "learning_rate": 9.74333081748986e-06,
      "loss": 0.2277,
      "step": 1809
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4883461524651147,
      "learning_rate": 9.742964311024396e-06,
      "loss": 0.2335,
      "step": 1810
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3784403434462695,
      "learning_rate": 9.74259754997545e-06,
      "loss": 0.1979,
      "step": 1811
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.43399893158558,
      "learning_rate": 9.742230534362708e-06,
      "loss": 0.2105,
      "step": 1812
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5747044306847444,
      "learning_rate": 9.741863264205873e-06,
      "loss": 0.2587,
      "step": 1813
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6369310886332042,
      "learning_rate": 9.741495739524655e-06,
      "loss": 0.2665,
      "step": 1814
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4148010433064986,
      "learning_rate": 9.741127960338784e-06,
      "loss": 0.2588,
      "step": 1815
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.383135311959032,
      "learning_rate": 9.740759926667998e-06,
      "loss": 0.7326,
      "step": 1816
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.215234690982681,
      "learning_rate": 9.740391638532055e-06,
      "loss": 0.7683,
      "step": 1817
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.521689426252949,
      "learning_rate": 9.74002309595072e-06,
      "loss": 0.6337,
      "step": 1818
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.219021824095078,
      "learning_rate": 9.739654298943775e-06,
      "loss": 0.6336,
      "step": 1819
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4671528860405594,
      "learning_rate": 9.739285247531019e-06,
      "loss": 0.2943,
      "step": 1820
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.950679909288861,
      "learning_rate": 9.738915941732258e-06,
      "loss": 0.5464,
      "step": 1821
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.263604446988797,
      "learning_rate": 9.738546381567315e-06,
      "loss": 0.1694,
      "step": 1822
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.735655068451781,
      "learning_rate": 9.73817656705603e-06,
      "loss": 0.2754,
      "step": 1823
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3394602050267703,
      "learning_rate": 9.737806498218249e-06,
      "loss": 0.1876,
      "step": 1824
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3135586864657471,
      "learning_rate": 9.737436175073837e-06,
      "loss": 0.2064,
      "step": 1825
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3710058616784464,
      "learning_rate": 9.737065597642674e-06,
      "loss": 0.2041,
      "step": 1826
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3735647173601169,
      "learning_rate": 9.736694765944648e-06,
      "loss": 0.213,
      "step": 1827
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.575970707511317,
      "learning_rate": 9.736323679999666e-06,
      "loss": 0.1764,
      "step": 1828
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4874345342814497,
      "learning_rate": 9.735952339827643e-06,
      "loss": 0.2594,
      "step": 1829
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4487034296171157,
      "learning_rate": 9.735580745448514e-06,
      "loss": 0.2587,
      "step": 1830
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6928386745538273,
      "learning_rate": 9.735208896882225e-06,
      "loss": 0.2471,
      "step": 1831
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6533601239995719,
      "learning_rate": 9.734836794148734e-06,
      "loss": 0.2599,
      "step": 1832
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.793279344838839,
      "learning_rate": 9.734464437268014e-06,
      "loss": 0.7486,
      "step": 1833
    },
    {
      "epoch": 0.13,
      "grad_norm": 23.526035949817015,
      "learning_rate": 9.734091826260053e-06,
      "loss": 0.6702,
      "step": 1834
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5635677219351236,
      "learning_rate": 9.733718961144849e-06,
      "loss": 0.2625,
      "step": 1835
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5729201459566307,
      "learning_rate": 9.733345841942418e-06,
      "loss": 0.2621,
      "step": 1836
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4512507834508095,
      "learning_rate": 9.732972468672788e-06,
      "loss": 0.2568,
      "step": 1837
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.81825127612839,
      "learning_rate": 9.732598841355997e-06,
      "loss": 0.7093,
      "step": 1838
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.935854508103519,
      "learning_rate": 9.732224960012103e-06,
      "loss": 0.448,
      "step": 1839
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2361647924980592,
      "learning_rate": 9.731850824661171e-06,
      "loss": 0.1815,
      "step": 1840
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.510432719016519,
      "learning_rate": 9.731476435323289e-06,
      "loss": 0.2244,
      "step": 1841
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5664921794746007,
      "learning_rate": 9.731101792018547e-06,
      "loss": 0.2267,
      "step": 1842
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4119985931063965,
      "learning_rate": 9.730726894767056e-06,
      "loss": 0.2467,
      "step": 1843
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4895625353438624,
      "learning_rate": 9.73035174358894e-06,
      "loss": 0.2377,
      "step": 1844
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.415260683378231,
      "learning_rate": 9.729976338504336e-06,
      "loss": 0.255,
      "step": 1845
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6206785462449451,
      "learning_rate": 9.729600679533392e-06,
      "loss": 0.236,
      "step": 1846
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4862274100663593,
      "learning_rate": 9.729224766696274e-06,
      "loss": 0.2049,
      "step": 1847
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3233284851217337,
      "learning_rate": 9.72884860001316e-06,
      "loss": 0.2223,
      "step": 1848
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.507432629772718,
      "learning_rate": 9.728472179504239e-06,
      "loss": 0.1975,
      "step": 1849
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.664902547746061,
      "learning_rate": 9.728095505189714e-06,
      "loss": 0.246,
      "step": 1850
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4211299756869158,
      "learning_rate": 9.72771857708981e-06,
      "loss": 0.2365,
      "step": 1851
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.154853910062455,
      "learning_rate": 9.727341395224751e-06,
      "loss": 0.7025,
      "step": 1852
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.537067176261166,
      "learning_rate": 9.72696395961479e-06,
      "loss": 0.2221,
      "step": 1853
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3677475424495202,
      "learning_rate": 9.726586270280181e-06,
      "loss": 0.2353,
      "step": 1854
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7353022976423134,
      "learning_rate": 9.7262083272412e-06,
      "loss": 0.3053,
      "step": 1855
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3710181128537797,
      "learning_rate": 9.72583013051813e-06,
      "loss": 0.1966,
      "step": 1856
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9894691453887685,
      "learning_rate": 9.725451680131278e-06,
      "loss": 0.2328,
      "step": 1857
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.431340194383833,
      "learning_rate": 9.725072976100949e-06,
      "loss": 0.2569,
      "step": 1858
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4015519374624992,
      "learning_rate": 9.724694018447475e-06,
      "loss": 0.2183,
      "step": 1859
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4059921260587471,
      "learning_rate": 9.724314807191197e-06,
      "loss": 0.1968,
      "step": 1860
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.464275152148016,
      "learning_rate": 9.723935342352468e-06,
      "loss": 0.1826,
      "step": 1861
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.449170754043741,
      "learning_rate": 9.723555623951657e-06,
      "loss": 0.2015,
      "step": 1862
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4958189850488863,
      "learning_rate": 9.723175652009148e-06,
      "loss": 0.2439,
      "step": 1863
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5739353607165942,
      "learning_rate": 9.72279542654533e-06,
      "loss": 0.2216,
      "step": 1864
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.545987404161147,
      "learning_rate": 9.72241494758062e-06,
      "loss": 0.2347,
      "step": 1865
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.67002781279628,
      "learning_rate": 9.722034215135436e-06,
      "loss": 0.7898,
      "step": 1866
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.490860802093944,
      "learning_rate": 9.721653229230214e-06,
      "loss": 0.2081,
      "step": 1867
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4295548205115338,
      "learning_rate": 9.721271989885405e-06,
      "loss": 0.1908,
      "step": 1868
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.405048012704226,
      "learning_rate": 9.720890497121473e-06,
      "loss": 0.2098,
      "step": 1869
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2924687409460553,
      "learning_rate": 9.720508750958892e-06,
      "loss": 0.218,
      "step": 1870
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4178717374792467,
      "learning_rate": 9.720126751418157e-06,
      "loss": 0.244,
      "step": 1871
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.519848553701147,
      "learning_rate": 9.71974449851977e-06,
      "loss": 0.2592,
      "step": 1872
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4704563787455887,
      "learning_rate": 9.719361992284248e-06,
      "loss": 0.2538,
      "step": 1873
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6823052719706368,
      "learning_rate": 9.718979232732125e-06,
      "loss": 0.2319,
      "step": 1874
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5352781424097124,
      "learning_rate": 9.718596219883942e-06,
      "loss": 0.2258,
      "step": 1875
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4898589395872692,
      "learning_rate": 9.718212953760261e-06,
      "loss": 0.2002,
      "step": 1876
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7217510529734357,
      "learning_rate": 9.717829434381653e-06,
      "loss": 0.2771,
      "step": 1877
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5160913837771142,
      "learning_rate": 9.717445661768703e-06,
      "loss": 0.2289,
      "step": 1878
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3547828002842095,
      "learning_rate": 9.717061635942014e-06,
      "loss": 0.189,
      "step": 1879
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.254461770583257,
      "learning_rate": 9.716677356922193e-06,
      "loss": 0.6469,
      "step": 1880
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5028762322461113,
      "learning_rate": 9.716292824729872e-06,
      "loss": 0.2349,
      "step": 1881
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6669656601377145,
      "learning_rate": 9.715908039385688e-06,
      "loss": 0.2003,
      "step": 1882
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3608523474912946,
      "learning_rate": 9.715523000910296e-06,
      "loss": 0.2579,
      "step": 1883
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4953629827375976,
      "learning_rate": 9.715137709324363e-06,
      "loss": 0.2258,
      "step": 1884
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.590519300516705,
      "learning_rate": 9.714752164648571e-06,
      "loss": 0.302,
      "step": 1885
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4235103783129495,
      "learning_rate": 9.714366366903612e-06,
      "loss": 0.2176,
      "step": 1886
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4928235981142612,
      "learning_rate": 9.713980316110196e-06,
      "loss": 0.213,
      "step": 1887
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7248317156192252,
      "learning_rate": 9.713594012289045e-06,
      "loss": 0.246,
      "step": 1888
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4034855318944837,
      "learning_rate": 9.713207455460893e-06,
      "loss": 0.2058,
      "step": 1889
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4278403851695654,
      "learning_rate": 9.71282064564649e-06,
      "loss": 0.2199,
      "step": 1890
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.469329707855562,
      "learning_rate": 9.712433582866596e-06,
      "loss": 0.2497,
      "step": 1891
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3010679363763809,
      "learning_rate": 9.71204626714199e-06,
      "loss": 0.2031,
      "step": 1892
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4156084758475338,
      "learning_rate": 9.71165869849346e-06,
      "loss": 0.2283,
      "step": 1893
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7838289407957961,
      "learning_rate": 9.71127087694181e-06,
      "loss": 0.2436,
      "step": 1894
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5887673362043404,
      "learning_rate": 9.710882802507857e-06,
      "loss": 0.261,
      "step": 1895
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.409210802499972,
      "learning_rate": 9.710494475212429e-06,
      "loss": 0.193,
      "step": 1896
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4623938364806417,
      "learning_rate": 9.710105895076372e-06,
      "loss": 0.1991,
      "step": 1897
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7049493576296055,
      "learning_rate": 9.709717062120542e-06,
      "loss": 0.2438,
      "step": 1898
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4520075655164966,
      "learning_rate": 9.709327976365813e-06,
      "loss": 0.2494,
      "step": 1899
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.165840308429936,
      "learning_rate": 9.708938637833065e-06,
      "loss": 0.5862,
      "step": 1900
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.924872560397852,
      "learning_rate": 9.708549046543201e-06,
      "loss": 0.7531,
      "step": 1901
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.411684373014523,
      "learning_rate": 9.708159202517129e-06,
      "loss": 0.2362,
      "step": 1902
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6666750908200851,
      "learning_rate": 9.707769105775774e-06,
      "loss": 0.2554,
      "step": 1903
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4960331275406946,
      "learning_rate": 9.707378756340079e-06,
      "loss": 0.2552,
      "step": 1904
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4987551615369414,
      "learning_rate": 9.706988154230992e-06,
      "loss": 0.2537,
      "step": 1905
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3951431210224616,
      "learning_rate": 9.70659729946948e-06,
      "loss": 0.2613,
      "step": 1906
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5313122108829853,
      "learning_rate": 9.706206192076527e-06,
      "loss": 0.2024,
      "step": 1907
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4271304891750085,
      "learning_rate": 9.705814832073118e-06,
      "loss": 0.2184,
      "step": 1908
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3742564968571074,
      "learning_rate": 9.705423219480265e-06,
      "loss": 0.2041,
      "step": 1909
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.369953197872659,
      "learning_rate": 9.70503135431899e-06,
      "loss": 0.2157,
      "step": 1910
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4623807136452425,
      "learning_rate": 9.704639236610321e-06,
      "loss": 0.2142,
      "step": 1911
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3650250852724641,
      "learning_rate": 9.704246866375309e-06,
      "loss": 0.2139,
      "step": 1912
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3946951198554776,
      "learning_rate": 9.703854243635015e-06,
      "loss": 0.2611,
      "step": 1913
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3217007436002144,
      "learning_rate": 9.703461368410512e-06,
      "loss": 0.1748,
      "step": 1914
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.420273891053106,
      "learning_rate": 9.70306824072289e-06,
      "loss": 0.2193,
      "step": 1915
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.465617760705863,
      "learning_rate": 9.702674860593248e-06,
      "loss": 0.2026,
      "step": 1916
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4119543226623383,
      "learning_rate": 9.702281228042703e-06,
      "loss": 0.2331,
      "step": 1917
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6226099691583857,
      "learning_rate": 9.701887343092381e-06,
      "loss": 0.2662,
      "step": 1918
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.351811391037114,
      "learning_rate": 9.701493205763427e-06,
      "loss": 0.7424,
      "step": 1919
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.543343158235312,
      "learning_rate": 9.701098816076995e-06,
      "loss": 0.2553,
      "step": 1920
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6108706193671603,
      "learning_rate": 9.700704174054257e-06,
      "loss": 0.1923,
      "step": 1921
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6448967344527,
      "learning_rate": 9.700309279716394e-06,
      "loss": 0.2349,
      "step": 1922
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.95404167783207,
      "learning_rate": 9.6999141330846e-06,
      "loss": 0.4977,
      "step": 1923
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7987358515060412,
      "learning_rate": 9.699518734180088e-06,
      "loss": 0.2859,
      "step": 1924
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4880639584777857,
      "learning_rate": 9.69912308302408e-06,
      "loss": 0.2307,
      "step": 1925
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5366616380761902,
      "learning_rate": 9.698727179637815e-06,
      "loss": 0.2411,
      "step": 1926
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.599479460646312,
      "learning_rate": 9.69833102404254e-06,
      "loss": 0.2604,
      "step": 1927
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3877101222843358,
      "learning_rate": 9.697934616259523e-06,
      "loss": 0.239,
      "step": 1928
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4709492509117583,
      "learning_rate": 9.69753795631004e-06,
      "loss": 0.2611,
      "step": 1929
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.029819809309242,
      "learning_rate": 9.697141044215379e-06,
      "loss": 0.2449,
      "step": 1930
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6259809200983182,
      "learning_rate": 9.696743879996851e-06,
      "loss": 0.2308,
      "step": 1931
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3304292583048278,
      "learning_rate": 9.696346463675767e-06,
      "loss": 0.2612,
      "step": 1932
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4982271274022423,
      "learning_rate": 9.695948795273465e-06,
      "loss": 0.2718,
      "step": 1933
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.811316470748526,
      "learning_rate": 9.695550874811288e-06,
      "loss": 0.65,
      "step": 1934
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3713588643410768,
      "learning_rate": 9.695152702310592e-06,
      "loss": 0.2332,
      "step": 1935
    },
    {
      "epoch": 0.14,
      "grad_norm": 15.045125078879652,
      "learning_rate": 9.694754277792754e-06,
      "loss": 0.5881,
      "step": 1936
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5304467441851728,
      "learning_rate": 9.694355601279157e-06,
      "loss": 0.2416,
      "step": 1937
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3919126923257714,
      "learning_rate": 9.693956672791201e-06,
      "loss": 0.2425,
      "step": 1938
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4548562351366578,
      "learning_rate": 9.6935574923503e-06,
      "loss": 0.2542,
      "step": 1939
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.070726763963831,
      "learning_rate": 9.693158059977879e-06,
      "loss": 0.5373,
      "step": 1940
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4969554025609875,
      "learning_rate": 9.692758375695377e-06,
      "loss": 0.2646,
      "step": 1941
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4569167815753479,
      "learning_rate": 9.692358439524248e-06,
      "loss": 0.2615,
      "step": 1942
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3328553208445544,
      "learning_rate": 9.691958251485962e-06,
      "loss": 0.2341,
      "step": 1943
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5713704154765922,
      "learning_rate": 9.691557811601996e-06,
      "loss": 0.2424,
      "step": 1944
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.39665779910806,
      "learning_rate": 9.691157119893846e-06,
      "loss": 0.2273,
      "step": 1945
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.529280500863797,
      "learning_rate": 9.690756176383019e-06,
      "loss": 0.2533,
      "step": 1946
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8566822154673406,
      "learning_rate": 9.690354981091033e-06,
      "loss": 0.1869,
      "step": 1947
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4771582292509915,
      "learning_rate": 9.689953534039428e-06,
      "loss": 0.2655,
      "step": 1948
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.368097707546508,
      "learning_rate": 9.689551835249749e-06,
      "loss": 0.6712,
      "step": 1949
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4129571833399497,
      "learning_rate": 9.689149884743558e-06,
      "loss": 0.233,
      "step": 1950
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4836749349537037,
      "learning_rate": 9.68874768254243e-06,
      "loss": 0.2901,
      "step": 1951
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3570903349848955,
      "learning_rate": 9.688345228667955e-06,
      "loss": 0.2312,
      "step": 1952
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3881651633940988,
      "learning_rate": 9.687942523141732e-06,
      "loss": 0.2316,
      "step": 1953
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5997011592852977,
      "learning_rate": 9.687539565985379e-06,
      "loss": 0.228,
      "step": 1954
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5082993978627361,
      "learning_rate": 9.687136357220525e-06,
      "loss": 0.2123,
      "step": 1955
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.698329092523647,
      "learning_rate": 9.686732896868814e-06,
      "loss": 0.6586,
      "step": 1956
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.405525969744624,
      "learning_rate": 9.686329184951899e-06,
      "loss": 0.2509,
      "step": 1957
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5297734761278114,
      "learning_rate": 9.685925221491452e-06,
      "loss": 0.2187,
      "step": 1958
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3260628762739894,
      "learning_rate": 9.685521006509153e-06,
      "loss": 0.1754,
      "step": 1959
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3450722509545328,
      "learning_rate": 9.685116540026703e-06,
      "loss": 0.2111,
      "step": 1960
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.332705349393018,
      "learning_rate": 9.68471182206581e-06,
      "loss": 0.1755,
      "step": 1961
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4101069743023689,
      "learning_rate": 9.684306852648198e-06,
      "loss": 0.1847,
      "step": 1962
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.416647702595143,
      "learning_rate": 9.683901631795602e-06,
      "loss": 0.233,
      "step": 1963
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.366678062343066,
      "learning_rate": 9.683496159529777e-06,
      "loss": 0.2035,
      "step": 1964
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6153968253300508,
      "learning_rate": 9.683090435872484e-06,
      "loss": 0.2598,
      "step": 1965
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3860988775998824,
      "learning_rate": 9.6826844608455e-06,
      "loss": 0.2118,
      "step": 1966
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4716406762575418,
      "learning_rate": 9.68227823447062e-06,
      "loss": 0.1927,
      "step": 1967
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4200399721936707,
      "learning_rate": 9.681871756769642e-06,
      "loss": 0.208,
      "step": 1968
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7434795267719347,
      "learning_rate": 9.681465027764391e-06,
      "loss": 0.2438,
      "step": 1969
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2958351514517805,
      "learning_rate": 9.681058047476696e-06,
      "loss": 0.186,
      "step": 1970
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3496401609989983,
      "learning_rate": 9.6806508159284e-06,
      "loss": 0.2105,
      "step": 1971
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4094482272930087,
      "learning_rate": 9.680243333141365e-06,
      "loss": 0.1977,
      "step": 1972
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4193954446257815,
      "learning_rate": 9.679835599137462e-06,
      "loss": 0.2458,
      "step": 1973
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5013428454716475,
      "learning_rate": 9.679427613938575e-06,
      "loss": 0.2398,
      "step": 1974
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5175489094371841,
      "learning_rate": 9.679019377566605e-06,
      "loss": 0.2572,
      "step": 1975
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.545944330114262,
      "learning_rate": 9.678610890043462e-06,
      "loss": 0.2081,
      "step": 1976
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5315826750207684,
      "learning_rate": 9.678202151391075e-06,
      "loss": 0.218,
      "step": 1977
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3493208140573416,
      "learning_rate": 9.677793161631381e-06,
      "loss": 0.2126,
      "step": 1978
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3913479211246853,
      "learning_rate": 9.677383920786335e-06,
      "loss": 0.245,
      "step": 1979
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3725360521474659,
      "learning_rate": 9.6769744288779e-06,
      "loss": 0.2038,
      "step": 1980
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1460072122737435,
      "learning_rate": 9.676564685928063e-06,
      "loss": 0.1857,
      "step": 1981
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4165996092298085,
      "learning_rate": 9.676154691958807e-06,
      "loss": 0.2542,
      "step": 1982
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3913252584818068,
      "learning_rate": 9.67574444699215e-06,
      "loss": 0.2647,
      "step": 1983
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3721701264030706,
      "learning_rate": 9.675333951050103e-06,
      "loss": 0.2327,
      "step": 1984
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2781860681468193,
      "learning_rate": 9.674923204154704e-06,
      "loss": 0.2235,
      "step": 1985
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4157902387139727,
      "learning_rate": 9.674512206328002e-06,
      "loss": 0.2058,
      "step": 1986
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5593694121486974,
      "learning_rate": 9.674100957592054e-06,
      "loss": 0.2096,
      "step": 1987
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.255870521935859,
      "learning_rate": 9.673689457968936e-06,
      "loss": 0.2097,
      "step": 1988
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.506996381722429,
      "learning_rate": 9.673277707480735e-06,
      "loss": 0.7056,
      "step": 1989
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.450737558060842,
      "learning_rate": 9.672865706149553e-06,
      "loss": 0.588,
      "step": 1990
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3361370022074133,
      "learning_rate": 9.672453453997506e-06,
      "loss": 0.2201,
      "step": 1991
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.25532987521443,
      "learning_rate": 9.672040951046717e-06,
      "loss": 0.2024,
      "step": 1992
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4253200274503304,
      "learning_rate": 9.671628197319333e-06,
      "loss": 0.2343,
      "step": 1993
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5189963947102203,
      "learning_rate": 9.671215192837507e-06,
      "loss": 0.2203,
      "step": 1994
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6863477956428774,
      "learning_rate": 9.670801937623406e-06,
      "loss": 0.2358,
      "step": 1995
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6022919721304347,
      "learning_rate": 9.670388431699214e-06,
      "loss": 0.2825,
      "step": 1996
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3422471162623222,
      "learning_rate": 9.669974675087124e-06,
      "loss": 0.2141,
      "step": 1997
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3506876497482363,
      "learning_rate": 9.669560667809348e-06,
      "loss": 0.2183,
      "step": 1998
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.841728687217585,
      "learning_rate": 9.669146409888104e-06,
      "loss": 0.5604,
      "step": 1999
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3005855706238183,
      "learning_rate": 9.668731901345632e-06,
      "loss": 0.1854,
      "step": 2000
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4462792549244696,
      "learning_rate": 9.66831714220418e-06,
      "loss": 0.2029,
      "step": 2001
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.385734966334063,
      "learning_rate": 9.667902132486009e-06,
      "loss": 0.1894,
      "step": 2002
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4070308823112274,
      "learning_rate": 9.667486872213396e-06,
      "loss": 0.2589,
      "step": 2003
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.554737030225706,
      "learning_rate": 9.667071361408633e-06,
      "loss": 0.2924,
      "step": 2004
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4899583573738209,
      "learning_rate": 9.666655600094019e-06,
      "loss": 0.2363,
      "step": 2005
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.390947278730927,
      "learning_rate": 9.66623958829187e-06,
      "loss": 0.2298,
      "step": 2006
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4999926233597811,
      "learning_rate": 9.66582332602452e-06,
      "loss": 0.2058,
      "step": 2007
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3877316040988428,
      "learning_rate": 9.665406813314311e-06,
      "loss": 0.254,
      "step": 2008
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4413783819472263,
      "learning_rate": 9.664990050183598e-06,
      "loss": 0.2224,
      "step": 2009
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.434013906942006,
      "learning_rate": 9.664573036654751e-06,
      "loss": 0.2018,
      "step": 2010
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4753947352626866,
      "learning_rate": 9.664155772750157e-06,
      "loss": 0.2852,
      "step": 2011
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.176507368366487,
      "learning_rate": 9.663738258492208e-06,
      "loss": 0.6371,
      "step": 2012
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5175244668684245,
      "learning_rate": 9.66332049390332e-06,
      "loss": 0.2448,
      "step": 2013
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2154787302290333,
      "learning_rate": 9.662902479005913e-06,
      "loss": 0.167,
      "step": 2014
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5608394424198524,
      "learning_rate": 9.662484213822425e-06,
      "loss": 0.2736,
      "step": 2015
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3993343109913665,
      "learning_rate": 9.662065698375309e-06,
      "loss": 0.2153,
      "step": 2016
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4328077159315211,
      "learning_rate": 9.661646932687028e-06,
      "loss": 0.1873,
      "step": 2017
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4972597308547662,
      "learning_rate": 9.661227916780058e-06,
      "loss": 0.2281,
      "step": 2018
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3933492041177549,
      "learning_rate": 9.660808650676891e-06,
      "loss": 0.2377,
      "step": 2019
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.948451495998756,
      "learning_rate": 9.660389134400034e-06,
      "loss": 0.2419,
      "step": 2020
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.076089257238804,
      "learning_rate": 9.659969367972003e-06,
      "loss": 0.696,
      "step": 2021
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3860670402533746,
      "learning_rate": 9.659549351415327e-06,
      "loss": 0.2409,
      "step": 2022
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3658511006484433,
      "learning_rate": 9.659129084752554e-06,
      "loss": 0.2203,
      "step": 2023
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9841676260377752,
      "learning_rate": 9.658708568006244e-06,
      "loss": 0.2502,
      "step": 2024
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.402650641094453,
      "learning_rate": 9.658287801198963e-06,
      "loss": 0.2201,
      "step": 2025
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5170963103634403,
      "learning_rate": 9.6578667843533e-06,
      "loss": 0.2155,
      "step": 2026
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6663798867931658,
      "learning_rate": 9.657445517491852e-06,
      "loss": 0.275,
      "step": 2027
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.524758584013468,
      "learning_rate": 9.657024000637235e-06,
      "loss": 0.2107,
      "step": 2028
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4588984892968944,
      "learning_rate": 9.656602233812068e-06,
      "loss": 0.2032,
      "step": 2029
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.229226181325648,
      "learning_rate": 9.656180217038994e-06,
      "loss": 0.1808,
      "step": 2030
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5829505974249733,
      "learning_rate": 9.655757950340662e-06,
      "loss": 0.3236,
      "step": 2031
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.961639738709765,
      "learning_rate": 9.655335433739742e-06,
      "loss": 0.6644,
      "step": 2032
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5655971095139682,
      "learning_rate": 9.65491266725891e-06,
      "loss": 0.232,
      "step": 2033
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4092100892962622,
      "learning_rate": 9.65448965092086e-06,
      "loss": 0.2122,
      "step": 2034
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6776571274439038,
      "learning_rate": 9.654066384748296e-06,
      "loss": 0.2329,
      "step": 2035
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.365196581000431,
      "learning_rate": 9.653642868763937e-06,
      "loss": 0.6796,
      "step": 2036
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6792110434028609,
      "learning_rate": 9.653219102990517e-06,
      "loss": 0.2155,
      "step": 2037
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.479195429892465,
      "learning_rate": 9.652795087450781e-06,
      "loss": 0.1914,
      "step": 2038
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4301722559567636,
      "learning_rate": 9.652370822167491e-06,
      "loss": 0.2001,
      "step": 2039
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3291237840731893,
      "learning_rate": 9.651946307163417e-06,
      "loss": 0.2065,
      "step": 2040
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6486920412271882,
      "learning_rate": 9.651521542461348e-06,
      "loss": 0.2274,
      "step": 2041
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.71291968455386,
      "learning_rate": 9.651096528084082e-06,
      "loss": 0.6185,
      "step": 2042
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.613592932321471,
      "learning_rate": 9.650671264054432e-06,
      "loss": 0.1721,
      "step": 2043
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7377470959957462,
      "learning_rate": 9.650245750395224e-06,
      "loss": 0.2261,
      "step": 2044
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.917527601020814,
      "learning_rate": 9.649819987129299e-06,
      "loss": 0.6601,
      "step": 2045
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6884395572607929,
      "learning_rate": 9.649393974279509e-06,
      "loss": 0.2185,
      "step": 2046
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.09780037709546,
      "learning_rate": 9.648967711868722e-06,
      "loss": 0.5147,
      "step": 2047
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.364056002031816,
      "learning_rate": 9.648541199919818e-06,
      "loss": 0.2168,
      "step": 2048
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4875827416470355,
      "learning_rate": 9.648114438455689e-06,
      "loss": 0.2507,
      "step": 2049
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3779775470439601,
      "learning_rate": 9.647687427499243e-06,
      "loss": 0.2555,
      "step": 2050
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.59606789347987,
      "learning_rate": 9.647260167073402e-06,
      "loss": 0.6518,
      "step": 2051
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5265644765208641,
      "learning_rate": 9.646832657201097e-06,
      "loss": 0.2329,
      "step": 2052
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.556613570163405,
      "learning_rate": 9.646404897905276e-06,
      "loss": 0.2526,
      "step": 2053
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4951310932493223,
      "learning_rate": 9.645976889208896e-06,
      "loss": 0.2319,
      "step": 2054
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.514169998107707,
      "learning_rate": 9.645548631134938e-06,
      "loss": 0.223,
      "step": 2055
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.378561275734485,
      "learning_rate": 9.645120123706384e-06,
      "loss": 0.1908,
      "step": 2056
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.529367858055682,
      "learning_rate": 9.644691366946238e-06,
      "loss": 0.2378,
      "step": 2057
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.749999058452197,
      "learning_rate": 9.644262360877507e-06,
      "loss": 0.4636,
      "step": 2058
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6691840195675718,
      "learning_rate": 9.643833105523227e-06,
      "loss": 0.234,
      "step": 2059
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9044867492531274,
      "learning_rate": 9.643403600906433e-06,
      "loss": 0.2205,
      "step": 2060
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3131553159943004,
      "learning_rate": 9.642973847050183e-06,
      "loss": 0.1943,
      "step": 2061
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5074633831157127,
      "learning_rate": 9.64254384397754e-06,
      "loss": 0.231,
      "step": 2062
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3843823957738763,
      "learning_rate": 9.64211359171159e-06,
      "loss": 0.2478,
      "step": 2063
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3891638643362163,
      "learning_rate": 9.641683090275423e-06,
      "loss": 0.2428,
      "step": 2064
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.559516381594344,
      "learning_rate": 9.641252339692146e-06,
      "loss": 0.2612,
      "step": 2065
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.515013793356278,
      "learning_rate": 9.640821339984885e-06,
      "loss": 0.1976,
      "step": 2066
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.866356190073929,
      "learning_rate": 9.640390091176772e-06,
      "loss": 0.46,
      "step": 2067
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5610310042587374,
      "learning_rate": 9.639958593290952e-06,
      "loss": 0.2414,
      "step": 2068
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4735061792704163,
      "learning_rate": 9.63952684635059e-06,
      "loss": 0.2261,
      "step": 2069
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3647505077613498,
      "learning_rate": 9.639094850378859e-06,
      "loss": 0.2069,
      "step": 2070
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4755872866428412,
      "learning_rate": 9.638662605398944e-06,
      "loss": 0.2249,
      "step": 2071
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3241105884689077,
      "learning_rate": 9.638230111434051e-06,
      "loss": 0.1963,
      "step": 2072
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3181839956390904,
      "learning_rate": 9.63779736850739e-06,
      "loss": 0.2152,
      "step": 2073
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4283543373259668,
      "learning_rate": 9.637364376642194e-06,
      "loss": 0.2011,
      "step": 2074
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3601027013785114,
      "learning_rate": 9.6369311358617e-06,
      "loss": 0.2161,
      "step": 2075
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.503669911642347,
      "learning_rate": 9.636497646189165e-06,
      "loss": 0.2513,
      "step": 2076
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.570466148709927,
      "learning_rate": 9.636063907647856e-06,
      "loss": 0.2315,
      "step": 2077
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3434407784692561,
      "learning_rate": 9.635629920261054e-06,
      "loss": 0.1897,
      "step": 2078
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.229893335694586,
      "learning_rate": 9.635195684052055e-06,
      "loss": 0.2284,
      "step": 2079
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5305370019155835,
      "learning_rate": 9.634761199044165e-06,
      "loss": 0.2025,
      "step": 2080
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3285144875002792,
      "learning_rate": 9.634326465260708e-06,
      "loss": 0.1952,
      "step": 2081
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2965342054696503,
      "learning_rate": 9.633891482725017e-06,
      "loss": 0.1583,
      "step": 2082
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.305288213156061,
      "learning_rate": 9.633456251460443e-06,
      "loss": 0.7332,
      "step": 2083
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6089621017595082,
      "learning_rate": 9.633020771490343e-06,
      "loss": 0.2435,
      "step": 2084
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.691289866192461,
      "learning_rate": 9.632585042838092e-06,
      "loss": 0.2419,
      "step": 2085
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6023878068408524,
      "learning_rate": 9.632149065527083e-06,
      "loss": 0.2501,
      "step": 2086
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3745010211116593,
      "learning_rate": 9.631712839580713e-06,
      "loss": 0.2383,
      "step": 2087
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3786220071975555,
      "learning_rate": 9.631276365022402e-06,
      "loss": 0.2281,
      "step": 2088
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3901687473012596,
      "learning_rate": 9.630839641875571e-06,
      "loss": 0.2288,
      "step": 2089
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.42254037784379,
      "learning_rate": 9.630402670163668e-06,
      "loss": 0.6686,
      "step": 2090
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2932780366574195,
      "learning_rate": 9.629965449910144e-06,
      "loss": 0.1824,
      "step": 2091
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4673187630162479,
      "learning_rate": 9.629527981138468e-06,
      "loss": 0.2451,
      "step": 2092
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8261084867685116,
      "learning_rate": 9.629090263872124e-06,
      "loss": 0.241,
      "step": 2093
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3730726716899075,
      "learning_rate": 9.628652298134601e-06,
      "loss": 0.2222,
      "step": 2094
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.145929770886053,
      "learning_rate": 9.628214083949415e-06,
      "loss": 0.5604,
      "step": 2095
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4894372629415926,
      "learning_rate": 9.627775621340083e-06,
      "loss": 0.2757,
      "step": 2096
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.373150962284338,
      "learning_rate": 9.62733691033014e-06,
      "loss": 0.2243,
      "step": 2097
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3972287139307273,
      "learning_rate": 9.626897950943137e-06,
      "loss": 0.2046,
      "step": 2098
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6202438846569491,
      "learning_rate": 9.626458743202632e-06,
      "loss": 0.2971,
      "step": 2099
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4444902930510832,
      "learning_rate": 9.626019287132202e-06,
      "loss": 0.225,
      "step": 2100
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3667574563987765,
      "learning_rate": 9.625579582755436e-06,
      "loss": 0.2324,
      "step": 2101
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7630522812761482,
      "learning_rate": 9.625139630095933e-06,
      "loss": 0.2863,
      "step": 2102
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.046378017571318,
      "learning_rate": 9.624699429177308e-06,
      "loss": 0.7104,
      "step": 2103
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5391312427463832,
      "learning_rate": 9.624258980023194e-06,
      "loss": 0.257,
      "step": 2104
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4103236786796092,
      "learning_rate": 9.623818282657227e-06,
      "loss": 0.2571,
      "step": 2105
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.325287397409188,
      "learning_rate": 9.623377337103066e-06,
      "loss": 0.2191,
      "step": 2106
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5456451763862133,
      "learning_rate": 9.622936143384374e-06,
      "loss": 0.2527,
      "step": 2107
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.421445026338252,
      "learning_rate": 9.622494701524838e-06,
      "loss": 0.262,
      "step": 2108
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3826173054696445,
      "learning_rate": 9.622053011548151e-06,
      "loss": 0.2099,
      "step": 2109
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3288567571280328,
      "learning_rate": 9.62161107347802e-06,
      "loss": 0.2145,
      "step": 2110
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3770629542494632,
      "learning_rate": 9.621168887338168e-06,
      "loss": 0.2492,
      "step": 2111
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.929837125462641,
      "learning_rate": 9.620726453152327e-06,
      "loss": 0.6346,
      "step": 2112
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3735618465396864,
      "learning_rate": 9.62028377094425e-06,
      "loss": 0.2055,
      "step": 2113
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9083863886016605,
      "learning_rate": 9.619840840737693e-06,
      "loss": 0.2497,
      "step": 2114
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4099356283735647,
      "learning_rate": 9.619397662556434e-06,
      "loss": 0.2478,
      "step": 2115
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3228810757601803,
      "learning_rate": 9.618954236424261e-06,
      "loss": 0.2158,
      "step": 2116
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5925570223070498,
      "learning_rate": 9.618510562364976e-06,
      "loss": 0.2082,
      "step": 2117
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.388580412686704,
      "learning_rate": 9.618066640402391e-06,
      "loss": 0.2177,
      "step": 2118
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3804031371423213,
      "learning_rate": 9.617622470560334e-06,
      "loss": 0.1707,
      "step": 2119
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4422215330788848,
      "learning_rate": 9.617178052862648e-06,
      "loss": 0.1889,
      "step": 2120
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.392170256819738,
      "learning_rate": 9.616733387333189e-06,
      "loss": 0.7937,
      "step": 2121
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3966604831005234,
      "learning_rate": 9.616288473995822e-06,
      "loss": 0.1766,
      "step": 2122
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.466240918188498,
      "learning_rate": 9.61584331287443e-06,
      "loss": 0.1906,
      "step": 2123
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4200473256976338,
      "learning_rate": 9.615397903992906e-06,
      "loss": 0.1935,
      "step": 2124
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6893118449916187,
      "learning_rate": 9.614952247375158e-06,
      "loss": 0.2318,
      "step": 2125
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.4697299949041565,
      "learning_rate": 9.614506343045109e-06,
      "loss": 0.7798,
      "step": 2126
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.484481472000296,
      "learning_rate": 9.614060191026691e-06,
      "loss": 0.1986,
      "step": 2127
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4329206635683895,
      "learning_rate": 9.613613791343852e-06,
      "loss": 0.2473,
      "step": 2128
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3978220860278685,
      "learning_rate": 9.613167144020554e-06,
      "loss": 0.2318,
      "step": 2129
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.385632726926145,
      "learning_rate": 9.612720249080772e-06,
      "loss": 0.1472,
      "step": 2130
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4644422568869802,
      "learning_rate": 9.612273106548492e-06,
      "loss": 0.2431,
      "step": 2131
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.921743720429038,
      "learning_rate": 9.611825716447716e-06,
      "loss": 0.7197,
      "step": 2132
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3735798927026437,
      "learning_rate": 9.611378078802456e-06,
      "loss": 0.1952,
      "step": 2133
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3420663158371056,
      "learning_rate": 9.610930193636741e-06,
      "loss": 0.22,
      "step": 2134
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3740066697000843,
      "learning_rate": 9.61048206097461e-06,
      "loss": 0.2451,
      "step": 2135
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.5179349080410685,
      "learning_rate": 9.61003368084012e-06,
      "loss": 0.5987,
      "step": 2136
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2803752877068484,
      "learning_rate": 9.609585053257338e-06,
      "loss": 0.1842,
      "step": 2137
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2988091439402814,
      "learning_rate": 9.609136178250341e-06,
      "loss": 0.19,
      "step": 2138
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5157033116022467,
      "learning_rate": 9.608687055843227e-06,
      "loss": 0.2495,
      "step": 2139
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4391736210882573,
      "learning_rate": 9.608237686060099e-06,
      "loss": 0.2192,
      "step": 2140
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3465171887198473,
      "learning_rate": 9.607788068925082e-06,
      "loss": 0.1989,
      "step": 2141
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6297583298179183,
      "learning_rate": 9.607338204462304e-06,
      "loss": 0.2385,
      "step": 2142
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4628956113042402,
      "learning_rate": 9.606888092695918e-06,
      "loss": 0.2199,
      "step": 2143
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4508188429794666,
      "learning_rate": 9.60643773365008e-06,
      "loss": 0.2295,
      "step": 2144
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.1619909242988555,
      "learning_rate": 9.605987127348966e-06,
      "loss": 0.6645,
      "step": 2145
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5262800427538348,
      "learning_rate": 9.605536273816761e-06,
      "loss": 0.2347,
      "step": 2146
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.403326403440822,
      "learning_rate": 9.605085173077665e-06,
      "loss": 0.1822,
      "step": 2147
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6236197888154558,
      "learning_rate": 9.604633825155894e-06,
      "loss": 0.2784,
      "step": 2148
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5668373164226215,
      "learning_rate": 9.604182230075671e-06,
      "loss": 0.2576,
      "step": 2149
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5103560692643814,
      "learning_rate": 9.603730387861236e-06,
      "loss": 0.2306,
      "step": 2150
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.446110275513765,
      "learning_rate": 9.603278298536845e-06,
      "loss": 0.2435,
      "step": 2151
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7267610687185202,
      "learning_rate": 9.602825962126763e-06,
      "loss": 0.3105,
      "step": 2152
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4063152208624514,
      "learning_rate": 9.60237337865527e-06,
      "loss": 0.1869,
      "step": 2153
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.467851398019143,
      "learning_rate": 9.601920548146657e-06,
      "loss": 0.2376,
      "step": 2154
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.9872533928032836,
      "learning_rate": 9.601467470625232e-06,
      "loss": 0.4816,
      "step": 2155
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.555080988945281,
      "learning_rate": 9.601014146115313e-06,
      "loss": 0.2054,
      "step": 2156
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5961190183763232,
      "learning_rate": 9.600560574641234e-06,
      "loss": 0.2371,
      "step": 2157
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4573350717067362,
      "learning_rate": 9.60010675622734e-06,
      "loss": 0.2573,
      "step": 2158
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5400737107352143,
      "learning_rate": 9.599652690897991e-06,
      "loss": 0.2676,
      "step": 2159
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4364573214618583,
      "learning_rate": 9.599198378677559e-06,
      "loss": 0.2496,
      "step": 2160
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4147367597953493,
      "learning_rate": 9.598743819590431e-06,
      "loss": 0.2285,
      "step": 2161
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7288629502779715,
      "learning_rate": 9.598289013661002e-06,
      "loss": 0.2787,
      "step": 2162
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.510985342804304,
      "learning_rate": 9.59783396091369e-06,
      "loss": 0.2115,
      "step": 2163
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5818735681933305,
      "learning_rate": 9.597378661372914e-06,
      "loss": 0.2377,
      "step": 2164
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4556521929735387,
      "learning_rate": 9.596923115063116e-06,
      "loss": 0.201,
      "step": 2165
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.947415319075834,
      "learning_rate": 9.596467322008752e-06,
      "loss": 0.7998,
      "step": 2166
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.377206368544645,
      "learning_rate": 9.59601128223428e-06,
      "loss": 0.1867,
      "step": 2167
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3843544245966157,
      "learning_rate": 9.59555499576418e-06,
      "loss": 0.2023,
      "step": 2168
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2793364459015888,
      "learning_rate": 9.595098462622947e-06,
      "loss": 0.2068,
      "step": 2169
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.334474245759089,
      "learning_rate": 9.594641682835082e-06,
      "loss": 0.194,
      "step": 2170
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4099435673567542,
      "learning_rate": 9.594184656425108e-06,
      "loss": 0.2998,
      "step": 2171
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4805270547796043,
      "learning_rate": 9.593727383417551e-06,
      "loss": 0.2253,
      "step": 2172
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4647741759507256,
      "learning_rate": 9.593269863836958e-06,
      "loss": 0.2541,
      "step": 2173
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.471328968626989,
      "learning_rate": 9.592812097707887e-06,
      "loss": 0.2403,
      "step": 2174
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5119445369724385,
      "learning_rate": 9.592354085054909e-06,
      "loss": 0.2601,
      "step": 2175
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.620294134942793,
      "learning_rate": 9.591895825902608e-06,
      "loss": 0.2765,
      "step": 2176
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4184184102522184,
      "learning_rate": 9.591437320275582e-06,
      "loss": 0.2414,
      "step": 2177
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3041445292109501,
      "learning_rate": 9.590978568198442e-06,
      "loss": 0.1913,
      "step": 2178
    },
    {
      "epoch": 0.16,
      "grad_norm": 23.208818809455042,
      "learning_rate": 9.590519569695811e-06,
      "loss": 0.591,
      "step": 2179
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4886497907249971,
      "learning_rate": 9.590060324792328e-06,
      "loss": 0.2725,
      "step": 2180
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3647301572498627,
      "learning_rate": 9.58960083351264e-06,
      "loss": 0.2245,
      "step": 2181
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4477646794058292,
      "learning_rate": 9.589141095881414e-06,
      "loss": 0.2022,
      "step": 2182
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6160086935823856,
      "learning_rate": 9.588681111923325e-06,
      "loss": 0.2522,
      "step": 2183
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4444482884581678,
      "learning_rate": 9.588220881663065e-06,
      "loss": 0.2176,
      "step": 2184
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3620014487010388,
      "learning_rate": 9.587760405125334e-06,
      "loss": 0.1712,
      "step": 2185
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.359346968575899,
      "learning_rate": 9.587299682334852e-06,
      "loss": 0.206,
      "step": 2186
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4371545126014544,
      "learning_rate": 9.586838713316347e-06,
      "loss": 0.238,
      "step": 2187
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.354100736177541,
      "learning_rate": 9.586377498094562e-06,
      "loss": 0.1835,
      "step": 2188
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.780242886017275,
      "learning_rate": 9.585916036694255e-06,
      "loss": 0.8067,
      "step": 2189
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4978420944151323,
      "learning_rate": 9.585454329140194e-06,
      "loss": 0.2111,
      "step": 2190
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.393404039702627,
      "learning_rate": 9.58499237545716e-06,
      "loss": 0.2367,
      "step": 2191
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.399419646336928,
      "learning_rate": 9.58453017566995e-06,
      "loss": 0.2605,
      "step": 2192
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4901533352427132,
      "learning_rate": 9.584067729803376e-06,
      "loss": 0.2039,
      "step": 2193
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7904135191761406,
      "learning_rate": 9.583605037882257e-06,
      "loss": 0.261,
      "step": 2194
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4080158745645563,
      "learning_rate": 9.583142099931429e-06,
      "loss": 0.2319,
      "step": 2195
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.299051816783134,
      "learning_rate": 9.582678915975741e-06,
      "loss": 0.4772,
      "step": 2196
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2750240411130134,
      "learning_rate": 9.582215486040054e-06,
      "loss": 0.179,
      "step": 2197
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2674226676161087,
      "learning_rate": 9.581751810149244e-06,
      "loss": 0.1828,
      "step": 2198
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5105202069001844,
      "learning_rate": 9.581287888328201e-06,
      "loss": 0.2369,
      "step": 2199
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4065515902343977,
      "learning_rate": 9.580823720601824e-06,
      "loss": 0.2012,
      "step": 2200
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.308827828233978,
      "learning_rate": 9.580359306995027e-06,
      "loss": 0.5322,
      "step": 2201
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4106749700153844,
      "learning_rate": 9.579894647532739e-06,
      "loss": 0.2052,
      "step": 2202
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.339725912282564,
      "learning_rate": 9.579429742239902e-06,
      "loss": 0.207,
      "step": 2203
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.805895094178684,
      "learning_rate": 9.57896459114147e-06,
      "loss": 0.4639,
      "step": 2204
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.3076711726276855,
      "learning_rate": 9.578499194262411e-06,
      "loss": 0.6018,
      "step": 2205
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.400498012085324,
      "learning_rate": 9.578033551627704e-06,
      "loss": 0.2205,
      "step": 2206
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.771131356279034,
      "learning_rate": 9.577567663262342e-06,
      "loss": 0.689,
      "step": 2207
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3879651581087789,
      "learning_rate": 9.577101529191335e-06,
      "loss": 0.1982,
      "step": 2208
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3198941720292972,
      "learning_rate": 9.576635149439702e-06,
      "loss": 0.213,
      "step": 2209
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4885534014756785,
      "learning_rate": 9.576168524032475e-06,
      "loss": 0.2252,
      "step": 2210
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7797932984174016,
      "learning_rate": 9.575701652994702e-06,
      "loss": 0.2704,
      "step": 2211
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3779436149619189,
      "learning_rate": 9.575234536351443e-06,
      "loss": 0.2143,
      "step": 2212
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3806700320484169,
      "learning_rate": 9.57476717412777e-06,
      "loss": 0.1828,
      "step": 2213
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4395511285494844,
      "learning_rate": 9.57429956634877e-06,
      "loss": 0.1982,
      "step": 2214
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4340096249209484,
      "learning_rate": 9.573831713039543e-06,
      "loss": 0.2749,
      "step": 2215
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3586778787873939,
      "learning_rate": 9.573363614225199e-06,
      "loss": 0.2008,
      "step": 2216
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4606402869453705,
      "learning_rate": 9.572895269930865e-06,
      "loss": 0.2245,
      "step": 2217
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3884084579078353,
      "learning_rate": 9.572426680181681e-06,
      "loss": 0.2175,
      "step": 2218
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.312240645966458,
      "learning_rate": 9.571957845002797e-06,
      "loss": 0.1707,
      "step": 2219
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4394167967969698,
      "learning_rate": 9.571488764419381e-06,
      "loss": 0.2166,
      "step": 2220
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4679816795649474,
      "learning_rate": 9.571019438456609e-06,
      "loss": 0.2219,
      "step": 2221
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4034172922758295,
      "learning_rate": 9.570549867139673e-06,
      "loss": 0.1969,
      "step": 2222
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6579497121557165,
      "learning_rate": 9.570080050493779e-06,
      "loss": 0.2022,
      "step": 2223
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3932417901644483,
      "learning_rate": 9.569609988544142e-06,
      "loss": 0.2058,
      "step": 2224
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.8150973154133854,
      "learning_rate": 9.569139681315996e-06,
      "loss": 0.5729,
      "step": 2225
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6828550068840442,
      "learning_rate": 9.568669128834584e-06,
      "loss": 0.2496,
      "step": 2226
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4118980168705837,
      "learning_rate": 9.568198331125163e-06,
      "loss": 0.2049,
      "step": 2227
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4423502717163474,
      "learning_rate": 9.567727288213005e-06,
      "loss": 0.2388,
      "step": 2228
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5837169968597165,
      "learning_rate": 9.567256000123393e-06,
      "loss": 0.2616,
      "step": 2229
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.422403736377754,
      "learning_rate": 9.566784466881622e-06,
      "loss": 0.2369,
      "step": 2230
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.39328323126279,
      "learning_rate": 9.566312688513006e-06,
      "loss": 0.2045,
      "step": 2231
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.265500887789139,
      "learning_rate": 9.565840665042865e-06,
      "loss": 0.2155,
      "step": 2232
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6337787653184104,
      "learning_rate": 9.565368396496535e-06,
      "loss": 0.2264,
      "step": 2233
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4370524952815555,
      "learning_rate": 9.564895882899368e-06,
      "loss": 0.2311,
      "step": 2234
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4402081614968072,
      "learning_rate": 9.564423124276726e-06,
      "loss": 0.2275,
      "step": 2235
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.581527249382296,
      "learning_rate": 9.563950120653982e-06,
      "loss": 0.2116,
      "step": 2236
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4870635290804242,
      "learning_rate": 9.563476872056527e-06,
      "loss": 0.2483,
      "step": 2237
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4491257163556845,
      "learning_rate": 9.563003378509766e-06,
      "loss": 0.2012,
      "step": 2238
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4315753392964596,
      "learning_rate": 9.56252964003911e-06,
      "loss": 0.2164,
      "step": 2239
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.663032144229193,
      "learning_rate": 9.562055656669988e-06,
      "loss": 0.4086,
      "step": 2240
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.49489835845098,
      "learning_rate": 9.561581428427845e-06,
      "loss": 0.2615,
      "step": 2241
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5743286836998294,
      "learning_rate": 9.56110695533813e-06,
      "loss": 0.2424,
      "step": 2242
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4463748757917514,
      "learning_rate": 9.560632237426315e-06,
      "loss": 0.2166,
      "step": 2243
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4105702113471734,
      "learning_rate": 9.56015727471788e-06,
      "loss": 0.2005,
      "step": 2244
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3193286373739412,
      "learning_rate": 9.55968206723832e-06,
      "loss": 0.19,
      "step": 2245
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5854581037097646,
      "learning_rate": 9.559206615013142e-06,
      "loss": 0.2252,
      "step": 2246
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3729059019083576,
      "learning_rate": 9.558730918067862e-06,
      "loss": 0.2139,
      "step": 2247
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4088530824548053,
      "learning_rate": 9.558254976428022e-06,
      "loss": 0.2354,
      "step": 2248
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6816387302356723,
      "learning_rate": 9.55777879011916e-06,
      "loss": 0.2182,
      "step": 2249
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.606334913339843,
      "learning_rate": 9.55730235916684e-06,
      "loss": 0.8392,
      "step": 2250
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3951464900498858,
      "learning_rate": 9.556825683596635e-06,
      "loss": 0.1977,
      "step": 2251
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3480952096502246,
      "learning_rate": 9.556348763434133e-06,
      "loss": 0.2074,
      "step": 2252
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.589784112360952,
      "learning_rate": 9.555871598704929e-06,
      "loss": 0.2202,
      "step": 2253
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3191573243599104,
      "learning_rate": 9.555394189434636e-06,
      "loss": 0.1743,
      "step": 2254
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3393415736325,
      "learning_rate": 9.554916535648882e-06,
      "loss": 0.2058,
      "step": 2255
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5263805213583772,
      "learning_rate": 9.554438637373304e-06,
      "loss": 0.2283,
      "step": 2256
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5427083119975071,
      "learning_rate": 9.553960494633552e-06,
      "loss": 0.2608,
      "step": 2257
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.473010941163431,
      "learning_rate": 9.553482107455296e-06,
      "loss": 0.2608,
      "step": 2258
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3096066081261102,
      "learning_rate": 9.553003475864207e-06,
      "loss": 0.2059,
      "step": 2259
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5464659159062626,
      "learning_rate": 9.552524599885982e-06,
      "loss": 0.2202,
      "step": 2260
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4793049959418219,
      "learning_rate": 9.552045479546322e-06,
      "loss": 0.2254,
      "step": 2261
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4201878945966728,
      "learning_rate": 9.551566114870945e-06,
      "loss": 0.237,
      "step": 2262
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6063762767942602,
      "learning_rate": 9.55108650588558e-06,
      "loss": 0.2365,
      "step": 2263
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.6444395662999645,
      "learning_rate": 9.550606652615973e-06,
      "loss": 0.585,
      "step": 2264
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.603321336434924,
      "learning_rate": 9.55012655508788e-06,
      "loss": 0.6692,
      "step": 2265
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3341201529460056,
      "learning_rate": 9.549646213327069e-06,
      "loss": 0.2236,
      "step": 2266
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3508297180273938,
      "learning_rate": 9.549165627359324e-06,
      "loss": 0.2057,
      "step": 2267
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.098366150594217,
      "learning_rate": 9.548684797210444e-06,
      "loss": 0.6355,
      "step": 2268
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4796161663181557,
      "learning_rate": 9.548203722906231e-06,
      "loss": 0.2184,
      "step": 2269
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5883090232935988,
      "learning_rate": 9.547722404472514e-06,
      "loss": 0.2564,
      "step": 2270
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4319785113897798,
      "learning_rate": 9.547240841935124e-06,
      "loss": 0.2121,
      "step": 2271
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3735242154725285,
      "learning_rate": 9.546759035319911e-06,
      "loss": 0.2086,
      "step": 2272
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.300715134219384,
      "learning_rate": 9.546276984652736e-06,
      "loss": 0.2026,
      "step": 2273
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3207040995008266,
      "learning_rate": 9.545794689959476e-06,
      "loss": 0.2287,
      "step": 2274
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3267397631179043,
      "learning_rate": 9.545312151266015e-06,
      "loss": 0.193,
      "step": 2275
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.193934742994709,
      "learning_rate": 9.544829368598254e-06,
      "loss": 0.522,
      "step": 2276
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4319616589425652,
      "learning_rate": 9.54434634198211e-06,
      "loss": 0.2054,
      "step": 2277
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3046697439284143,
      "learning_rate": 9.543863071443508e-06,
      "loss": 0.1814,
      "step": 2278
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5200117496686427,
      "learning_rate": 9.543379557008388e-06,
      "loss": 0.2269,
      "step": 2279
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.791963277922103,
      "learning_rate": 9.542895798702702e-06,
      "loss": 0.5867,
      "step": 2280
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.451737868958123,
      "learning_rate": 9.542411796552418e-06,
      "loss": 0.2409,
      "step": 2281
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.451508274389207,
      "learning_rate": 9.541927550583516e-06,
      "loss": 0.22,
      "step": 2282
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.311063570759236,
      "learning_rate": 9.541443060821986e-06,
      "loss": 0.1757,
      "step": 2283
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.439030798929904,
      "learning_rate": 9.540958327293835e-06,
      "loss": 0.2265,
      "step": 2284
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5174749622001882,
      "learning_rate": 9.540473350025082e-06,
      "loss": 0.2811,
      "step": 2285
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3029114176835457,
      "learning_rate": 9.539988129041757e-06,
      "loss": 0.2066,
      "step": 2286
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.460803829409455,
      "learning_rate": 9.539502664369905e-06,
      "loss": 0.2088,
      "step": 2287
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.607682359154701,
      "learning_rate": 9.539016956035588e-06,
      "loss": 0.7175,
      "step": 2288
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5433623722890235,
      "learning_rate": 9.538531004064869e-06,
      "loss": 0.2437,
      "step": 2289
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.649202903570783,
      "learning_rate": 9.538044808483839e-06,
      "loss": 0.2503,
      "step": 2290
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3853485820054525,
      "learning_rate": 9.537558369318591e-06,
      "loss": 0.2041,
      "step": 2291
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3820991853368472,
      "learning_rate": 9.537071686595237e-06,
      "loss": 0.1705,
      "step": 2292
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6489403547185333,
      "learning_rate": 9.5365847603399e-06,
      "loss": 0.2372,
      "step": 2293
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5512020947576006,
      "learning_rate": 9.536097590578715e-06,
      "loss": 0.2594,
      "step": 2294
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.391941919949618,
      "learning_rate": 9.535610177337833e-06,
      "loss": 0.2189,
      "step": 2295
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.656919616128035,
      "learning_rate": 9.535122520643415e-06,
      "loss": 0.2066,
      "step": 2296
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2612389920459641,
      "learning_rate": 9.53463462052164e-06,
      "loss": 0.2261,
      "step": 2297
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5060811977949595,
      "learning_rate": 9.53414647699869e-06,
      "loss": 0.2145,
      "step": 2298
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.475487046889574,
      "learning_rate": 9.533658090100772e-06,
      "loss": 0.2033,
      "step": 2299
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.298480149515627,
      "learning_rate": 9.5331694598541e-06,
      "loss": 0.2327,
      "step": 2300
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4230620937762193,
      "learning_rate": 9.532680586284898e-06,
      "loss": 0.2116,
      "step": 2301
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.614627635963967,
      "learning_rate": 9.532191469419412e-06,
      "loss": 0.2549,
      "step": 2302
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.157347237314944,
      "learning_rate": 9.531702109283892e-06,
      "loss": 0.6041,
      "step": 2303
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.331892603121996,
      "learning_rate": 9.531212505904606e-06,
      "loss": 0.233,
      "step": 2304
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4304896330775478,
      "learning_rate": 9.530722659307836e-06,
      "loss": 0.1883,
      "step": 2305
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3229709592839418,
      "learning_rate": 9.530232569519871e-06,
      "loss": 0.2369,
      "step": 2306
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5418363645160602,
      "learning_rate": 9.529742236567022e-06,
      "loss": 0.2712,
      "step": 2307
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.319075302257437,
      "learning_rate": 9.529251660475602e-06,
      "loss": 0.2089,
      "step": 2308
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5588370866192451,
      "learning_rate": 9.528760841271949e-06,
      "loss": 0.2331,
      "step": 2309
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4886050158019533,
      "learning_rate": 9.528269778982407e-06,
      "loss": 0.2104,
      "step": 2310
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.450455333423547,
      "learning_rate": 9.527778473633331e-06,
      "loss": 0.2342,
      "step": 2311
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4106105093077694,
      "learning_rate": 9.527286925251095e-06,
      "loss": 0.2478,
      "step": 2312
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5206971409482772,
      "learning_rate": 9.526795133862083e-06,
      "loss": 0.2183,
      "step": 2313
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4685942438699602,
      "learning_rate": 9.52630309949269e-06,
      "loss": 0.1986,
      "step": 2314
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.640275774890013,
      "learning_rate": 9.525810822169332e-06,
      "loss": 0.5119,
      "step": 2315
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.480412117304701,
      "learning_rate": 9.525318301918427e-06,
      "loss": 0.2569,
      "step": 2316
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4772556125522698,
      "learning_rate": 9.524825538766417e-06,
      "loss": 0.2553,
      "step": 2317
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.670378961226955,
      "learning_rate": 9.524332532739745e-06,
      "loss": 0.689,
      "step": 2318
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.420136427022906,
      "learning_rate": 9.523839283864878e-06,
      "loss": 0.2035,
      "step": 2319
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4263542873993964,
      "learning_rate": 9.52334579216829e-06,
      "loss": 0.2283,
      "step": 2320
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.421509811738969,
      "learning_rate": 9.52285205767647e-06,
      "loss": 0.2099,
      "step": 2321
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.626621512711122,
      "learning_rate": 9.52235808041592e-06,
      "loss": 0.2682,
      "step": 2322
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3813766297591432,
      "learning_rate": 9.521863860413154e-06,
      "loss": 0.2186,
      "step": 2323
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.695258738944951,
      "learning_rate": 9.521369397694702e-06,
      "loss": 0.5931,
      "step": 2324
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5706953274986557,
      "learning_rate": 9.520874692287102e-06,
      "loss": 0.2432,
      "step": 2325
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4804008829469815,
      "learning_rate": 9.52037974421691e-06,
      "loss": 0.2149,
      "step": 2326
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3059048351611338,
      "learning_rate": 9.519884553510691e-06,
      "loss": 0.2414,
      "step": 2327
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5082721546336773,
      "learning_rate": 9.519389120195026e-06,
      "loss": 0.1941,
      "step": 2328
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3995581653430642,
      "learning_rate": 9.518893444296508e-06,
      "loss": 0.2479,
      "step": 2329
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4276169521955404,
      "learning_rate": 9.518397525841742e-06,
      "loss": 0.1997,
      "step": 2330
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4850931045413314,
      "learning_rate": 9.517901364857347e-06,
      "loss": 0.2036,
      "step": 2331
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4840980805794286,
      "learning_rate": 9.517404961369956e-06,
      "loss": 0.2858,
      "step": 2332
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5050591665709097,
      "learning_rate": 9.516908315406212e-06,
      "loss": 0.2082,
      "step": 2333
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.994365483778152,
      "learning_rate": 9.516411426992777e-06,
      "loss": 0.5814,
      "step": 2334
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5322474640885728,
      "learning_rate": 9.515914296156318e-06,
      "loss": 0.2359,
      "step": 2335
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1119811688905632,
      "learning_rate": 9.51541692292352e-06,
      "loss": 0.1584,
      "step": 2336
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.683876452595023,
      "learning_rate": 9.514919307321081e-06,
      "loss": 0.2507,
      "step": 2337
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4362818724730777,
      "learning_rate": 9.514421449375711e-06,
      "loss": 0.2034,
      "step": 2338
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.38403440517796,
      "learning_rate": 9.513923349114131e-06,
      "loss": 0.2209,
      "step": 2339
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4170858126054215,
      "learning_rate": 9.51342500656308e-06,
      "loss": 0.2178,
      "step": 2340
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2949232073157406,
      "learning_rate": 9.512926421749305e-06,
      "loss": 0.2384,
      "step": 2341
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4943835263499077,
      "learning_rate": 9.512427594699567e-06,
      "loss": 0.2751,
      "step": 2342
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3302230009239429,
      "learning_rate": 9.511928525440644e-06,
      "loss": 0.1798,
      "step": 2343
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4805598699583216,
      "learning_rate": 9.511429213999323e-06,
      "loss": 0.222,
      "step": 2344
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3386830201887079,
      "learning_rate": 9.510929660402404e-06,
      "loss": 0.1468,
      "step": 2345
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5417678324419908,
      "learning_rate": 9.510429864676702e-06,
      "loss": 0.3038,
      "step": 2346
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.158266688800714,
      "learning_rate": 9.509929826849044e-06,
      "loss": 0.9248,
      "step": 2347
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.5280252168687865,
      "learning_rate": 9.509429546946269e-06,
      "loss": 0.7129,
      "step": 2348
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4210087367325064,
      "learning_rate": 9.508929024995233e-06,
      "loss": 0.2514,
      "step": 2349
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.414547666014133,
      "learning_rate": 9.508428261022798e-06,
      "loss": 0.2425,
      "step": 2350
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4292607312526646,
      "learning_rate": 9.507927255055847e-06,
      "loss": 0.2218,
      "step": 2351
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6571863115922023,
      "learning_rate": 9.507426007121268e-06,
      "loss": 0.262,
      "step": 2352
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3796186401412138,
      "learning_rate": 9.506924517245968e-06,
      "loss": 0.2312,
      "step": 2353
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.411856797638386,
      "learning_rate": 9.506422785456865e-06,
      "loss": 0.2176,
      "step": 2354
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3839233728749893,
      "learning_rate": 9.505920811780892e-06,
      "loss": 0.2041,
      "step": 2355
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2814223901795698,
      "learning_rate": 9.505418596244989e-06,
      "loss": 0.205,
      "step": 2356
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4325012609939118,
      "learning_rate": 9.504916138876115e-06,
      "loss": 0.2397,
      "step": 2357
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.969135896429005,
      "learning_rate": 9.504413439701241e-06,
      "loss": 0.5681,
      "step": 2358
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5145294294113256,
      "learning_rate": 9.503910498747348e-06,
      "loss": 0.2047,
      "step": 2359
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.525319043785778,
      "learning_rate": 9.503407316041432e-06,
      "loss": 0.2468,
      "step": 2360
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.389277161363658,
      "learning_rate": 9.502903891610502e-06,
      "loss": 0.6212,
      "step": 2361
    },
    {
      "epoch": 0.17,
      "grad_norm": 12.682134048053712,
      "learning_rate": 9.502400225481582e-06,
      "loss": 0.4795,
      "step": 2362
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5882725996764977,
      "learning_rate": 9.501896317681703e-06,
      "loss": 0.2165,
      "step": 2363
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.501100114827806,
      "learning_rate": 9.501392168237914e-06,
      "loss": 0.1984,
      "step": 2364
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7440042880270596,
      "learning_rate": 9.50088777717728e-06,
      "loss": 0.2512,
      "step": 2365
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.427499099272823,
      "learning_rate": 9.500383144526868e-06,
      "loss": 0.2257,
      "step": 2366
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.127610435126591,
      "learning_rate": 9.49987827031377e-06,
      "loss": 0.5424,
      "step": 2367
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4848357718694933,
      "learning_rate": 9.499373154565083e-06,
      "loss": 0.1879,
      "step": 2368
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4921283913114312,
      "learning_rate": 9.49886779730792e-06,
      "loss": 0.2291,
      "step": 2369
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4147138680821634,
      "learning_rate": 9.498362198569407e-06,
      "loss": 0.2286,
      "step": 2370
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4997754989648011,
      "learning_rate": 9.497856358376681e-06,
      "loss": 0.2574,
      "step": 2371
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4167895328550235,
      "learning_rate": 9.497350276756895e-06,
      "loss": 0.206,
      "step": 2372
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3337212853600846,
      "learning_rate": 9.496843953737214e-06,
      "loss": 0.1751,
      "step": 2373
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5150807321607511,
      "learning_rate": 9.496337389344814e-06,
      "loss": 0.2264,
      "step": 2374
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3414188971494687,
      "learning_rate": 9.495830583606887e-06,
      "loss": 0.2495,
      "step": 2375
    },
    {
      "epoch": 0.17,
      "grad_norm": 13.046149377746191,
      "learning_rate": 9.495323536550633e-06,
      "loss": 0.7607,
      "step": 2376
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5449413778195826,
      "learning_rate": 9.494816248203272e-06,
      "loss": 0.2557,
      "step": 2377
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.38756764183356,
      "learning_rate": 9.494308718592032e-06,
      "loss": 0.2076,
      "step": 2378
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.64804574743153,
      "learning_rate": 9.493800947744154e-06,
      "loss": 0.2555,
      "step": 2379
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4306783394738056,
      "learning_rate": 9.493292935686896e-06,
      "loss": 0.2287,
      "step": 2380
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5434781545497618,
      "learning_rate": 9.492784682447521e-06,
      "loss": 0.1985,
      "step": 2381
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.429361794605198,
      "learning_rate": 9.492276188053315e-06,
      "loss": 0.2095,
      "step": 2382
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4119075397388192,
      "learning_rate": 9.491767452531571e-06,
      "loss": 0.1956,
      "step": 2383
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5556630616540017,
      "learning_rate": 9.491258475909593e-06,
      "loss": 0.2357,
      "step": 2384
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.290056654928571,
      "learning_rate": 9.490749258214704e-06,
      "loss": 0.1729,
      "step": 2385
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2675972315226016,
      "learning_rate": 9.490239799474237e-06,
      "loss": 0.1736,
      "step": 2386
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5234740070937312,
      "learning_rate": 9.489730099715534e-06,
      "loss": 0.2445,
      "step": 2387
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3683308225226825,
      "learning_rate": 9.489220158965957e-06,
      "loss": 0.2021,
      "step": 2388
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3673711689337822,
      "learning_rate": 9.488709977252876e-06,
      "loss": 0.1986,
      "step": 2389
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6042250249500396,
      "learning_rate": 9.488199554603676e-06,
      "loss": 0.2391,
      "step": 2390
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6592866741284222,
      "learning_rate": 9.487688891045756e-06,
      "loss": 0.2754,
      "step": 2391
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4318310830364822,
      "learning_rate": 9.487177986606525e-06,
      "loss": 0.2279,
      "step": 2392
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4519554997433146,
      "learning_rate": 9.486666841313404e-06,
      "loss": 0.2068,
      "step": 2393
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4027138201387752,
      "learning_rate": 9.486155455193833e-06,
      "loss": 0.2508,
      "step": 2394
    },
    {
      "epoch": 0.17,
      "grad_norm": 13.617430246264899,
      "learning_rate": 9.485643828275259e-06,
      "loss": 0.625,
      "step": 2395
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5400345259920187,
      "learning_rate": 9.485131960585146e-06,
      "loss": 0.2669,
      "step": 2396
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3174053728272423,
      "learning_rate": 9.484619852150967e-06,
      "loss": 0.19,
      "step": 2397
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4571704252550244,
      "learning_rate": 9.484107503000211e-06,
      "loss": 0.2282,
      "step": 2398
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3275148469213154,
      "learning_rate": 9.48359491316038e-06,
      "loss": 0.2118,
      "step": 2399
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5524317178502713,
      "learning_rate": 9.483082082658984e-06,
      "loss": 0.2523,
      "step": 2400
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.278269845411248,
      "learning_rate": 9.482569011523553e-06,
      "loss": 0.1922,
      "step": 2401
    },
    {
      "epoch": 0.17,
      "grad_norm": 19.28521233933641,
      "learning_rate": 9.482055699781625e-06,
      "loss": 0.5399,
      "step": 2402
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5315347033034197,
      "learning_rate": 9.481542147460756e-06,
      "loss": 0.2118,
      "step": 2403
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3555767356865305,
      "learning_rate": 9.481028354588506e-06,
      "loss": 0.1871,
      "step": 2404
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4678777874055562,
      "learning_rate": 9.480514321192458e-06,
      "loss": 0.2247,
      "step": 2405
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.159926088978146,
      "learning_rate": 9.4800000473002e-06,
      "loss": 0.6156,
      "step": 2406
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3424263045263272,
      "learning_rate": 9.479485532939338e-06,
      "loss": 0.2224,
      "step": 2407
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3877855771842444,
      "learning_rate": 9.478970778137488e-06,
      "loss": 0.2031,
      "step": 2408
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3809372649674194,
      "learning_rate": 9.478455782922282e-06,
      "loss": 0.228,
      "step": 2409
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4611839123048342,
      "learning_rate": 9.477940547321361e-06,
      "loss": 0.2212,
      "step": 2410
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4558949819790852,
      "learning_rate": 9.47742507136238e-06,
      "loss": 0.2385,
      "step": 2411
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.565317403500844,
      "learning_rate": 9.476909355073012e-06,
      "loss": 0.2503,
      "step": 2412
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.645823851278893,
      "learning_rate": 9.476393398480933e-06,
      "loss": 0.5941,
      "step": 2413
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5182986320540979,
      "learning_rate": 9.47587720161384e-06,
      "loss": 0.2642,
      "step": 2414
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.234095356823936,
      "learning_rate": 9.475360764499442e-06,
      "loss": 0.2427,
      "step": 2415
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3656075962229723,
      "learning_rate": 9.474844087165458e-06,
      "loss": 0.1975,
      "step": 2416
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.232820518586282,
      "learning_rate": 9.47432716963962e-06,
      "loss": 0.6512,
      "step": 2417
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4423249203985848,
      "learning_rate": 9.473810011949676e-06,
      "loss": 0.2517,
      "step": 2418
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5993249880140628,
      "learning_rate": 9.473292614123384e-06,
      "loss": 0.1969,
      "step": 2419
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3806214622715274,
      "learning_rate": 9.472774976188515e-06,
      "loss": 0.2072,
      "step": 2420
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3504150106556574,
      "learning_rate": 9.472257098172856e-06,
      "loss": 0.2504,
      "step": 2421
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3256305389782437,
      "learning_rate": 9.471738980104202e-06,
      "loss": 0.2501,
      "step": 2422
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.893047176166037,
      "learning_rate": 9.471220622010366e-06,
      "loss": 0.2438,
      "step": 2423
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.276670629899183,
      "learning_rate": 9.47070202391917e-06,
      "loss": 0.162,
      "step": 2424
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5753352557350828,
      "learning_rate": 9.470183185858447e-06,
      "loss": 0.2216,
      "step": 2425
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.394180061701923,
      "learning_rate": 9.469664107856054e-06,
      "loss": 0.2301,
      "step": 2426
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3917635561478825,
      "learning_rate": 9.469144789939848e-06,
      "loss": 0.2055,
      "step": 2427
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5224918506057952,
      "learning_rate": 9.468625232137703e-06,
      "loss": 0.2842,
      "step": 2428
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4027595752722253,
      "learning_rate": 9.46810543447751e-06,
      "loss": 0.1792,
      "step": 2429
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5610562148275493,
      "learning_rate": 9.467585396987168e-06,
      "loss": 0.2396,
      "step": 2430
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5809477351904575,
      "learning_rate": 9.467065119694589e-06,
      "loss": 0.2505,
      "step": 2431
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4558217715441395,
      "learning_rate": 9.466544602627703e-06,
      "loss": 0.189,
      "step": 2432
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7046440626205586,
      "learning_rate": 9.466023845814446e-06,
      "loss": 0.2188,
      "step": 2433
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3646207248336062,
      "learning_rate": 9.465502849282773e-06,
      "loss": 0.1969,
      "step": 2434
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3955047251744517,
      "learning_rate": 9.464981613060648e-06,
      "loss": 0.1948,
      "step": 2435
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.394756048873777,
      "learning_rate": 9.464460137176047e-06,
      "loss": 0.1959,
      "step": 2436
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5448350162022106,
      "learning_rate": 9.463938421656962e-06,
      "loss": 0.2363,
      "step": 2437
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4074420286439586,
      "learning_rate": 9.463416466531397e-06,
      "loss": 0.1976,
      "step": 2438
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.397981760304576,
      "learning_rate": 9.46289427182737e-06,
      "loss": 0.2439,
      "step": 2439
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.5910820732962385,
      "learning_rate": 9.462371837572907e-06,
      "loss": 0.8099,
      "step": 2440
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3424003112067515,
      "learning_rate": 9.461849163796052e-06,
      "loss": 0.2006,
      "step": 2441
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3212945336196555,
      "learning_rate": 9.461326250524861e-06,
      "loss": 0.2031,
      "step": 2442
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4978260459150183,
      "learning_rate": 9.460803097787398e-06,
      "loss": 0.2161,
      "step": 2443
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.386586616899618,
      "learning_rate": 9.460279705611748e-06,
      "loss": 0.2189,
      "step": 2444
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.765708090831872,
      "learning_rate": 9.459756074026002e-06,
      "loss": 0.5152,
      "step": 2445
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4784525288847616,
      "learning_rate": 9.45923220305827e-06,
      "loss": 0.2042,
      "step": 2446
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4685467977243203,
      "learning_rate": 9.458708092736669e-06,
      "loss": 0.2203,
      "step": 2447
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.210446188253907,
      "learning_rate": 9.45818374308933e-06,
      "loss": 0.2261,
      "step": 2448
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4903827313726299,
      "learning_rate": 9.457659154144398e-06,
      "loss": 0.1725,
      "step": 2449
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.729920456377542,
      "learning_rate": 9.457134325930034e-06,
      "loss": 0.2202,
      "step": 2450
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3130165283240416,
      "learning_rate": 9.456609258474406e-06,
      "loss": 0.2471,
      "step": 2451
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4556616212347169,
      "learning_rate": 9.456083951805698e-06,
      "loss": 0.2284,
      "step": 2452
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4520219612286838,
      "learning_rate": 9.455558405952105e-06,
      "loss": 0.2317,
      "step": 2453
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4547885933196265,
      "learning_rate": 9.45503262094184e-06,
      "loss": 0.2229,
      "step": 2454
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3152056381543786,
      "learning_rate": 9.454506596803122e-06,
      "loss": 0.2263,
      "step": 2455
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3395566946378628,
      "learning_rate": 9.453980333564187e-06,
      "loss": 0.191,
      "step": 2456
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3764276354089426,
      "learning_rate": 9.453453831253282e-06,
      "loss": 0.214,
      "step": 2457
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.47030020912995,
      "learning_rate": 9.452927089898667e-06,
      "loss": 0.1772,
      "step": 2458
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3515915272953936,
      "learning_rate": 9.452400109528619e-06,
      "loss": 0.2668,
      "step": 2459
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4709508758961785,
      "learning_rate": 9.451872890171419e-06,
      "loss": 0.202,
      "step": 2460
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5699363811627245,
      "learning_rate": 9.45134543185537e-06,
      "loss": 0.2342,
      "step": 2461
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4367649720743676,
      "learning_rate": 9.45081773460878e-06,
      "loss": 0.2186,
      "step": 2462
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5257438842110773,
      "learning_rate": 9.45028979845998e-06,
      "loss": 0.2453,
      "step": 2463
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.554377577091035,
      "learning_rate": 9.449761623437302e-06,
      "loss": 0.2319,
      "step": 2464
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.334766203130036,
      "learning_rate": 9.449233209569098e-06,
      "loss": 0.2022,
      "step": 2465
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6710315748597055,
      "learning_rate": 9.448704556883731e-06,
      "loss": 0.2428,
      "step": 2466
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.93872037486805,
      "learning_rate": 9.448175665409578e-06,
      "loss": 0.5796,
      "step": 2467
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4304675392597739,
      "learning_rate": 9.447646535175027e-06,
      "loss": 0.1764,
      "step": 2468
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5910803662028452,
      "learning_rate": 9.447117166208477e-06,
      "loss": 0.2214,
      "step": 2469
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.475957015041202,
      "learning_rate": 9.446587558538346e-06,
      "loss": 0.2208,
      "step": 2470
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4375360789950442,
      "learning_rate": 9.446057712193063e-06,
      "loss": 0.2124,
      "step": 2471
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3816179743264985,
      "learning_rate": 9.445527627201064e-06,
      "loss": 0.2246,
      "step": 2472
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4172040387055247,
      "learning_rate": 9.444997303590802e-06,
      "loss": 0.2267,
      "step": 2473
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.55519095085811,
      "learning_rate": 9.444466741390743e-06,
      "loss": 0.2109,
      "step": 2474
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.417476806387846,
      "learning_rate": 9.443935940629368e-06,
      "loss": 0.2168,
      "step": 2475
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.458826150167142,
      "learning_rate": 9.443404901335167e-06,
      "loss": 0.2439,
      "step": 2476
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4846456027069463,
      "learning_rate": 9.442873623536642e-06,
      "loss": 0.2677,
      "step": 2477
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2692713615137534,
      "learning_rate": 9.44234210726231e-06,
      "loss": 0.1849,
      "step": 2478
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3879901088283517,
      "learning_rate": 9.441810352540705e-06,
      "loss": 0.168,
      "step": 2479
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8379916711988598,
      "learning_rate": 9.441278359400366e-06,
      "loss": 0.2587,
      "step": 2480
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2768327501037542,
      "learning_rate": 9.440746127869849e-06,
      "loss": 0.194,
      "step": 2481
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6528708633444582,
      "learning_rate": 9.440213657977721e-06,
      "loss": 0.2457,
      "step": 2482
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.582112741107105,
      "learning_rate": 9.439680949752566e-06,
      "loss": 0.7004,
      "step": 2483
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5742574477787226,
      "learning_rate": 9.439148003222973e-06,
      "loss": 0.2496,
      "step": 2484
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3700644895289067,
      "learning_rate": 9.438614818417553e-06,
      "loss": 0.225,
      "step": 2485
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3939017135069647,
      "learning_rate": 9.438081395364922e-06,
      "loss": 0.224,
      "step": 2486
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2183299732774795,
      "learning_rate": 9.437547734093714e-06,
      "loss": 0.1613,
      "step": 2487
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4507348011498014,
      "learning_rate": 9.437013834632573e-06,
      "loss": 0.2167,
      "step": 2488
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3593412855339093,
      "learning_rate": 9.436479697010156e-06,
      "loss": 0.2003,
      "step": 2489
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3522448468440682,
      "learning_rate": 9.435945321255135e-06,
      "loss": 0.1857,
      "step": 2490
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.446281182995164,
      "learning_rate": 9.435410707396192e-06,
      "loss": 0.1955,
      "step": 2491
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3487644674149917,
      "learning_rate": 9.434875855462023e-06,
      "loss": 0.2415,
      "step": 2492
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.477806549360147,
      "learning_rate": 9.434340765481335e-06,
      "loss": 0.1968,
      "step": 2493
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.950906532631057,
      "learning_rate": 9.433805437482854e-06,
      "loss": 0.7637,
      "step": 2494
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.309629615185627,
      "learning_rate": 9.433269871495311e-06,
      "loss": 0.2364,
      "step": 2495
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3308815833565133,
      "learning_rate": 9.432734067547454e-06,
      "loss": 0.1978,
      "step": 2496
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5893063479965046,
      "learning_rate": 9.432198025668043e-06,
      "loss": 0.2799,
      "step": 2497
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3364069514862345,
      "learning_rate": 9.431661745885848e-06,
      "loss": 0.2023,
      "step": 2498
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.394018826942934,
      "learning_rate": 9.431125228229658e-06,
      "loss": 0.218,
      "step": 2499
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.502472282375056,
      "learning_rate": 9.430588472728271e-06,
      "loss": 0.2273,
      "step": 2500
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.353080094648928,
      "learning_rate": 9.430051479410493e-06,
      "loss": 0.2206,
      "step": 2501
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.450405257071622,
      "learning_rate": 9.429514248305154e-06,
      "loss": 0.2033,
      "step": 2502
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4796355156151553,
      "learning_rate": 9.428976779441089e-06,
      "loss": 0.2095,
      "step": 2503
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4731564633678975,
      "learning_rate": 9.428439072847142e-06,
      "loss": 0.1713,
      "step": 2504
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3858268422265942,
      "learning_rate": 9.427901128552182e-06,
      "loss": 0.224,
      "step": 2505
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4999664683552298,
      "learning_rate": 9.427362946585078e-06,
      "loss": 0.2538,
      "step": 2506
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.443221153338652,
      "learning_rate": 9.426824526974721e-06,
      "loss": 0.232,
      "step": 2507
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5154799523932865,
      "learning_rate": 9.426285869750012e-06,
      "loss": 0.2653,
      "step": 2508
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7375101278234002,
      "learning_rate": 9.42574697493986e-06,
      "loss": 0.2563,
      "step": 2509
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7278190930598922,
      "learning_rate": 9.425207842573193e-06,
      "loss": 0.2373,
      "step": 2510
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.112438188896315,
      "learning_rate": 9.42466847267895e-06,
      "loss": 0.7665,
      "step": 2511
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3670854493249174,
      "learning_rate": 9.424128865286082e-06,
      "loss": 0.2108,
      "step": 2512
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.386900899115097,
      "learning_rate": 9.423589020423554e-06,
      "loss": 0.2055,
      "step": 2513
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3364404562962722,
      "learning_rate": 9.423048938120338e-06,
      "loss": 0.2303,
      "step": 2514
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5204239142157632,
      "learning_rate": 9.422508618405431e-06,
      "loss": 0.2188,
      "step": 2515
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.447814845796347,
      "learning_rate": 9.421968061307826e-06,
      "loss": 0.24,
      "step": 2516
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.345980152762442,
      "learning_rate": 9.421427266856546e-06,
      "loss": 0.2072,
      "step": 2517
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4858305049235843,
      "learning_rate": 9.420886235080616e-06,
      "loss": 0.2051,
      "step": 2518
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.953166338104311,
      "learning_rate": 9.420344966009076e-06,
      "loss": 0.6718,
      "step": 2519
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4302932235919987,
      "learning_rate": 9.41980345967098e-06,
      "loss": 0.2218,
      "step": 2520
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5175740000448021,
      "learning_rate": 9.419261716095392e-06,
      "loss": 0.2535,
      "step": 2521
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3118175933393332,
      "learning_rate": 9.41871973531139e-06,
      "loss": 0.1868,
      "step": 2522
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5676814743103566,
      "learning_rate": 9.41817751734807e-06,
      "loss": 0.2724,
      "step": 2523
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.482636525205323,
      "learning_rate": 9.417635062234533e-06,
      "loss": 0.8307,
      "step": 2524
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.297961969341316,
      "learning_rate": 9.417092369999897e-06,
      "loss": 0.1962,
      "step": 2525
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4125358528613452,
      "learning_rate": 9.416549440673288e-06,
      "loss": 0.1973,
      "step": 2526
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.296176583011786,
      "learning_rate": 9.416006274283853e-06,
      "loss": 0.1854,
      "step": 2527
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3313935994406978,
      "learning_rate": 9.415462870860746e-06,
      "loss": 0.2461,
      "step": 2528
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6780353280629303,
      "learning_rate": 9.41491923043313e-06,
      "loss": 0.2574,
      "step": 2529
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.499101368110134,
      "learning_rate": 9.414375353030192e-06,
      "loss": 0.6757,
      "step": 2530
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4853714488403271,
      "learning_rate": 9.413831238681124e-06,
      "loss": 0.2355,
      "step": 2531
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.569123284488001,
      "learning_rate": 9.413286887415128e-06,
      "loss": 0.197,
      "step": 2532
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4169325458493869,
      "learning_rate": 9.412742299261426e-06,
      "loss": 0.251,
      "step": 2533
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.510651441098901,
      "learning_rate": 9.412197474249247e-06,
      "loss": 0.2637,
      "step": 2534
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2067362394542676,
      "learning_rate": 9.411652412407837e-06,
      "loss": 0.2162,
      "step": 2535
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2477273083006541,
      "learning_rate": 9.411107113766452e-06,
      "loss": 0.1857,
      "step": 2536
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4250817536231168,
      "learning_rate": 9.410561578354362e-06,
      "loss": 0.2028,
      "step": 2537
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1801288799728686,
      "learning_rate": 9.410015806200849e-06,
      "loss": 0.2395,
      "step": 2538
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4909444786860953,
      "learning_rate": 9.409469797335206e-06,
      "loss": 0.2295,
      "step": 2539
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.516145414170348,
      "learning_rate": 9.408923551786742e-06,
      "loss": 0.2365,
      "step": 2540
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4758053953734014,
      "learning_rate": 9.40837706958478e-06,
      "loss": 0.2026,
      "step": 2541
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4286178469646953,
      "learning_rate": 9.40783035075865e-06,
      "loss": 0.2184,
      "step": 2542
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4604527487138599,
      "learning_rate": 9.407283395337696e-06,
      "loss": 0.2179,
      "step": 2543
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7185156550509677,
      "learning_rate": 9.406736203351278e-06,
      "loss": 0.2346,
      "step": 2544
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.512872717059322,
      "learning_rate": 9.406188774828771e-06,
      "loss": 0.1901,
      "step": 2545
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4936408127657042,
      "learning_rate": 9.405641109799554e-06,
      "loss": 0.2307,
      "step": 2546
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.667225573679939,
      "learning_rate": 9.405093208293023e-06,
      "loss": 0.2691,
      "step": 2547
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2487556753117102,
      "learning_rate": 9.404545070338593e-06,
      "loss": 0.2199,
      "step": 2548
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5825009255579274,
      "learning_rate": 9.403996695965677e-06,
      "loss": 0.2049,
      "step": 2549
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5170348902208557,
      "learning_rate": 9.403448085203718e-06,
      "loss": 0.2035,
      "step": 2550
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5473668278518455,
      "learning_rate": 9.402899238082159e-06,
      "loss": 0.215,
      "step": 2551
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.275240360202992,
      "learning_rate": 9.402350154630462e-06,
      "loss": 0.1691,
      "step": 2552
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3418197112154044,
      "learning_rate": 9.401800834878096e-06,
      "loss": 0.2019,
      "step": 2553
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4542036000172531,
      "learning_rate": 9.401251278854551e-06,
      "loss": 0.2382,
      "step": 2554
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2648990973797447,
      "learning_rate": 9.40070148658932e-06,
      "loss": 0.2167,
      "step": 2555
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3670632272309051,
      "learning_rate": 9.400151458111918e-06,
      "loss": 0.2268,
      "step": 2556
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.337470764442655,
      "learning_rate": 9.399601193451866e-06,
      "loss": 0.4662,
      "step": 2557
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4892041476995646,
      "learning_rate": 9.3990506926387e-06,
      "loss": 0.2519,
      "step": 2558
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.119669690320454,
      "learning_rate": 9.39849995570197e-06,
      "loss": 0.6525,
      "step": 2559
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.465976325894581,
      "learning_rate": 9.397948982671237e-06,
      "loss": 0.27,
      "step": 2560
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.08568928988736,
      "learning_rate": 9.397397773576074e-06,
      "loss": 0.6543,
      "step": 2561
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5189782675403085,
      "learning_rate": 9.39684632844607e-06,
      "loss": 0.242,
      "step": 2562
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4071735096660711,
      "learning_rate": 9.396294647310822e-06,
      "loss": 0.19,
      "step": 2563
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4370055552296672,
      "learning_rate": 9.395742730199942e-06,
      "loss": 0.1654,
      "step": 2564
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.970414512865405,
      "learning_rate": 9.395190577143057e-06,
      "loss": 0.749,
      "step": 2565
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.484066099046267,
      "learning_rate": 9.394638188169803e-06,
      "loss": 0.2086,
      "step": 2566
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.486353336139996,
      "learning_rate": 9.394085563309827e-06,
      "loss": 0.2414,
      "step": 2567
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4810219803692295,
      "learning_rate": 9.393532702592798e-06,
      "loss": 0.2556,
      "step": 2568
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4140330510795132,
      "learning_rate": 9.392979606048386e-06,
      "loss": 0.2422,
      "step": 2569
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3740066127090602,
      "learning_rate": 9.392426273706284e-06,
      "loss": 0.2165,
      "step": 2570
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3848248012270683,
      "learning_rate": 9.391872705596186e-06,
      "loss": 0.2233,
      "step": 2571
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.238710975646007,
      "learning_rate": 9.391318901747812e-06,
      "loss": 0.1966,
      "step": 2572
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.467387934226765,
      "learning_rate": 9.390764862190885e-06,
      "loss": 0.2275,
      "step": 2573
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3381964919802931,
      "learning_rate": 9.390210586955143e-06,
      "loss": 0.2334,
      "step": 2574
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.150413931274772,
      "learning_rate": 9.389656076070337e-06,
      "loss": 0.3801,
      "step": 2575
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6358963938446924,
      "learning_rate": 9.389101329566234e-06,
      "loss": 0.238,
      "step": 2576
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3506830108362546,
      "learning_rate": 9.388546347472608e-06,
      "loss": 0.185,
      "step": 2577
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.395977514191067,
      "learning_rate": 9.387991129819249e-06,
      "loss": 0.2077,
      "step": 2578
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4421513975374125,
      "learning_rate": 9.387435676635958e-06,
      "loss": 0.2438,
      "step": 2579
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5283789095280744,
      "learning_rate": 9.386879987952549e-06,
      "loss": 0.271,
      "step": 2580
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5573160095369922,
      "learning_rate": 9.386324063798853e-06,
      "loss": 0.2826,
      "step": 2581
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6341628290442851,
      "learning_rate": 9.385767904204705e-06,
      "loss": 0.2357,
      "step": 2582
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2991602293304791,
      "learning_rate": 9.385211509199961e-06,
      "loss": 0.2001,
      "step": 2583
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3820289229756992,
      "learning_rate": 9.384654878814485e-06,
      "loss": 0.194,
      "step": 2584
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5704484318379797,
      "learning_rate": 9.384098013078153e-06,
      "loss": 0.2075,
      "step": 2585
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4569078715468275,
      "learning_rate": 9.38354091202086e-06,
      "loss": 0.2265,
      "step": 2586
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.366742029026146,
      "learning_rate": 9.382983575672502e-06,
      "loss": 0.2244,
      "step": 2587
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4285443750863607,
      "learning_rate": 9.382426004063e-06,
      "loss": 0.2129,
      "step": 2588
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4159106749452377,
      "learning_rate": 9.381868197222279e-06,
      "loss": 0.1942,
      "step": 2589
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.47367493253568,
      "learning_rate": 9.381310155180283e-06,
      "loss": 0.2815,
      "step": 2590
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5011937629642231,
      "learning_rate": 9.380751877966964e-06,
      "loss": 0.2164,
      "step": 2591
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.465458499746767,
      "learning_rate": 9.380193365612288e-06,
      "loss": 0.2406,
      "step": 2592
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.220150935569685,
      "learning_rate": 9.379634618146234e-06,
      "loss": 0.657,
      "step": 2593
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3745933997369417,
      "learning_rate": 9.379075635598792e-06,
      "loss": 0.1996,
      "step": 2594
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5565525687039836,
      "learning_rate": 9.378516417999967e-06,
      "loss": 0.2278,
      "step": 2595
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3656471769207637,
      "learning_rate": 9.377956965379776e-06,
      "loss": 0.21,
      "step": 2596
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.485108983419183,
      "learning_rate": 9.377397277768247e-06,
      "loss": 0.2394,
      "step": 2597
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5186262861840791,
      "learning_rate": 9.376837355195424e-06,
      "loss": 0.2379,
      "step": 2598
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3456319799277998,
      "learning_rate": 9.37627719769136e-06,
      "loss": 0.2136,
      "step": 2599
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4797305119689288,
      "learning_rate": 9.375716805286122e-06,
      "loss": 0.2223,
      "step": 2600
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3784933414834113,
      "learning_rate": 9.375156178009787e-06,
      "loss": 0.2112,
      "step": 2601
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.362812963726172,
      "learning_rate": 9.374595315892453e-06,
      "loss": 0.2467,
      "step": 2602
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5282408902874565,
      "learning_rate": 9.37403421896422e-06,
      "loss": 0.2369,
      "step": 2603
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4662727727613907,
      "learning_rate": 9.373472887255209e-06,
      "loss": 0.2563,
      "step": 2604
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.392805867502153,
      "learning_rate": 9.372911320795548e-06,
      "loss": 0.6633,
      "step": 2605
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.411979198973745,
      "learning_rate": 9.372349519615378e-06,
      "loss": 0.6402,
      "step": 2606
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3296598634771524,
      "learning_rate": 9.371787483744858e-06,
      "loss": 0.2245,
      "step": 2607
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2074160163078957,
      "learning_rate": 9.371225213214155e-06,
      "loss": 0.1935,
      "step": 2608
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2345187064089502,
      "learning_rate": 9.370662708053446e-06,
      "loss": 0.2026,
      "step": 2609
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5670469034123782,
      "learning_rate": 9.370099968292928e-06,
      "loss": 0.3105,
      "step": 2610
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2736884433859044,
      "learning_rate": 9.369536993962805e-06,
      "loss": 0.1929,
      "step": 2611
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.552111752667459,
      "learning_rate": 9.368973785093294e-06,
      "loss": 0.2218,
      "step": 2612
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2903047429914232,
      "learning_rate": 9.36841034171463e-06,
      "loss": 0.1725,
      "step": 2613
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.316133816347541,
      "learning_rate": 9.367846663857052e-06,
      "loss": 0.6253,
      "step": 2614
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5343751127083,
      "learning_rate": 9.367282751550818e-06,
      "loss": 0.2547,
      "step": 2615
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3955394327221744,
      "learning_rate": 9.366718604826196e-06,
      "loss": 0.2411,
      "step": 2616
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.310393300970099,
      "learning_rate": 9.366154223713468e-06,
      "loss": 0.1985,
      "step": 2617
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3973836522775696,
      "learning_rate": 9.365589608242925e-06,
      "loss": 0.2059,
      "step": 2618
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3238252565612803,
      "learning_rate": 9.365024758444876e-06,
      "loss": 0.211,
      "step": 2619
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.608485287603873,
      "learning_rate": 9.364459674349642e-06,
      "loss": 0.7084,
      "step": 2620
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.54372780776565,
      "learning_rate": 9.363894355987548e-06,
      "loss": 0.2312,
      "step": 2621
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4399688844193446,
      "learning_rate": 9.363328803388945e-06,
      "loss": 0.2341,
      "step": 2622
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.555308097709079,
      "learning_rate": 9.362763016584185e-06,
      "loss": 0.2128,
      "step": 2623
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6054150228149366,
      "learning_rate": 9.362196995603638e-06,
      "loss": 0.2671,
      "step": 2624
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.7441345999217495,
      "learning_rate": 9.361630740477687e-06,
      "loss": 0.748,
      "step": 2625
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3898886491048759,
      "learning_rate": 9.361064251236727e-06,
      "loss": 0.238,
      "step": 2626
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7681876860381698,
      "learning_rate": 9.360497527911161e-06,
      "loss": 0.2653,
      "step": 2627
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3830518890928463,
      "learning_rate": 9.359930570531412e-06,
      "loss": 0.1947,
      "step": 2628
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.721124030620209,
      "learning_rate": 9.359363379127913e-06,
      "loss": 0.6674,
      "step": 2629
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3715638671589498,
      "learning_rate": 9.358795953731105e-06,
      "loss": 0.2118,
      "step": 2630
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.300925773108132,
      "learning_rate": 9.358228294371447e-06,
      "loss": 0.574,
      "step": 2631
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.470608078411285,
      "learning_rate": 9.357660401079407e-06,
      "loss": 0.1929,
      "step": 2632
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2712053947491184,
      "learning_rate": 9.357092273885471e-06,
      "loss": 0.2207,
      "step": 2633
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.343966932686822,
      "learning_rate": 9.35652391282013e-06,
      "loss": 0.2021,
      "step": 2634
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4397501971606592,
      "learning_rate": 9.355955317913895e-06,
      "loss": 0.2335,
      "step": 2635
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3173720124582087,
      "learning_rate": 9.35538648919728e-06,
      "loss": 0.167,
      "step": 2636
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3250624739124315,
      "learning_rate": 9.354817426700823e-06,
      "loss": 0.1982,
      "step": 2637
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4240700305697667,
      "learning_rate": 9.35424813045507e-06,
      "loss": 0.2177,
      "step": 2638
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3009564031440501,
      "learning_rate": 9.353678600490574e-06,
      "loss": 0.2176,
      "step": 2639
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1809864114445163,
      "learning_rate": 9.353108836837907e-06,
      "loss": 0.1814,
      "step": 2640
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6284671722401147,
      "learning_rate": 9.35253883952765e-06,
      "loss": 0.2631,
      "step": 2641
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.584967221164042,
      "learning_rate": 9.351968608590403e-06,
      "loss": 0.2671,
      "step": 2642
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6393939829993323,
      "learning_rate": 9.351398144056766e-06,
      "loss": 0.1816,
      "step": 2643
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2956942014534252,
      "learning_rate": 9.350827445957369e-06,
      "loss": 0.1869,
      "step": 2644
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3632702962493306,
      "learning_rate": 9.350256514322836e-06,
      "loss": 0.2217,
      "step": 2645
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.605739316717559,
      "learning_rate": 9.349685349183817e-06,
      "loss": 0.5691,
      "step": 2646
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3461219548050736,
      "learning_rate": 9.349113950570969e-06,
      "loss": 0.1882,
      "step": 2647
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.41257908296118,
      "learning_rate": 9.348542318514963e-06,
      "loss": 0.2178,
      "step": 2648
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.683448129892471,
      "learning_rate": 9.34797045304648e-06,
      "loss": 0.7057,
      "step": 2649
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.540053669782913,
      "learning_rate": 9.347398354196218e-06,
      "loss": 0.1963,
      "step": 2650
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3848077156393017,
      "learning_rate": 9.346826021994882e-06,
      "loss": 0.1656,
      "step": 2651
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7088747074613753,
      "learning_rate": 9.346253456473196e-06,
      "loss": 0.1953,
      "step": 2652
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.356861984625284,
      "learning_rate": 9.345680657661889e-06,
      "loss": 0.2016,
      "step": 2653
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3702899839318514,
      "learning_rate": 9.34510762559171e-06,
      "loss": 0.1937,
      "step": 2654
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5373665955594087,
      "learning_rate": 9.344534360293416e-06,
      "loss": 0.2262,
      "step": 2655
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3508817893590057,
      "learning_rate": 9.343960861797776e-06,
      "loss": 0.2074,
      "step": 2656
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.668990845109574,
      "learning_rate": 9.343387130135578e-06,
      "loss": 0.2209,
      "step": 2657
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2624700293295148,
      "learning_rate": 9.34281316533761e-06,
      "loss": 0.1696,
      "step": 2658
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.371125337285241,
      "learning_rate": 9.342238967434687e-06,
      "loss": 0.2321,
      "step": 2659
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.3154912091480195,
      "learning_rate": 9.341664536457626e-06,
      "loss": 0.6463,
      "step": 2660
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.228365937805923,
      "learning_rate": 9.34108987243726e-06,
      "loss": 0.6574,
      "step": 2661
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5222433283254058,
      "learning_rate": 9.34051497540444e-06,
      "loss": 0.2486,
      "step": 2662
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.186250406634907,
      "learning_rate": 9.339939845390017e-06,
      "loss": 0.1595,
      "step": 2663
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2586567404213749,
      "learning_rate": 9.339364482424865e-06,
      "loss": 0.1787,
      "step": 2664
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4708089042739207,
      "learning_rate": 9.338788886539867e-06,
      "loss": 0.2319,
      "step": 2665
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4473028547669557,
      "learning_rate": 9.33821305776592e-06,
      "loss": 0.2114,
      "step": 2666
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.008082694813588,
      "learning_rate": 9.337636996133928e-06,
      "loss": 0.6501,
      "step": 2667
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3928793156903505,
      "learning_rate": 9.337060701674817e-06,
      "loss": 0.1992,
      "step": 2668
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.127217601180208,
      "learning_rate": 9.336484174419517e-06,
      "loss": 0.7187,
      "step": 2669
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3621883220535445,
      "learning_rate": 9.335907414398974e-06,
      "loss": 0.2167,
      "step": 2670
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5112950897491666,
      "learning_rate": 9.335330421644149e-06,
      "loss": 0.268,
      "step": 2671
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.6009733635284285,
      "learning_rate": 9.334753196186008e-06,
      "loss": 0.6594,
      "step": 2672
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4341769174052572,
      "learning_rate": 9.334175738055536e-06,
      "loss": 0.2215,
      "step": 2673
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.537329764551531,
      "learning_rate": 9.333598047283732e-06,
      "loss": 0.2354,
      "step": 2674
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4103874617643597,
      "learning_rate": 9.333020123901598e-06,
      "loss": 0.2131,
      "step": 2675
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4867447840007968,
      "learning_rate": 9.332441967940161e-06,
      "loss": 0.2322,
      "step": 2676
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.422852635270182,
      "learning_rate": 9.331863579430448e-06,
      "loss": 0.2232,
      "step": 2677
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5005484131513953,
      "learning_rate": 9.331284958403508e-06,
      "loss": 0.2768,
      "step": 2678
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6300623814846578,
      "learning_rate": 9.3307061048904e-06,
      "loss": 0.2521,
      "step": 2679
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3535516016391143,
      "learning_rate": 9.330127018922195e-06,
      "loss": 0.2567,
      "step": 2680
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.134663939258145,
      "learning_rate": 9.32954770052997e-06,
      "loss": 0.4074,
      "step": 2681
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4924927869139148,
      "learning_rate": 9.328968149744828e-06,
      "loss": 0.2501,
      "step": 2682
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6153668058356008,
      "learning_rate": 9.328388366597873e-06,
      "loss": 0.2364,
      "step": 2683
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.247401935794908,
      "learning_rate": 9.327808351120226e-06,
      "loss": 0.1843,
      "step": 2684
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2616115513981114,
      "learning_rate": 9.327228103343021e-06,
      "loss": 0.2139,
      "step": 2685
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5528563534981583,
      "learning_rate": 9.326647623297402e-06,
      "loss": 0.1905,
      "step": 2686
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3720780262954007,
      "learning_rate": 9.326066911014527e-06,
      "loss": 0.2023,
      "step": 2687
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3948592966233728,
      "learning_rate": 9.325485966525567e-06,
      "loss": 0.2033,
      "step": 2688
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4118822665499795,
      "learning_rate": 9.324904789861707e-06,
      "loss": 0.1953,
      "step": 2689
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.457205728546168,
      "learning_rate": 9.324323381054137e-06,
      "loss": 0.2092,
      "step": 2690
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.521618661494352,
      "learning_rate": 9.323741740134066e-06,
      "loss": 0.2268,
      "step": 2691
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5302387460297349,
      "learning_rate": 9.32315986713272e-06,
      "loss": 0.2808,
      "step": 2692
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4056407809955174,
      "learning_rate": 9.322577762081324e-06,
      "loss": 0.2364,
      "step": 2693
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5812229698886506,
      "learning_rate": 9.321995425011127e-06,
      "loss": 0.2411,
      "step": 2694
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4778914911432797,
      "learning_rate": 9.321412855953386e-06,
      "loss": 0.2259,
      "step": 2695
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.5953610546342505,
      "learning_rate": 9.320830054939373e-06,
      "loss": 0.4609,
      "step": 2696
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3463001817745606,
      "learning_rate": 9.320247022000366e-06,
      "loss": 0.1609,
      "step": 2697
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4620690972729318,
      "learning_rate": 9.319663757167663e-06,
      "loss": 0.2232,
      "step": 2698
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.571207864020747,
      "learning_rate": 9.319080260472571e-06,
      "loss": 0.2458,
      "step": 2699
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.594287483968256,
      "learning_rate": 9.318496531946411e-06,
      "loss": 0.5159,
      "step": 2700
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5639165793111875,
      "learning_rate": 9.31791257162051e-06,
      "loss": 0.2586,
      "step": 2701
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6678175206244428,
      "learning_rate": 9.317328379526219e-06,
      "loss": 0.2199,
      "step": 2702
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.411568760918074,
      "learning_rate": 9.316743955694892e-06,
      "loss": 0.1969,
      "step": 2703
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6006313297847388,
      "learning_rate": 9.316159300157899e-06,
      "loss": 0.2492,
      "step": 2704
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.439379058915103,
      "learning_rate": 9.31557441294662e-06,
      "loss": 0.2225,
      "step": 2705
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.375470478177647,
      "learning_rate": 9.314989294092453e-06,
      "loss": 0.2047,
      "step": 2706
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.432408221958686,
      "learning_rate": 9.314403943626804e-06,
      "loss": 0.246,
      "step": 2707
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.424167195985073,
      "learning_rate": 9.31381836158109e-06,
      "loss": 0.1968,
      "step": 2708
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5602232022364202,
      "learning_rate": 9.313232547986744e-06,
      "loss": 0.2459,
      "step": 2709
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.430354091928059,
      "learning_rate": 9.312646502875213e-06,
      "loss": 0.7736,
      "step": 2710
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5224049519185867,
      "learning_rate": 9.312060226277948e-06,
      "loss": 0.2555,
      "step": 2711
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5217102315763986,
      "learning_rate": 9.311473718226422e-06,
      "loss": 0.2181,
      "step": 2712
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3757371671563503,
      "learning_rate": 9.310886978752116e-06,
      "loss": 0.247,
      "step": 2713
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4554764511425178,
      "learning_rate": 9.310300007886523e-06,
      "loss": 0.2572,
      "step": 2714
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4223498539004633,
      "learning_rate": 9.30971280566115e-06,
      "loss": 0.257,
      "step": 2715
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4603009274575902,
      "learning_rate": 9.309125372107514e-06,
      "loss": 0.198,
      "step": 2716
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3885932419394922,
      "learning_rate": 9.308537707257148e-06,
      "loss": 0.2471,
      "step": 2717
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3665536893236059,
      "learning_rate": 9.307949811141595e-06,
      "loss": 0.1904,
      "step": 2718
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7724286761669077,
      "learning_rate": 9.30736168379241e-06,
      "loss": 0.2754,
      "step": 2719
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.468305458607527,
      "learning_rate": 9.306773325241161e-06,
      "loss": 0.4609,
      "step": 2720
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5025273147906886,
      "learning_rate": 9.306184735519432e-06,
      "loss": 0.2285,
      "step": 2721
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.866954256090008,
      "learning_rate": 9.305595914658813e-06,
      "loss": 0.5086,
      "step": 2722
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4961517640542386,
      "learning_rate": 9.30500686269091e-06,
      "loss": 0.2187,
      "step": 2723
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4009562373951943,
      "learning_rate": 9.304417579647343e-06,
      "loss": 0.2207,
      "step": 2724
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.52417488568084,
      "learning_rate": 9.303828065559739e-06,
      "loss": 0.2299,
      "step": 2725
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3922647711060259,
      "learning_rate": 9.303238320459743e-06,
      "loss": 0.2243,
      "step": 2726
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4158905057994566,
      "learning_rate": 9.302648344379011e-06,
      "loss": 0.2448,
      "step": 2727
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4007514263350775,
      "learning_rate": 9.302058137349208e-06,
      "loss": 0.224,
      "step": 2728
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4415310237479846,
      "learning_rate": 9.301467699402018e-06,
      "loss": 0.1995,
      "step": 2729
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.547873263596923,
      "learning_rate": 9.300877030569129e-06,
      "loss": 0.2651,
      "step": 2730
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3112016602667436,
      "learning_rate": 9.300286130882247e-06,
      "loss": 0.2253,
      "step": 2731
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3200207159178472,
      "learning_rate": 9.29969500037309e-06,
      "loss": 0.1865,
      "step": 2732
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.484029897637912,
      "learning_rate": 9.299103639073388e-06,
      "loss": 0.1804,
      "step": 2733
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3514755449299012,
      "learning_rate": 9.298512047014881e-06,
      "loss": 0.2162,
      "step": 2734
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3133749155000263,
      "learning_rate": 9.297920224229324e-06,
      "loss": 0.1811,
      "step": 2735
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6680523931386122,
      "learning_rate": 9.297328170748484e-06,
      "loss": 0.2187,
      "step": 2736
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3641498118584952,
      "learning_rate": 9.29673588660414e-06,
      "loss": 0.1825,
      "step": 2737
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3983912375089784,
      "learning_rate": 9.296143371828085e-06,
      "loss": 0.1868,
      "step": 2738
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4709086835658272,
      "learning_rate": 9.295550626452121e-06,
      "loss": 0.2417,
      "step": 2739
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4447720380731979,
      "learning_rate": 9.294957650508065e-06,
      "loss": 0.2407,
      "step": 2740
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.351024778018581,
      "learning_rate": 9.294364444027746e-06,
      "loss": 0.2056,
      "step": 2741
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4556672287236467,
      "learning_rate": 9.293771007043002e-06,
      "loss": 0.2441,
      "step": 2742
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4059252715979254,
      "learning_rate": 9.293177339585691e-06,
      "loss": 0.2102,
      "step": 2743
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.393370760501809,
      "learning_rate": 9.292583441687675e-06,
      "loss": 0.211,
      "step": 2744
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3850214815757624,
      "learning_rate": 9.291989313380833e-06,
      "loss": 0.2089,
      "step": 2745
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5452408158663233,
      "learning_rate": 9.291394954697058e-06,
      "loss": 0.2431,
      "step": 2746
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.54379478589557,
      "learning_rate": 9.290800365668246e-06,
      "loss": 0.2444,
      "step": 2747
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4235216445906556,
      "learning_rate": 9.29020554632632e-06,
      "loss": 0.2208,
      "step": 2748
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5615086986420386,
      "learning_rate": 9.289610496703206e-06,
      "loss": 0.1903,
      "step": 2749
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.980866950673275,
      "learning_rate": 9.289015216830839e-06,
      "loss": 0.4589,
      "step": 2750
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4709741365835498,
      "learning_rate": 9.288419706741177e-06,
      "loss": 0.2732,
      "step": 2751
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4944737467566327,
      "learning_rate": 9.28782396646618e-06,
      "loss": 0.247,
      "step": 2752
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2533161569754208,
      "learning_rate": 9.287227996037829e-06,
      "loss": 0.2166,
      "step": 2753
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5398154363558294,
      "learning_rate": 9.28663179548811e-06,
      "loss": 0.2799,
      "step": 2754
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.366561861880614,
      "learning_rate": 9.286035364849027e-06,
      "loss": 0.2144,
      "step": 2755
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.302162884394884,
      "learning_rate": 9.285438704152594e-06,
      "loss": 0.1653,
      "step": 2756
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3622804246308315,
      "learning_rate": 9.284841813430834e-06,
      "loss": 0.212,
      "step": 2757
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.824199291912395,
      "learning_rate": 9.28424469271579e-06,
      "loss": 0.6597,
      "step": 2758
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.8210438744484754,
      "learning_rate": 9.28364734203951e-06,
      "loss": 0.6027,
      "step": 2759
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3327106212448299,
      "learning_rate": 9.283049761434059e-06,
      "loss": 0.2064,
      "step": 2760
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.564747013139255,
      "learning_rate": 9.282451950931515e-06,
      "loss": 0.2123,
      "step": 2761
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4655292731330085,
      "learning_rate": 9.28185391056396e-06,
      "loss": 0.2672,
      "step": 2762
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.734441567397226,
      "learning_rate": 9.2812556403635e-06,
      "loss": 0.6468,
      "step": 2763
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.035917829056862,
      "learning_rate": 9.280657140362246e-06,
      "loss": 0.7761,
      "step": 2764
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3411146166064634,
      "learning_rate": 9.28005841059232e-06,
      "loss": 0.1654,
      "step": 2765
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.130826316965804,
      "learning_rate": 9.279459451085866e-06,
      "loss": 0.1416,
      "step": 2766
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4208572742343444,
      "learning_rate": 9.278860261875029e-06,
      "loss": 0.252,
      "step": 2767
    },
    {
      "epoch": 0.2,
      "grad_norm": 13.110058786456102,
      "learning_rate": 9.278260842991971e-06,
      "loss": 0.5862,
      "step": 2768
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.109746481413057,
      "learning_rate": 9.277661194468867e-06,
      "loss": 0.5593,
      "step": 2769
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.267914966286527,
      "learning_rate": 9.277061316337908e-06,
      "loss": 0.2155,
      "step": 2770
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5116837340734872,
      "learning_rate": 9.276461208631286e-06,
      "loss": 0.2528,
      "step": 2771
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3129335946906224,
      "learning_rate": 9.275860871381217e-06,
      "loss": 0.1985,
      "step": 2772
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7569515488462752,
      "learning_rate": 9.275260304619922e-06,
      "loss": 0.2886,
      "step": 2773
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5389205055054571,
      "learning_rate": 9.27465950837964e-06,
      "loss": 0.2095,
      "step": 2774
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3729770457397474,
      "learning_rate": 9.274058482692618e-06,
      "loss": 0.1806,
      "step": 2775
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.457133959654698,
      "learning_rate": 9.273457227591117e-06,
      "loss": 0.1955,
      "step": 2776
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3837547158127559,
      "learning_rate": 9.272855743107408e-06,
      "loss": 0.2296,
      "step": 2777
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4770409181636541,
      "learning_rate": 9.272254029273779e-06,
      "loss": 0.246,
      "step": 2778
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.50953406094558,
      "learning_rate": 9.271652086122527e-06,
      "loss": 0.1948,
      "step": 2779
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4094134858899505,
      "learning_rate": 9.27104991368596e-06,
      "loss": 0.206,
      "step": 2780
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.743157530124243,
      "learning_rate": 9.270447511996402e-06,
      "loss": 0.5884,
      "step": 2781
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4585970236953467,
      "learning_rate": 9.269844881086187e-06,
      "loss": 0.2232,
      "step": 2782
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2420764128207955,
      "learning_rate": 9.269242020987663e-06,
      "loss": 0.1763,
      "step": 2783
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6389088435882875,
      "learning_rate": 9.268638931733188e-06,
      "loss": 0.2664,
      "step": 2784
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4173813235755677,
      "learning_rate": 9.268035613355133e-06,
      "loss": 0.2072,
      "step": 2785
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5163241281216688,
      "learning_rate": 9.267432065885881e-06,
      "loss": 0.2257,
      "step": 2786
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.913317759367872,
      "learning_rate": 9.266828289357831e-06,
      "loss": 0.6326,
      "step": 2787
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3921344945068963,
      "learning_rate": 9.266224283803388e-06,
      "loss": 0.2147,
      "step": 2788
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3765929934884522,
      "learning_rate": 9.265620049254977e-06,
      "loss": 0.1904,
      "step": 2789
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4151804306094544,
      "learning_rate": 9.265015585745028e-06,
      "loss": 0.1913,
      "step": 2790
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.730188153764662,
      "learning_rate": 9.264410893305984e-06,
      "loss": 0.4471,
      "step": 2791
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4275896451456935,
      "learning_rate": 9.263805971970305e-06,
      "loss": 0.2021,
      "step": 2792
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.400286947194806,
      "learning_rate": 9.263200821770462e-06,
      "loss": 0.5824,
      "step": 2793
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.525565134603284,
      "learning_rate": 9.262595442738936e-06,
      "loss": 0.6484,
      "step": 2794
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4376229811971102,
      "learning_rate": 9.26198983490822e-06,
      "loss": 0.2188,
      "step": 2795
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5562788641439695,
      "learning_rate": 9.261383998310822e-06,
      "loss": 0.2453,
      "step": 2796
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4095990593929484,
      "learning_rate": 9.26077793297926e-06,
      "loss": 0.2341,
      "step": 2797
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3918150164673484,
      "learning_rate": 9.260171638946065e-06,
      "loss": 0.207,
      "step": 2798
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.383177749722103,
      "learning_rate": 9.259565116243782e-06,
      "loss": 0.2418,
      "step": 2799
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5421088406739354,
      "learning_rate": 9.258958364904966e-06,
      "loss": 0.2294,
      "step": 2800
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4101165407802647,
      "learning_rate": 9.258351384962187e-06,
      "loss": 0.1934,
      "step": 2801
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.808644490871377,
      "learning_rate": 9.25774417644802e-06,
      "loss": 0.5176,
      "step": 2802
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2975173269739633,
      "learning_rate": 9.257136739395063e-06,
      "loss": 0.1771,
      "step": 2803
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.301827000773767,
      "learning_rate": 9.256529073835917e-06,
      "loss": 0.2105,
      "step": 2804
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5622858386793308,
      "learning_rate": 9.255921179803202e-06,
      "loss": 0.241,
      "step": 2805
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4943515687214168,
      "learning_rate": 9.255313057329544e-06,
      "loss": 0.2104,
      "step": 2806
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4273896313848267,
      "learning_rate": 9.254704706447588e-06,
      "loss": 0.2469,
      "step": 2807
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3350369604624892,
      "learning_rate": 9.254096127189987e-06,
      "loss": 0.1907,
      "step": 2808
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.22600804276605,
      "learning_rate": 9.253487319589405e-06,
      "loss": 0.6303,
      "step": 2809
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.199166773009833,
      "learning_rate": 9.252878283678524e-06,
      "loss": 0.1561,
      "step": 2810
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4541966938026452,
      "learning_rate": 9.25226901949003e-06,
      "loss": 0.2152,
      "step": 2811
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.418355352415606,
      "learning_rate": 9.251659527056629e-06,
      "loss": 0.2227,
      "step": 2812
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3098082635392974,
      "learning_rate": 9.251049806411037e-06,
      "loss": 0.2175,
      "step": 2813
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4981322906904728,
      "learning_rate": 9.250439857585977e-06,
      "loss": 0.2571,
      "step": 2814
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4681854896439621,
      "learning_rate": 9.249829680614195e-06,
      "loss": 0.2471,
      "step": 2815
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3584326514908769,
      "learning_rate": 9.249219275528438e-06,
      "loss": 0.1892,
      "step": 2816
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.43939624957332,
      "learning_rate": 9.248608642361471e-06,
      "loss": 0.2305,
      "step": 2817
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6473957806283077,
      "learning_rate": 9.247997781146071e-06,
      "loss": 0.2225,
      "step": 2818
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3492870624845066,
      "learning_rate": 9.247386691915027e-06,
      "loss": 0.168,
      "step": 2819
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4929351976082126,
      "learning_rate": 9.246775374701139e-06,
      "loss": 0.2252,
      "step": 2820
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.415205746908608,
      "learning_rate": 9.246163829537221e-06,
      "loss": 0.2468,
      "step": 2821
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5294410991062586,
      "learning_rate": 9.245552056456098e-06,
      "loss": 0.22,
      "step": 2822
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5060850201811555,
      "learning_rate": 9.244940055490605e-06,
      "loss": 0.2207,
      "step": 2823
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5442673706591972,
      "learning_rate": 9.244327826673597e-06,
      "loss": 0.2992,
      "step": 2824
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3735292757117905,
      "learning_rate": 9.24371537003793e-06,
      "loss": 0.2383,
      "step": 2825
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4279895383303036,
      "learning_rate": 9.243102685616486e-06,
      "loss": 0.1851,
      "step": 2826
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3736542694389804,
      "learning_rate": 9.242489773442142e-06,
      "loss": 0.1692,
      "step": 2827
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4050323736809658,
      "learning_rate": 9.241876633547803e-06,
      "loss": 0.2119,
      "step": 2828
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6507496021649768,
      "learning_rate": 9.24126326596638e-06,
      "loss": 0.2093,
      "step": 2829
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3120092849258171,
      "learning_rate": 9.240649670730793e-06,
      "loss": 0.1951,
      "step": 2830
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.484924776215179,
      "learning_rate": 9.240035847873979e-06,
      "loss": 0.7142,
      "step": 2831
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.982837490805566,
      "learning_rate": 9.239421797428884e-06,
      "loss": 0.6435,
      "step": 2832
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5965601464560937,
      "learning_rate": 9.238807519428472e-06,
      "loss": 0.2262,
      "step": 2833
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.295069840397293,
      "learning_rate": 9.238193013905713e-06,
      "loss": 0.6014,
      "step": 2834
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4283721998912922,
      "learning_rate": 9.237578280893587e-06,
      "loss": 0.2141,
      "step": 2835
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.526082744237695,
      "learning_rate": 9.236963320425097e-06,
      "loss": 0.2357,
      "step": 2836
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.189494667861111,
      "learning_rate": 9.236348132533246e-06,
      "loss": 0.5008,
      "step": 2837
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4055438900908386,
      "learning_rate": 9.235732717251058e-06,
      "loss": 0.2041,
      "step": 2838
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.460036773370276,
      "learning_rate": 9.235117074611565e-06,
      "loss": 0.1947,
      "step": 2839
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4335338202480654,
      "learning_rate": 9.234501204647814e-06,
      "loss": 0.2658,
      "step": 2840
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4684474814845387,
      "learning_rate": 9.233885107392862e-06,
      "loss": 0.2263,
      "step": 2841
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.151358695531919,
      "learning_rate": 9.233268782879776e-06,
      "loss": 0.6527,
      "step": 2842
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4589339251718731,
      "learning_rate": 9.232652231141638e-06,
      "loss": 0.1888,
      "step": 2843
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5247053127117112,
      "learning_rate": 9.232035452211546e-06,
      "loss": 0.2671,
      "step": 2844
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5587615120413763,
      "learning_rate": 9.231418446122604e-06,
      "loss": 0.2018,
      "step": 2845
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2545277438154867,
      "learning_rate": 9.230801212907929e-06,
      "loss": 0.2125,
      "step": 2846
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.124026859415719,
      "learning_rate": 9.230183752600654e-06,
      "loss": 0.5483,
      "step": 2847
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3049019603588219,
      "learning_rate": 9.229566065233921e-06,
      "loss": 0.1826,
      "step": 2848
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4311304964041343,
      "learning_rate": 9.228948150840885e-06,
      "loss": 0.2233,
      "step": 2849
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4595755941107131,
      "learning_rate": 9.228330009454712e-06,
      "loss": 0.2417,
      "step": 2850
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3195441124832552,
      "learning_rate": 9.227711641108584e-06,
      "loss": 0.1847,
      "step": 2851
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2680987731607882,
      "learning_rate": 9.22709304583569e-06,
      "loss": 0.1779,
      "step": 2852
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.406926919757446,
      "learning_rate": 9.226474223669232e-06,
      "loss": 0.2046,
      "step": 2853
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4824867011955372,
      "learning_rate": 9.225855174642432e-06,
      "loss": 0.2482,
      "step": 2854
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3195907163082892,
      "learning_rate": 9.225235898788514e-06,
      "loss": 0.1813,
      "step": 2855
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.423444453793779,
      "learning_rate": 9.224616396140718e-06,
      "loss": 0.2536,
      "step": 2856
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.490820136277582,
      "learning_rate": 9.223996666732297e-06,
      "loss": 0.225,
      "step": 2857
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3734782446205307,
      "learning_rate": 9.223376710596517e-06,
      "loss": 0.1677,
      "step": 2858
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.320047194461662,
      "learning_rate": 9.222756527766655e-06,
      "loss": 0.2177,
      "step": 2859
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.724840604611111,
      "learning_rate": 9.222136118275996e-06,
      "loss": 0.565,
      "step": 2860
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5647166395838488,
      "learning_rate": 9.221515482157846e-06,
      "loss": 0.2416,
      "step": 2861
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5138161696762844,
      "learning_rate": 9.220894619445515e-06,
      "loss": 0.219,
      "step": 2862
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6641828045243603,
      "learning_rate": 9.22027353017233e-06,
      "loss": 0.2114,
      "step": 2863
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5068638410549247,
      "learning_rate": 9.219652214371628e-06,
      "loss": 0.2412,
      "step": 2864
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4088980677678526,
      "learning_rate": 9.219030672076759e-06,
      "loss": 0.1803,
      "step": 2865
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.445658100962736,
      "learning_rate": 9.218408903321083e-06,
      "loss": 0.2682,
      "step": 2866
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2967510704170178,
      "learning_rate": 9.217786908137977e-06,
      "loss": 0.2097,
      "step": 2867
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.2703517903031045,
      "learning_rate": 9.21716468656083e-06,
      "loss": 0.6153,
      "step": 2868
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4876279649956823,
      "learning_rate": 9.216542238623032e-06,
      "loss": 0.1947,
      "step": 2869
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4269034370104228,
      "learning_rate": 9.215919564358e-06,
      "loss": 0.218,
      "step": 2870
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.305710200853404,
      "learning_rate": 9.215296663799154e-06,
      "loss": 0.2031,
      "step": 2871
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.739863981747492,
      "learning_rate": 9.21467353697993e-06,
      "loss": 0.5491,
      "step": 2872
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5706629578445788,
      "learning_rate": 9.214050183933774e-06,
      "loss": 0.2676,
      "step": 2873
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.754683199847275,
      "learning_rate": 9.213426604694144e-06,
      "loss": 0.6966,
      "step": 2874
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.487998160727917,
      "learning_rate": 9.212802799294516e-06,
      "loss": 0.2161,
      "step": 2875
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.294755276412573,
      "learning_rate": 9.21217876776837e-06,
      "loss": 0.2157,
      "step": 2876
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4278918743110487,
      "learning_rate": 9.2115545101492e-06,
      "loss": 0.2308,
      "step": 2877
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5712930003039205,
      "learning_rate": 9.210930026470515e-06,
      "loss": 0.2741,
      "step": 2878
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1750212499757744,
      "learning_rate": 9.210305316765837e-06,
      "loss": 0.1516,
      "step": 2879
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5418971011845954,
      "learning_rate": 9.209680381068698e-06,
      "loss": 0.2803,
      "step": 2880
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5855413746728415,
      "learning_rate": 9.209055219412636e-06,
      "loss": 0.2345,
      "step": 2881
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4665870339453904,
      "learning_rate": 9.208429831831216e-06,
      "loss": 0.2554,
      "step": 2882
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.596048384415915,
      "learning_rate": 9.207804218357998e-06,
      "loss": 0.5125,
      "step": 2883
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3744818120320499,
      "learning_rate": 9.207178379026568e-06,
      "loss": 0.2148,
      "step": 2884
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.391466740522246,
      "learning_rate": 9.206552313870518e-06,
      "loss": 0.2165,
      "step": 2885
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2245404745787873,
      "learning_rate": 9.20592602292345e-06,
      "loss": 0.2041,
      "step": 2886
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3303813780140332,
      "learning_rate": 9.205299506218984e-06,
      "loss": 0.1737,
      "step": 2887
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.491862920075917,
      "learning_rate": 9.204672763790746e-06,
      "loss": 0.8073,
      "step": 2888
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.324213034678787,
      "learning_rate": 9.204045795672378e-06,
      "loss": 0.2126,
      "step": 2889
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4647247173433193,
      "learning_rate": 9.203418601897533e-06,
      "loss": 0.2243,
      "step": 2890
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4114448082721172,
      "learning_rate": 9.202791182499877e-06,
      "loss": 0.2025,
      "step": 2891
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4730769243053359,
      "learning_rate": 9.202163537513088e-06,
      "loss": 0.2568,
      "step": 2892
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.510226858794718,
      "learning_rate": 9.201535666970853e-06,
      "loss": 0.2336,
      "step": 2893
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6217562039263327,
      "learning_rate": 9.200907570906879e-06,
      "loss": 0.2609,
      "step": 2894
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.187671247670847,
      "learning_rate": 9.200279249354871e-06,
      "loss": 0.5835,
      "step": 2895
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.468417463546273,
      "learning_rate": 9.199650702348564e-06,
      "loss": 0.1782,
      "step": 2896
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.494521617811187,
      "learning_rate": 9.19902192992169e-06,
      "loss": 0.2289,
      "step": 2897
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3300920353043386,
      "learning_rate": 9.198392932108e-06,
      "loss": 0.2353,
      "step": 2898
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3033020956747912,
      "learning_rate": 9.197763708941256e-06,
      "loss": 0.2116,
      "step": 2899
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2522625146157715,
      "learning_rate": 9.197134260455233e-06,
      "loss": 0.1925,
      "step": 2900
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3614017231829951,
      "learning_rate": 9.196504586683719e-06,
      "loss": 0.249,
      "step": 2901
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6164602045948744,
      "learning_rate": 9.195874687660512e-06,
      "loss": 0.2349,
      "step": 2902
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3772405635756395,
      "learning_rate": 9.195244563419418e-06,
      "loss": 0.2351,
      "step": 2903
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.205161490884727,
      "learning_rate": 9.194614213994263e-06,
      "loss": 0.1672,
      "step": 2904
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3001416974410283,
      "learning_rate": 9.193983639418882e-06,
      "loss": 0.2357,
      "step": 2905
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.8481986565466895,
      "learning_rate": 9.193352839727122e-06,
      "loss": 0.5973,
      "step": 2906
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.347811544501337,
      "learning_rate": 9.192721814952838e-06,
      "loss": 0.2281,
      "step": 2907
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6103013379328681,
      "learning_rate": 9.192090565129907e-06,
      "loss": 0.2011,
      "step": 2908
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5320720317510643,
      "learning_rate": 9.191459090292207e-06,
      "loss": 0.265,
      "step": 2909
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.524764620528247,
      "learning_rate": 9.190827390473636e-06,
      "loss": 0.3148,
      "step": 2910
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4294818226140642,
      "learning_rate": 9.190195465708098e-06,
      "loss": 0.2371,
      "step": 2911
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.305243088148744,
      "learning_rate": 9.189563316029515e-06,
      "loss": 0.1709,
      "step": 2912
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3934715579723185,
      "learning_rate": 9.188930941471818e-06,
      "loss": 0.1621,
      "step": 2913
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2784602542474226,
      "learning_rate": 9.18829834206895e-06,
      "loss": 0.2014,
      "step": 2914
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.449436754164341,
      "learning_rate": 9.187665517854866e-06,
      "loss": 0.227,
      "step": 2915
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.396498785691227,
      "learning_rate": 9.187032468863532e-06,
      "loss": 0.2509,
      "step": 2916
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4228358880798186,
      "learning_rate": 9.186399195128932e-06,
      "loss": 0.2181,
      "step": 2917
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4272453857546465,
      "learning_rate": 9.185765696685053e-06,
      "loss": 0.2256,
      "step": 2918
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4407259463789588,
      "learning_rate": 9.185131973565901e-06,
      "loss": 0.1922,
      "step": 2919
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4248450625479225,
      "learning_rate": 9.184498025805493e-06,
      "loss": 0.2231,
      "step": 2920
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.497140119027999,
      "learning_rate": 9.183863853437854e-06,
      "loss": 0.179,
      "step": 2921
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8461566417740474,
      "learning_rate": 9.183229456497024e-06,
      "loss": 0.219,
      "step": 2922
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.367967088702171,
      "learning_rate": 9.182594835017056e-06,
      "loss": 0.2146,
      "step": 2923
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4650821723213976,
      "learning_rate": 9.181959989032014e-06,
      "loss": 0.2203,
      "step": 2924
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2576111863158799,
      "learning_rate": 9.181324918575974e-06,
      "loss": 0.1856,
      "step": 2925
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6514945340486962,
      "learning_rate": 9.180689623683024e-06,
      "loss": 0.2462,
      "step": 2926
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4877699404389335,
      "learning_rate": 9.180054104387264e-06,
      "loss": 0.1958,
      "step": 2927
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3572596692971515,
      "learning_rate": 9.179418360722807e-06,
      "loss": 0.188,
      "step": 2928
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4844144616604722,
      "learning_rate": 9.178782392723775e-06,
      "loss": 0.1828,
      "step": 2929
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.464110868713132,
      "learning_rate": 9.178146200424306e-06,
      "loss": 0.2248,
      "step": 2930
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.382279362877647,
      "learning_rate": 9.177509783858548e-06,
      "loss": 0.1875,
      "step": 2931
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3265848300436374,
      "learning_rate": 9.176873143060661e-06,
      "loss": 0.1784,
      "step": 2932
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.560245132429068,
      "learning_rate": 9.176236278064818e-06,
      "loss": 0.2118,
      "step": 2933
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.480926863352633,
      "learning_rate": 9.175599188905203e-06,
      "loss": 0.2001,
      "step": 2934
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5788512760334066,
      "learning_rate": 9.17496187561601e-06,
      "loss": 0.2392,
      "step": 2935
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.603343800948272,
      "learning_rate": 9.174324338231452e-06,
      "loss": 0.2286,
      "step": 2936
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4771134268020356,
      "learning_rate": 9.173686576785747e-06,
      "loss": 0.2529,
      "step": 2937
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.51347460813198,
      "learning_rate": 9.173048591313128e-06,
      "loss": 0.235,
      "step": 2938
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.405034339120005,
      "learning_rate": 9.172410381847838e-06,
      "loss": 0.2068,
      "step": 2939
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.930089226392536,
      "learning_rate": 9.171771948424138e-06,
      "loss": 0.5473,
      "step": 2940
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3219373958911942,
      "learning_rate": 9.171133291076289e-06,
      "loss": 0.2156,
      "step": 2941
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5606383088826374,
      "learning_rate": 9.170494409838579e-06,
      "loss": 0.2023,
      "step": 2942
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3482254729619318,
      "learning_rate": 9.169855304745297e-06,
      "loss": 0.1606,
      "step": 2943
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4099946923179147,
      "learning_rate": 9.169215975830747e-06,
      "loss": 0.2064,
      "step": 2944
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4014200171900193,
      "learning_rate": 9.168576423129247e-06,
      "loss": 0.2339,
      "step": 2945
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5054583925355143,
      "learning_rate": 9.167936646675126e-06,
      "loss": 0.2391,
      "step": 2946
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3604018423659765,
      "learning_rate": 9.167296646502725e-06,
      "loss": 0.2,
      "step": 2947
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.545591963722861,
      "learning_rate": 9.166656422646396e-06,
      "loss": 0.7984,
      "step": 2948
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4563237314064592,
      "learning_rate": 9.166015975140502e-06,
      "loss": 0.1872,
      "step": 2949
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3093818155297914,
      "learning_rate": 9.165375304019422e-06,
      "loss": 0.2204,
      "step": 2950
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3301106967671248,
      "learning_rate": 9.164734409317542e-06,
      "loss": 0.2359,
      "step": 2951
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.394982635646508,
      "learning_rate": 9.164093291069267e-06,
      "loss": 0.2334,
      "step": 2952
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.523469364770285,
      "learning_rate": 9.163451949309005e-06,
      "loss": 0.2252,
      "step": 2953
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4400366451853166,
      "learning_rate": 9.162810384071186e-06,
      "loss": 0.2073,
      "step": 2954
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5001236193416456,
      "learning_rate": 9.16216859539024e-06,
      "loss": 0.228,
      "step": 2955
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5024858258321905,
      "learning_rate": 9.161526583300621e-06,
      "loss": 0.2313,
      "step": 2956
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3305208293726116,
      "learning_rate": 9.160884347836789e-06,
      "loss": 0.1792,
      "step": 2957
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1188431289538487,
      "learning_rate": 9.160241889033213e-06,
      "loss": 0.1732,
      "step": 2958
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4341096989102657,
      "learning_rate": 9.159599206924383e-06,
      "loss": 0.2159,
      "step": 2959
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5016750056550827,
      "learning_rate": 9.158956301544791e-06,
      "loss": 0.2743,
      "step": 2960
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2940109056668865,
      "learning_rate": 9.158313172928948e-06,
      "loss": 0.1973,
      "step": 2961
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2937943032426407,
      "learning_rate": 9.157669821111373e-06,
      "loss": 0.1955,
      "step": 2962
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5408680491649607,
      "learning_rate": 9.157026246126599e-06,
      "loss": 0.2481,
      "step": 2963
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5393186923767896,
      "learning_rate": 9.156382448009173e-06,
      "loss": 0.1958,
      "step": 2964
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4027958718970357,
      "learning_rate": 9.15573842679365e-06,
      "loss": 0.2148,
      "step": 2965
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.453649674160241,
      "learning_rate": 9.155094182514595e-06,
      "loss": 0.7086,
      "step": 2966
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4662910818257089,
      "learning_rate": 9.154449715206595e-06,
      "loss": 0.1844,
      "step": 2967
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3489516261763703,
      "learning_rate": 9.153805024904236e-06,
      "loss": 0.239,
      "step": 2968
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2137091547192778,
      "learning_rate": 9.153160111642127e-06,
      "loss": 0.1561,
      "step": 2969
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4749059306062013,
      "learning_rate": 9.152514975454884e-06,
      "loss": 0.2293,
      "step": 2970
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.393611101433033,
      "learning_rate": 9.151869616377132e-06,
      "loss": 0.212,
      "step": 2971
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6345451446085344,
      "learning_rate": 9.151224034443516e-06,
      "loss": 0.3214,
      "step": 2972
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7474050805689048,
      "learning_rate": 9.150578229688684e-06,
      "loss": 0.2856,
      "step": 2973
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.415333897237682,
      "learning_rate": 9.149932202147302e-06,
      "loss": 0.2072,
      "step": 2974
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.620666572430044,
      "learning_rate": 9.149285951854047e-06,
      "loss": 0.2297,
      "step": 2975
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.483948549363042,
      "learning_rate": 9.148639478843606e-06,
      "loss": 0.2149,
      "step": 2976
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5429731254719918,
      "learning_rate": 9.147992783150679e-06,
      "loss": 0.2402,
      "step": 2977
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2824180752272012,
      "learning_rate": 9.14734586480998e-06,
      "loss": 0.2279,
      "step": 2978
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.406260435718829,
      "learning_rate": 9.146698723856228e-06,
      "loss": 0.2296,
      "step": 2979
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4512049244634375,
      "learning_rate": 9.146051360324166e-06,
      "loss": 0.2134,
      "step": 2980
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4146936651181095,
      "learning_rate": 9.145403774248536e-06,
      "loss": 0.1947,
      "step": 2981
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.521867269846185,
      "learning_rate": 9.144755965664102e-06,
      "loss": 0.2491,
      "step": 2982
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2943458397664256,
      "learning_rate": 9.144107934605634e-06,
      "loss": 0.2182,
      "step": 2983
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4757967065460242,
      "learning_rate": 9.143459681107915e-06,
      "loss": 0.2219,
      "step": 2984
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3770888511956474,
      "learning_rate": 9.142811205205742e-06,
      "loss": 0.1957,
      "step": 2985
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3424948673508654,
      "learning_rate": 9.142162506933921e-06,
      "loss": 0.2403,
      "step": 2986
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4421371902080553,
      "learning_rate": 9.141513586327273e-06,
      "loss": 0.2366,
      "step": 2987
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4306553360803718,
      "learning_rate": 9.140864443420629e-06,
      "loss": 0.227,
      "step": 2988
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.139658732723522,
      "learning_rate": 9.140215078248833e-06,
      "loss": 0.7777,
      "step": 2989
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4169625821803593,
      "learning_rate": 9.13956549084674e-06,
      "loss": 0.2259,
      "step": 2990
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.34830775179381,
      "learning_rate": 9.138915681249217e-06,
      "loss": 0.2523,
      "step": 2991
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5019153909923473,
      "learning_rate": 9.138265649491142e-06,
      "loss": 0.2718,
      "step": 2992
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.461254835728262,
      "learning_rate": 9.137615395607409e-06,
      "loss": 0.5928,
      "step": 2993
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.418195033785992,
      "learning_rate": 9.136964919632918e-06,
      "loss": 0.2262,
      "step": 2994
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2327639140404898,
      "learning_rate": 9.136314221602585e-06,
      "loss": 0.2206,
      "step": 2995
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3451593453030868,
      "learning_rate": 9.135663301551339e-06,
      "loss": 0.1796,
      "step": 2996
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2715335402641175,
      "learning_rate": 9.135012159514116e-06,
      "loss": 0.1835,
      "step": 2997
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4349084599292932,
      "learning_rate": 9.134360795525869e-06,
      "loss": 0.225,
      "step": 2998
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4063227352155494,
      "learning_rate": 9.133709209621559e-06,
      "loss": 0.2413,
      "step": 2999
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3298396166778672,
      "learning_rate": 9.13305740183616e-06,
      "loss": 0.1999,
      "step": 3000
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4280543500594016,
      "learning_rate": 9.13240537220466e-06,
      "loss": 0.2132,
      "step": 3001
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7267700938759818,
      "learning_rate": 9.131753120762057e-06,
      "loss": 0.2595,
      "step": 3002
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4551700898693578,
      "learning_rate": 9.13110064754336e-06,
      "loss": 0.2482,
      "step": 3003
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.894896547989317,
      "learning_rate": 9.130447952583595e-06,
      "loss": 0.4624,
      "step": 3004
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.455237938140936,
      "learning_rate": 9.12979503591779e-06,
      "loss": 0.1876,
      "step": 3005
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3889517326581236,
      "learning_rate": 9.129141897580995e-06,
      "loss": 0.2361,
      "step": 3006
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4740452259439505,
      "learning_rate": 9.128488537608269e-06,
      "loss": 0.2019,
      "step": 3007
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3870506617078244,
      "learning_rate": 9.127834956034679e-06,
      "loss": 0.1916,
      "step": 3008
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2196918840690139,
      "learning_rate": 9.127181152895308e-06,
      "loss": 0.19,
      "step": 3009
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3911461653887733,
      "learning_rate": 9.126527128225247e-06,
      "loss": 0.2276,
      "step": 3010
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2707836670071562,
      "learning_rate": 9.125872882059605e-06,
      "loss": 0.2115,
      "step": 3011
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3710191784202057,
      "learning_rate": 9.125218414433498e-06,
      "loss": 0.1987,
      "step": 3012
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3219643107282224,
      "learning_rate": 9.124563725382056e-06,
      "loss": 0.1467,
      "step": 3013
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3106134955741449,
      "learning_rate": 9.123908814940418e-06,
      "loss": 0.2091,
      "step": 3014
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4328400388253453,
      "learning_rate": 9.123253683143738e-06,
      "loss": 0.2315,
      "step": 3015
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5183517507411208,
      "learning_rate": 9.122598330027183e-06,
      "loss": 0.2285,
      "step": 3016
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6896997381230359,
      "learning_rate": 9.121942755625925e-06,
      "loss": 0.2802,
      "step": 3017
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.748622611984189,
      "learning_rate": 9.121286959975159e-06,
      "loss": 0.24,
      "step": 3018
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4388623471874564,
      "learning_rate": 9.120630943110078e-06,
      "loss": 0.2264,
      "step": 3019
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.534226535168014,
      "learning_rate": 9.119974705065902e-06,
      "loss": 0.235,
      "step": 3020
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.458656626281266,
      "learning_rate": 9.119318245877849e-06,
      "loss": 0.2491,
      "step": 3021
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5062738683850154,
      "learning_rate": 9.11866156558116e-06,
      "loss": 0.2161,
      "step": 3022
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4615222759233526,
      "learning_rate": 9.118004664211078e-06,
      "loss": 0.2307,
      "step": 3023
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3951390025222694,
      "learning_rate": 9.117347541802868e-06,
      "loss": 0.2393,
      "step": 3024
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.562445744050878,
      "learning_rate": 9.116690198391797e-06,
      "loss": 0.2183,
      "step": 3025
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5446008757300853,
      "learning_rate": 9.116032634013154e-06,
      "loss": 0.2656,
      "step": 3026
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3613353560372505,
      "learning_rate": 9.11537484870223e-06,
      "loss": 0.1816,
      "step": 3027
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4850661023273644,
      "learning_rate": 9.114716842494333e-06,
      "loss": 0.2423,
      "step": 3028
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.963642580207885,
      "learning_rate": 9.114058615424783e-06,
      "loss": 0.567,
      "step": 3029
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4947142503305477,
      "learning_rate": 9.11340016752891e-06,
      "loss": 0.2161,
      "step": 3030
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4055485963174663,
      "learning_rate": 9.112741498842059e-06,
      "loss": 0.238,
      "step": 3031
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3717743001773361,
      "learning_rate": 9.112082609399585e-06,
      "loss": 0.1733,
      "step": 3032
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5924545538273276,
      "learning_rate": 9.111423499236849e-06,
      "loss": 0.2551,
      "step": 3033
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4718584536406234,
      "learning_rate": 9.110764168389236e-06,
      "loss": 0.2111,
      "step": 3034
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.475152152205069,
      "learning_rate": 9.110104616892132e-06,
      "loss": 0.2064,
      "step": 3035
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4302155103564766,
      "learning_rate": 9.109444844780942e-06,
      "loss": 0.1805,
      "step": 3036
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3017397429582174,
      "learning_rate": 9.108784852091079e-06,
      "loss": 0.2121,
      "step": 3037
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.517405486868771,
      "learning_rate": 9.108124638857967e-06,
      "loss": 0.1772,
      "step": 3038
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.341607582504751,
      "learning_rate": 9.107464205117047e-06,
      "loss": 0.2122,
      "step": 3039
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.165663380071189,
      "learning_rate": 9.106803550903765e-06,
      "loss": 0.4884,
      "step": 3040
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5200212390450671,
      "learning_rate": 9.106142676253584e-06,
      "loss": 0.2276,
      "step": 3041
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4380004076505872,
      "learning_rate": 9.105481581201978e-06,
      "loss": 0.1894,
      "step": 3042
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2954721613290607,
      "learning_rate": 9.10482026578443e-06,
      "loss": 0.1952,
      "step": 3043
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.551039069988955,
      "learning_rate": 9.104158730036439e-06,
      "loss": 0.2213,
      "step": 3044
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.372035457146638,
      "learning_rate": 9.10349697399351e-06,
      "loss": 0.2493,
      "step": 3045
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5605644629039561,
      "learning_rate": 9.102834997691167e-06,
      "loss": 0.2607,
      "step": 3046
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.43958960478813,
      "learning_rate": 9.102172801164942e-06,
      "loss": 0.2387,
      "step": 3047
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5586279069795774,
      "learning_rate": 9.101510384450377e-06,
      "loss": 0.2438,
      "step": 3048
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.570852692586721,
      "learning_rate": 9.10084774758303e-06,
      "loss": 0.24,
      "step": 3049
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4738828643951447,
      "learning_rate": 9.100184890598467e-06,
      "loss": 0.2331,
      "step": 3050
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5054705197867058,
      "learning_rate": 9.099521813532268e-06,
      "loss": 0.2253,
      "step": 3051
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5182933584250808,
      "learning_rate": 9.098858516420023e-06,
      "loss": 0.2501,
      "step": 3052
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.095098728044864,
      "learning_rate": 9.098194999297339e-06,
      "loss": 0.7189,
      "step": 3053
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3866920682300161,
      "learning_rate": 9.097531262199828e-06,
      "loss": 0.2272,
      "step": 3054
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4153347174841637,
      "learning_rate": 9.09686730516312e-06,
      "loss": 0.2542,
      "step": 3055
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4393718472581118,
      "learning_rate": 9.096203128222847e-06,
      "loss": 0.2262,
      "step": 3056
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3993008788829686,
      "learning_rate": 9.095538731414666e-06,
      "loss": 0.2188,
      "step": 3057
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3805482638289766,
      "learning_rate": 9.094874114774235e-06,
      "loss": 0.2086,
      "step": 3058
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.575546014928864,
      "learning_rate": 9.094209278337232e-06,
      "loss": 0.238,
      "step": 3059
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.302214325231813,
      "learning_rate": 9.093544222139338e-06,
      "loss": 0.1954,
      "step": 3060
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.375681807825601,
      "learning_rate": 9.092878946216252e-06,
      "loss": 0.2398,
      "step": 3061
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2959780151658282,
      "learning_rate": 9.092213450603687e-06,
      "loss": 0.214,
      "step": 3062
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.479116322897022,
      "learning_rate": 9.09154773533736e-06,
      "loss": 0.1788,
      "step": 3063
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3763315893801809,
      "learning_rate": 9.090881800453006e-06,
      "loss": 0.2154,
      "step": 3064
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.298242616584281,
      "learning_rate": 9.09021564598637e-06,
      "loss": 0.204,
      "step": 3065
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4167164560861139,
      "learning_rate": 9.089549271973207e-06,
      "loss": 0.2173,
      "step": 3066
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2812439930119288,
      "learning_rate": 9.088882678449285e-06,
      "loss": 0.2037,
      "step": 3067
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4978058371200331,
      "learning_rate": 9.088215865450386e-06,
      "loss": 0.2375,
      "step": 3068
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.392670287565296,
      "learning_rate": 9.087548833012301e-06,
      "loss": 0.2198,
      "step": 3069
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6134189057385846,
      "learning_rate": 9.086881581170834e-06,
      "loss": 0.2559,
      "step": 3070
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4402453810896767,
      "learning_rate": 9.0862141099618e-06,
      "loss": 0.1733,
      "step": 3071
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.583529395198567,
      "learning_rate": 9.085546419421026e-06,
      "loss": 0.2457,
      "step": 3072
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.862349412477856,
      "learning_rate": 9.08487850958435e-06,
      "loss": 0.6775,
      "step": 3073
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3777231530134035,
      "learning_rate": 9.084210380487624e-06,
      "loss": 0.188,
      "step": 3074
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3611501690441554,
      "learning_rate": 9.083542032166712e-06,
      "loss": 0.2019,
      "step": 3075
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4694260224150932,
      "learning_rate": 9.082873464657486e-06,
      "loss": 0.2426,
      "step": 3076
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2860551361951889,
      "learning_rate": 9.082204677995832e-06,
      "loss": 0.1779,
      "step": 3077
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.423730286399062,
      "learning_rate": 9.081535672217651e-06,
      "loss": 0.1782,
      "step": 3078
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.6547524252785575,
      "learning_rate": 9.080866447358849e-06,
      "loss": 0.6702,
      "step": 3079
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.343235357299407,
      "learning_rate": 9.080197003455347e-06,
      "loss": 0.2333,
      "step": 3080
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3404968854374828,
      "learning_rate": 9.079527340543082e-06,
      "loss": 0.2761,
      "step": 3081
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5419329235591352,
      "learning_rate": 9.078857458657995e-06,
      "loss": 0.22,
      "step": 3082
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5020062033996753,
      "learning_rate": 9.078187357836043e-06,
      "loss": 0.2159,
      "step": 3083
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2933799428719461,
      "learning_rate": 9.077517038113197e-06,
      "loss": 0.1797,
      "step": 3084
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5265142388755737,
      "learning_rate": 9.076846499525436e-06,
      "loss": 0.2151,
      "step": 3085
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3892483603033727,
      "learning_rate": 9.07617574210875e-06,
      "loss": 0.2239,
      "step": 3086
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3471434123066444,
      "learning_rate": 9.075504765899145e-06,
      "loss": 0.2144,
      "step": 3087
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6247715660885467,
      "learning_rate": 9.074833570932634e-06,
      "loss": 0.2442,
      "step": 3088
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.615856520882401,
      "learning_rate": 9.074162157245248e-06,
      "loss": 0.2323,
      "step": 3089
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4738229671936398,
      "learning_rate": 9.07349052487302e-06,
      "loss": 0.278,
      "step": 3090
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3576676047963705,
      "learning_rate": 9.072818673852006e-06,
      "loss": 0.1989,
      "step": 3091
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4049716898845126,
      "learning_rate": 9.072146604218266e-06,
      "loss": 0.2074,
      "step": 3092
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.399466741478571,
      "learning_rate": 9.071474316007874e-06,
      "loss": 0.2408,
      "step": 3093
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6050994153710094,
      "learning_rate": 9.070801809256915e-06,
      "loss": 0.2086,
      "step": 3094
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.634783221499501,
      "learning_rate": 9.070129084001486e-06,
      "loss": 0.2265,
      "step": 3095
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.820896781602804,
      "learning_rate": 9.0694561402777e-06,
      "loss": 0.8011,
      "step": 3096
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.550301927650956,
      "learning_rate": 9.068782978121675e-06,
      "loss": 0.2557,
      "step": 3097
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5067359719988793,
      "learning_rate": 9.068109597569544e-06,
      "loss": 0.2249,
      "step": 3098
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.313580573758077,
      "learning_rate": 9.06743599865745e-06,
      "loss": 0.5505,
      "step": 3099
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.391490799551258,
      "learning_rate": 9.066762181421552e-06,
      "loss": 0.1928,
      "step": 3100
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3587038662699662,
      "learning_rate": 9.066088145898017e-06,
      "loss": 0.2139,
      "step": 3101
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.393420355064994,
      "learning_rate": 9.065413892123024e-06,
      "loss": 0.2286,
      "step": 3102
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4738503317613607,
      "learning_rate": 9.064739420132761e-06,
      "loss": 0.2028,
      "step": 3103
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3833652833574654,
      "learning_rate": 9.06406472996344e-06,
      "loss": 0.2424,
      "step": 3104
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.982873133522027,
      "learning_rate": 9.063389821651265e-06,
      "loss": 0.7066,
      "step": 3105
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6506590188069235,
      "learning_rate": 9.06271469523247e-06,
      "loss": 0.2525,
      "step": 3106
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4261458714564441,
      "learning_rate": 9.06203935074329e-06,
      "loss": 0.2693,
      "step": 3107
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.461768602089681,
      "learning_rate": 9.061363788219975e-06,
      "loss": 0.2425,
      "step": 3108
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.909424870684166,
      "learning_rate": 9.060688007698787e-06,
      "loss": 0.7189,
      "step": 3109
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3541743352047728,
      "learning_rate": 9.060012009215999e-06,
      "loss": 0.2186,
      "step": 3110
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.860032310828481,
      "learning_rate": 9.059335792807896e-06,
      "loss": 0.6798,
      "step": 3111
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4382055948299062,
      "learning_rate": 9.058659358510774e-06,
      "loss": 0.2274,
      "step": 3112
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4517122439149788,
      "learning_rate": 9.057982706360942e-06,
      "loss": 0.2398,
      "step": 3113
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.528432676062184,
      "learning_rate": 9.05730583639472e-06,
      "loss": 0.2427,
      "step": 3114
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4671933443084157,
      "learning_rate": 9.05662874864844e-06,
      "loss": 0.1921,
      "step": 3115
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3892865370872574,
      "learning_rate": 9.055951443158445e-06,
      "loss": 0.207,
      "step": 3116
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2271431427823836,
      "learning_rate": 9.055273919961089e-06,
      "loss": 0.1942,
      "step": 3117
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4506253933662845,
      "learning_rate": 9.054596179092739e-06,
      "loss": 0.2402,
      "step": 3118
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.283514446520111,
      "learning_rate": 9.053918220589776e-06,
      "loss": 0.2037,
      "step": 3119
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4125661677551773,
      "learning_rate": 9.053240044488587e-06,
      "loss": 0.2184,
      "step": 3120
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3426478385223595,
      "learning_rate": 9.052561650825575e-06,
      "loss": 0.1755,
      "step": 3121
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.304310556183331,
      "learning_rate": 9.051883039637155e-06,
      "loss": 0.2106,
      "step": 3122
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4008051832117994,
      "learning_rate": 9.051204210959749e-06,
      "loss": 0.1821,
      "step": 3123
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2700401304587063,
      "learning_rate": 9.050525164829797e-06,
      "loss": 0.2093,
      "step": 3124
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3075651981892595,
      "learning_rate": 9.049845901283746e-06,
      "loss": 0.17,
      "step": 3125
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6028507093490816,
      "learning_rate": 9.049166420358056e-06,
      "loss": 0.2927,
      "step": 3126
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4419739490653496,
      "learning_rate": 9.048486722089197e-06,
      "loss": 0.1981,
      "step": 3127
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4834209522926776,
      "learning_rate": 9.047806806513657e-06,
      "loss": 0.228,
      "step": 3128
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2671611224554695,
      "learning_rate": 9.047126673667929e-06,
      "loss": 0.18,
      "step": 3129
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6589943518084107,
      "learning_rate": 9.046446323588518e-06,
      "loss": 0.2433,
      "step": 3130
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3639326500262245,
      "learning_rate": 9.045765756311944e-06,
      "loss": 0.2189,
      "step": 3131
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.579941761001767,
      "learning_rate": 9.045084971874738e-06,
      "loss": 0.5439,
      "step": 3132
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4443011167960502,
      "learning_rate": 9.04440397031344e-06,
      "loss": 0.2119,
      "step": 3133
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.408106286512159,
      "learning_rate": 9.043722751664606e-06,
      "loss": 0.5594,
      "step": 3134
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.455799147315538,
      "learning_rate": 9.043041315964797e-06,
      "loss": 0.1983,
      "step": 3135
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4318257825930656,
      "learning_rate": 9.042359663250595e-06,
      "loss": 0.2296,
      "step": 3136
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.496274749187072,
      "learning_rate": 9.041677793558586e-06,
      "loss": 0.189,
      "step": 3137
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4598534225806843,
      "learning_rate": 9.040995706925368e-06,
      "loss": 0.2001,
      "step": 3138
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.56329828889496,
      "learning_rate": 9.040313403387555e-06,
      "loss": 0.5713,
      "step": 3139
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4594863013418782,
      "learning_rate": 9.039630882981769e-06,
      "loss": 0.2085,
      "step": 3140
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.410552904152116,
      "learning_rate": 9.038948145744647e-06,
      "loss": 0.1698,
      "step": 3141
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4609510355601452,
      "learning_rate": 9.038265191712833e-06,
      "loss": 0.187,
      "step": 3142
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4617049072236215,
      "learning_rate": 9.037582020922988e-06,
      "loss": 0.2167,
      "step": 3143
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5031952306224532,
      "learning_rate": 9.036898633411781e-06,
      "loss": 0.2145,
      "step": 3144
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5017250537407745,
      "learning_rate": 9.036215029215892e-06,
      "loss": 0.248,
      "step": 3145
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.481497369269224,
      "learning_rate": 9.035531208372016e-06,
      "loss": 0.2096,
      "step": 3146
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3088068408624856,
      "learning_rate": 9.034847170916857e-06,
      "loss": 0.1892,
      "step": 3147
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5167000697990123,
      "learning_rate": 9.034162916887131e-06,
      "loss": 0.2303,
      "step": 3148
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3008429143131708,
      "learning_rate": 9.033478446319565e-06,
      "loss": 0.2114,
      "step": 3149
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3185034965123557,
      "learning_rate": 9.032793759250903e-06,
      "loss": 0.2099,
      "step": 3150
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4200025049597909,
      "learning_rate": 9.03210885571789e-06,
      "loss": 0.2076,
      "step": 3151
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4441790099506742,
      "learning_rate": 9.031423735757296e-06,
      "loss": 0.2449,
      "step": 3152
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4408584758246343,
      "learning_rate": 9.030738399405891e-06,
      "loss": 0.2127,
      "step": 3153
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.383397977199835,
      "learning_rate": 9.030052846700462e-06,
      "loss": 0.2482,
      "step": 3154
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4172496286710512,
      "learning_rate": 9.029367077677807e-06,
      "loss": 0.1669,
      "step": 3155
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3080856995265862,
      "learning_rate": 9.028681092374733e-06,
      "loss": 0.1898,
      "step": 3156
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3740473249753982,
      "learning_rate": 9.027994890828065e-06,
      "loss": 0.2183,
      "step": 3157
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.540079901954493,
      "learning_rate": 9.027308473074633e-06,
      "loss": 0.6943,
      "step": 3158
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.309360876920952,
      "learning_rate": 9.026621839151282e-06,
      "loss": 0.1587,
      "step": 3159
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5587679645842738,
      "learning_rate": 9.025934989094866e-06,
      "loss": 0.2262,
      "step": 3160
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.316631589828942,
      "learning_rate": 9.025247922942256e-06,
      "loss": 0.2196,
      "step": 3161
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4270498299951493,
      "learning_rate": 9.024560640730328e-06,
      "loss": 0.1895,
      "step": 3162
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4038980039618878,
      "learning_rate": 9.023873142495973e-06,
      "loss": 0.1801,
      "step": 3163
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3255108023193858,
      "learning_rate": 9.023185428276095e-06,
      "loss": 0.2354,
      "step": 3164
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4758133652302277,
      "learning_rate": 9.022497498107607e-06,
      "loss": 0.2417,
      "step": 3165
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3752511682461506,
      "learning_rate": 9.021809352027433e-06,
      "loss": 0.1927,
      "step": 3166
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.479281693261229,
      "learning_rate": 9.02112099007251e-06,
      "loss": 0.243,
      "step": 3167
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3087628232523358,
      "learning_rate": 9.020432412279789e-06,
      "loss": 0.184,
      "step": 3168
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4489142467153464,
      "learning_rate": 9.019743618686226e-06,
      "loss": 0.2348,
      "step": 3169
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4677376178577999,
      "learning_rate": 9.019054609328796e-06,
      "loss": 0.1848,
      "step": 3170
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.273660363404849,
      "learning_rate": 9.018365384244483e-06,
      "loss": 0.229,
      "step": 3171
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4744876308426846,
      "learning_rate": 9.01767594347028e-06,
      "loss": 0.2179,
      "step": 3172
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.545305547174912,
      "learning_rate": 9.016986287043193e-06,
      "loss": 0.2116,
      "step": 3173
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4915769035904196,
      "learning_rate": 9.016296415000241e-06,
      "loss": 0.2265,
      "step": 3174
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3687396335099447,
      "learning_rate": 9.015606327378453e-06,
      "loss": 0.2256,
      "step": 3175
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7114140537035492,
      "learning_rate": 9.01491602421487e-06,
      "loss": 0.2084,
      "step": 3176
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5365783826583845,
      "learning_rate": 9.014225505546545e-06,
      "loss": 0.2273,
      "step": 3177
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5898198881183163,
      "learning_rate": 9.013534771410544e-06,
      "loss": 0.2341,
      "step": 3178
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3571622408935828,
      "learning_rate": 9.012843821843941e-06,
      "loss": 0.2113,
      "step": 3179
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.697027857439972,
      "learning_rate": 9.012152656883824e-06,
      "loss": 0.2421,
      "step": 3180
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5212639457440849,
      "learning_rate": 9.01146127656729e-06,
      "loss": 0.1923,
      "step": 3181
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.374267377501372,
      "learning_rate": 9.010769680931454e-06,
      "loss": 0.2611,
      "step": 3182
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3577588440248787,
      "learning_rate": 9.010077870013435e-06,
      "loss": 0.2476,
      "step": 3183
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4891901492574346,
      "learning_rate": 9.009385843850366e-06,
      "loss": 0.2274,
      "step": 3184
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.451760575657211,
      "learning_rate": 9.008693602479392e-06,
      "loss": 0.225,
      "step": 3185
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.74594614702044,
      "learning_rate": 9.008001145937675e-06,
      "loss": 0.2439,
      "step": 3186
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4256278661675348,
      "learning_rate": 9.007308474262378e-06,
      "loss": 0.2215,
      "step": 3187
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3514782376673284,
      "learning_rate": 9.00661558749068e-06,
      "loss": 0.2308,
      "step": 3188
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2809703659553242,
      "learning_rate": 9.005922485659777e-06,
      "loss": 0.213,
      "step": 3189
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2160509839341365,
      "learning_rate": 9.005229168806869e-06,
      "loss": 0.1637,
      "step": 3190
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.133087376721232,
      "learning_rate": 9.00453563696917e-06,
      "loss": 0.6876,
      "step": 3191
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5037172568326707,
      "learning_rate": 9.003841890183908e-06,
      "loss": 0.2004,
      "step": 3192
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3167217856544602,
      "learning_rate": 9.00314792848832e-06,
      "loss": 0.1832,
      "step": 3193
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3041244312590856,
      "learning_rate": 9.002453751919655e-06,
      "loss": 0.1984,
      "step": 3194
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3127729295467228,
      "learning_rate": 9.001759360515175e-06,
      "loss": 0.2126,
      "step": 3195
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2492951653987943,
      "learning_rate": 9.001064754312148e-06,
      "loss": 0.1755,
      "step": 3196
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4802191001341758,
      "learning_rate": 9.00036993334786e-06,
      "loss": 0.2563,
      "step": 3197
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.110759055604152,
      "learning_rate": 8.999674897659609e-06,
      "loss": 0.6189,
      "step": 3198
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6213529442893517,
      "learning_rate": 8.998979647284698e-06,
      "loss": 0.2187,
      "step": 3199
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4059377209471042,
      "learning_rate": 8.998284182260448e-06,
      "loss": 0.2536,
      "step": 3200
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.395358619219628,
      "learning_rate": 8.997588502624186e-06,
      "loss": 0.2077,
      "step": 3201
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.080373494390886,
      "learning_rate": 8.996892608413254e-06,
      "loss": 0.2668,
      "step": 3202
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7763310644364663,
      "learning_rate": 8.99619649966501e-06,
      "loss": 0.2101,
      "step": 3203
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5822614452434214,
      "learning_rate": 8.995500176416809e-06,
      "loss": 0.2281,
      "step": 3204
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4259913307625192,
      "learning_rate": 8.994803638706033e-06,
      "loss": 0.2234,
      "step": 3205
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.731753048678132,
      "learning_rate": 8.994106886570069e-06,
      "loss": 0.674,
      "step": 3206
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4493100939504822,
      "learning_rate": 8.993409920046315e-06,
      "loss": 0.2885,
      "step": 3207
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4136468104460167,
      "learning_rate": 8.99271273917218e-06,
      "loss": 0.227,
      "step": 3208
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.214134518468838,
      "learning_rate": 8.99201534398509e-06,
      "loss": 0.1936,
      "step": 3209
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.161569171056517,
      "learning_rate": 8.991317734522475e-06,
      "loss": 0.6661,
      "step": 3210
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3443046610451856,
      "learning_rate": 8.990619910821779e-06,
      "loss": 0.2172,
      "step": 3211
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4391187234569074,
      "learning_rate": 8.989921872920462e-06,
      "loss": 0.1777,
      "step": 3212
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5870561010254274,
      "learning_rate": 8.989223620855989e-06,
      "loss": 0.2353,
      "step": 3213
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3162845902071516,
      "learning_rate": 8.988525154665842e-06,
      "loss": 0.1524,
      "step": 3214
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4499813871002178,
      "learning_rate": 8.987826474387507e-06,
      "loss": 0.2116,
      "step": 3215
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3898886209879526,
      "learning_rate": 8.987127580058492e-06,
      "loss": 0.2197,
      "step": 3216
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3683138112216187,
      "learning_rate": 8.986428471716308e-06,
      "loss": 0.1906,
      "step": 3217
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3725151241405702,
      "learning_rate": 8.98572914939848e-06,
      "loss": 0.1904,
      "step": 3218
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5775040302906564,
      "learning_rate": 8.985029613142548e-06,
      "loss": 0.26,
      "step": 3219
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.333948626162973,
      "learning_rate": 8.984329862986056e-06,
      "loss": 0.1987,
      "step": 3220
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.223090017271359,
      "learning_rate": 8.983629898966566e-06,
      "loss": 0.1752,
      "step": 3221
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.748776699790595,
      "learning_rate": 8.98292972112165e-06,
      "loss": 0.4016,
      "step": 3222
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3089830235128994,
      "learning_rate": 8.982229329488889e-06,
      "loss": 0.1943,
      "step": 3223
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4425652085101002,
      "learning_rate": 8.98152872410588e-06,
      "loss": 0.2473,
      "step": 3224
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3641048051139864,
      "learning_rate": 8.980827905010226e-06,
      "loss": 0.2242,
      "step": 3225
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.493343081710168,
      "learning_rate": 8.980126872239543e-06,
      "loss": 0.2482,
      "step": 3226
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6677565606161084,
      "learning_rate": 8.979425625831465e-06,
      "loss": 0.2223,
      "step": 3227
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.441680509273604,
      "learning_rate": 8.978724165823626e-06,
      "loss": 0.2414,
      "step": 3228
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2203954117430933,
      "learning_rate": 8.978022492253681e-06,
      "loss": 0.2068,
      "step": 3229
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.446225807207349,
      "learning_rate": 8.977320605159294e-06,
      "loss": 0.2245,
      "step": 3230
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3899562931464085,
      "learning_rate": 8.976618504578136e-06,
      "loss": 0.2392,
      "step": 3231
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3958861514752925,
      "learning_rate": 8.975916190547897e-06,
      "loss": 0.2091,
      "step": 3232
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3705652383212996,
      "learning_rate": 8.97521366310627e-06,
      "loss": 0.203,
      "step": 3233
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4331444789057235,
      "learning_rate": 8.974510922290968e-06,
      "loss": 0.2751,
      "step": 3234
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.500506293702792,
      "learning_rate": 8.97380796813971e-06,
      "loss": 0.71,
      "step": 3235
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2615768578871394,
      "learning_rate": 8.973104800690226e-06,
      "loss": 0.1716,
      "step": 3236
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.23751656026381,
      "learning_rate": 8.972401419980262e-06,
      "loss": 0.166,
      "step": 3237
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4511289385426775,
      "learning_rate": 8.97169782604757e-06,
      "loss": 0.2312,
      "step": 3238
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.282322235389946,
      "learning_rate": 8.970994018929917e-06,
      "loss": 0.1902,
      "step": 3239
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3983098449734295,
      "learning_rate": 8.970289998665083e-06,
      "loss": 0.2387,
      "step": 3240
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.405867915681503,
      "learning_rate": 8.969585765290853e-06,
      "loss": 0.2505,
      "step": 3241
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5374575437133131,
      "learning_rate": 8.96888131884503e-06,
      "loss": 0.2027,
      "step": 3242
    },
    {
      "epoch": 0.23,
      "grad_norm": 14.947900236737672,
      "learning_rate": 8.968176659365426e-06,
      "loss": 0.6052,
      "step": 3243
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.334367057617126,
      "learning_rate": 8.967471786889863e-06,
      "loss": 0.157,
      "step": 3244
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.394586182238013,
      "learning_rate": 8.966766701456177e-06,
      "loss": 0.2278,
      "step": 3245
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4403049322476118,
      "learning_rate": 8.966061403102213e-06,
      "loss": 0.225,
      "step": 3246
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3759637517081997,
      "learning_rate": 8.96535589186583e-06,
      "loss": 0.2144,
      "step": 3247
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.475982869310369,
      "learning_rate": 8.964650167784895e-06,
      "loss": 0.2369,
      "step": 3248
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3847048798412704,
      "learning_rate": 8.96394423089729e-06,
      "loss": 0.2176,
      "step": 3249
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.511481528160134,
      "learning_rate": 8.963238081240908e-06,
      "loss": 0.2209,
      "step": 3250
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3985750723212627,
      "learning_rate": 8.96253171885365e-06,
      "loss": 0.2128,
      "step": 3251
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2752337880192919,
      "learning_rate": 8.96182514377343e-06,
      "loss": 0.187,
      "step": 3252
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3249883079945826,
      "learning_rate": 8.961118356038179e-06,
      "loss": 0.1973,
      "step": 3253
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6044797164587958,
      "learning_rate": 8.96041135568583e-06,
      "loss": 0.2278,
      "step": 3254
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3220423255461502,
      "learning_rate": 8.959704142754331e-06,
      "loss": 0.2038,
      "step": 3255
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4906115572189955,
      "learning_rate": 8.958996717281648e-06,
      "loss": 0.2829,
      "step": 3256
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.445090684175075,
      "learning_rate": 8.958289079305747e-06,
      "loss": 0.2269,
      "step": 3257
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2387680616871306,
      "learning_rate": 8.957581228864616e-06,
      "loss": 0.2081,
      "step": 3258
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4255887515723153,
      "learning_rate": 8.956873165996245e-06,
      "loss": 0.1826,
      "step": 3259
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2772500349247426,
      "learning_rate": 8.956164890738643e-06,
      "loss": 0.2256,
      "step": 3260
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4934700634627525,
      "learning_rate": 8.955456403129828e-06,
      "loss": 0.2279,
      "step": 3261
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4457790992712745,
      "learning_rate": 8.954747703207826e-06,
      "loss": 0.2417,
      "step": 3262
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3033396036548048,
      "learning_rate": 8.95403879101068e-06,
      "loss": 0.1714,
      "step": 3263
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4536092736551012,
      "learning_rate": 8.953329666576439e-06,
      "loss": 0.2011,
      "step": 3264
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5226812385381971,
      "learning_rate": 8.952620329943168e-06,
      "loss": 0.2723,
      "step": 3265
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.983647940918482,
      "learning_rate": 8.95191078114894e-06,
      "loss": 0.4387,
      "step": 3266
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3554911560700096,
      "learning_rate": 8.95120102023184e-06,
      "loss": 0.2106,
      "step": 3267
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5086663228110664,
      "learning_rate": 8.95049104722997e-06,
      "loss": 0.2534,
      "step": 3268
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3854082469218247,
      "learning_rate": 8.949780862181432e-06,
      "loss": 0.2278,
      "step": 3269
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3804243408483254,
      "learning_rate": 8.949070465124352e-06,
      "loss": 0.2119,
      "step": 3270
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4561605996855538,
      "learning_rate": 8.948359856096855e-06,
      "loss": 0.1916,
      "step": 3271
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5314212914012642,
      "learning_rate": 8.94764903513709e-06,
      "loss": 0.2236,
      "step": 3272
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4282994510349154,
      "learning_rate": 8.946938002283206e-06,
      "loss": 0.2426,
      "step": 3273
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2951682280409567,
      "learning_rate": 8.946226757573371e-06,
      "loss": 0.2087,
      "step": 3274
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3099381063232904,
      "learning_rate": 8.945515301045759e-06,
      "loss": 0.2211,
      "step": 3275
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3063613448461173,
      "learning_rate": 8.944803632738563e-06,
      "loss": 0.2165,
      "step": 3276
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4192931220108251,
      "learning_rate": 8.944091752689979e-06,
      "loss": 0.2261,
      "step": 3277
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.307477581721814,
      "learning_rate": 8.943379660938219e-06,
      "loss": 0.6033,
      "step": 3278
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5015387143512329,
      "learning_rate": 8.942667357521505e-06,
      "loss": 0.2724,
      "step": 3279
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.224459884286943,
      "learning_rate": 8.941954842478071e-06,
      "loss": 0.1929,
      "step": 3280
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3979436724812062,
      "learning_rate": 8.94124211584616e-06,
      "loss": 0.219,
      "step": 3281
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.125885975141405,
      "learning_rate": 8.940529177664032e-06,
      "loss": 0.5862,
      "step": 3282
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3555718616648973,
      "learning_rate": 8.939816027969952e-06,
      "loss": 0.2132,
      "step": 3283
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3012619267263932,
      "learning_rate": 8.9391026668022e-06,
      "loss": 0.2212,
      "step": 3284
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4156025698408934,
      "learning_rate": 8.938389094199065e-06,
      "loss": 0.2666,
      "step": 3285
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.770113895552139,
      "learning_rate": 8.93767531019885e-06,
      "loss": 0.7606,
      "step": 3286
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.384838394540693,
      "learning_rate": 8.936961314839869e-06,
      "loss": 0.5548,
      "step": 3287
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.7104024976241554,
      "learning_rate": 8.936247108160444e-06,
      "loss": 0.4813,
      "step": 3288
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3202732823741723,
      "learning_rate": 8.935532690198912e-06,
      "loss": 0.216,
      "step": 3289
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4286467883584035,
      "learning_rate": 8.934818060993622e-06,
      "loss": 0.2235,
      "step": 3290
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.437572706600563,
      "learning_rate": 8.934103220582929e-06,
      "loss": 0.5074,
      "step": 3291
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4469278564531205,
      "learning_rate": 8.933388169005205e-06,
      "loss": 0.2296,
      "step": 3292
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3784087009010173,
      "learning_rate": 8.93267290629883e-06,
      "loss": 0.2095,
      "step": 3293
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3863299577825723,
      "learning_rate": 8.931957432502198e-06,
      "loss": 0.1965,
      "step": 3294
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4180808522187207,
      "learning_rate": 8.931241747653711e-06,
      "loss": 0.205,
      "step": 3295
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.736384856857584,
      "learning_rate": 8.930525851791787e-06,
      "loss": 0.6212,
      "step": 3296
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.049262037118591,
      "learning_rate": 8.929809744954849e-06,
      "loss": 0.5757,
      "step": 3297
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.717374439126013,
      "learning_rate": 8.929093427181336e-06,
      "loss": 0.7092,
      "step": 3298
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.472558876669526,
      "learning_rate": 8.928376898509698e-06,
      "loss": 0.2056,
      "step": 3299
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4783267715022645,
      "learning_rate": 8.927660158978392e-06,
      "loss": 0.2225,
      "step": 3300
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.376954662635892,
      "learning_rate": 8.926943208625896e-06,
      "loss": 0.2319,
      "step": 3301
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5247660656231647,
      "learning_rate": 8.92622604749069e-06,
      "loss": 0.2208,
      "step": 3302
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.357453395494915,
      "learning_rate": 8.925508675611264e-06,
      "loss": 0.2316,
      "step": 3303
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3519253453378373,
      "learning_rate": 8.92479109302613e-06,
      "loss": 0.2177,
      "step": 3304
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4200443074562479,
      "learning_rate": 8.9240732997738e-06,
      "loss": 0.1984,
      "step": 3305
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3521377155229775,
      "learning_rate": 8.923355295892808e-06,
      "loss": 0.2172,
      "step": 3306
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4058697888257647,
      "learning_rate": 8.92263708142169e-06,
      "loss": 0.2038,
      "step": 3307
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.733111647857325,
      "learning_rate": 8.921918656398995e-06,
      "loss": 0.6956,
      "step": 3308
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3219963398035541,
      "learning_rate": 8.921200020863289e-06,
      "loss": 0.1989,
      "step": 3309
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.502684391473349,
      "learning_rate": 8.920481174853144e-06,
      "loss": 0.2224,
      "step": 3310
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3099365778310341,
      "learning_rate": 8.919762118407146e-06,
      "loss": 0.1821,
      "step": 3311
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3467531667799149,
      "learning_rate": 8.919042851563888e-06,
      "loss": 0.2041,
      "step": 3312
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.611390824704519,
      "learning_rate": 8.91832337436198e-06,
      "loss": 0.5028,
      "step": 3313
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3578157689251673,
      "learning_rate": 8.917603686840039e-06,
      "loss": 0.2835,
      "step": 3314
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.567274345310595,
      "learning_rate": 8.916883789036697e-06,
      "loss": 0.6578,
      "step": 3315
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3639732581018305,
      "learning_rate": 8.916163680990593e-06,
      "loss": 0.1818,
      "step": 3316
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.223463220252573,
      "learning_rate": 8.915443362740383e-06,
      "loss": 0.2021,
      "step": 3317
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3470178564671258,
      "learning_rate": 8.914722834324725e-06,
      "loss": 0.1793,
      "step": 3318
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4195466440564755,
      "learning_rate": 8.914002095782301e-06,
      "loss": 0.2106,
      "step": 3319
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4707969756725279,
      "learning_rate": 8.913281147151793e-06,
      "loss": 0.2352,
      "step": 3320
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.0827904437108975,
      "learning_rate": 8.912559988471899e-06,
      "loss": 0.4202,
      "step": 3321
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4672794497351787,
      "learning_rate": 8.911838619781328e-06,
      "loss": 0.2628,
      "step": 3322
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.32036419332378,
      "learning_rate": 8.911117041118803e-06,
      "loss": 0.2029,
      "step": 3323
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3099676791027253,
      "learning_rate": 8.910395252523053e-06,
      "loss": 0.202,
      "step": 3324
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3370408628498869,
      "learning_rate": 8.909673254032818e-06,
      "loss": 0.2099,
      "step": 3325
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4901321999777943,
      "learning_rate": 8.908951045686858e-06,
      "loss": 0.2331,
      "step": 3326
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5409107704084781,
      "learning_rate": 8.908228627523934e-06,
      "loss": 0.2058,
      "step": 3327
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2499717556253962,
      "learning_rate": 8.907505999582823e-06,
      "loss": 0.2271,
      "step": 3328
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5495622911412996,
      "learning_rate": 8.906783161902316e-06,
      "loss": 0.2139,
      "step": 3329
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.188244342530435,
      "learning_rate": 8.906060114521207e-06,
      "loss": 0.6686,
      "step": 3330
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5192486021521592,
      "learning_rate": 8.905336857478311e-06,
      "loss": 0.2679,
      "step": 3331
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2846258833251372,
      "learning_rate": 8.904613390812449e-06,
      "loss": 0.1586,
      "step": 3332
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.443928316235056,
      "learning_rate": 8.903889714562449e-06,
      "loss": 0.1837,
      "step": 3333
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2734650330900155,
      "learning_rate": 8.90316582876716e-06,
      "loss": 0.2363,
      "step": 3334
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3793810358070389,
      "learning_rate": 8.902441733465436e-06,
      "loss": 0.2155,
      "step": 3335
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.359097273332423,
      "learning_rate": 8.901717428696142e-06,
      "loss": 0.2322,
      "step": 3336
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4164999258689257,
      "learning_rate": 8.900992914498159e-06,
      "loss": 0.1832,
      "step": 3337
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3130728001093355,
      "learning_rate": 8.900268190910373e-06,
      "loss": 0.1762,
      "step": 3338
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.430435786055622,
      "learning_rate": 8.899543257971686e-06,
      "loss": 0.2217,
      "step": 3339
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5286284222806887,
      "learning_rate": 8.898818115721009e-06,
      "loss": 0.2095,
      "step": 3340
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2846597620659843,
      "learning_rate": 8.898092764197264e-06,
      "loss": 0.1841,
      "step": 3341
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.468208556285966,
      "learning_rate": 8.897367203439386e-06,
      "loss": 0.1916,
      "step": 3342
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5392152150071345,
      "learning_rate": 8.89664143348632e-06,
      "loss": 0.1649,
      "step": 3343
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4236655158906615,
      "learning_rate": 8.895915454377021e-06,
      "loss": 0.2132,
      "step": 3344
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6583361987145053,
      "learning_rate": 8.895189266150462e-06,
      "loss": 0.2067,
      "step": 3345
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4962372209160975,
      "learning_rate": 8.894462868845615e-06,
      "loss": 0.1987,
      "step": 3346
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3378925360554126,
      "learning_rate": 8.893736262501473e-06,
      "loss": 0.2316,
      "step": 3347
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.625326675998737,
      "learning_rate": 8.893009447157039e-06,
      "loss": 0.2326,
      "step": 3348
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4716486335374517,
      "learning_rate": 8.892282422851321e-06,
      "loss": 0.195,
      "step": 3349
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6032885833363117,
      "learning_rate": 8.891555189623348e-06,
      "loss": 0.2324,
      "step": 3350
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.427129227724154,
      "learning_rate": 8.890827747512155e-06,
      "loss": 0.1912,
      "step": 3351
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2777478492271028,
      "learning_rate": 8.890100096556782e-06,
      "loss": 0.1619,
      "step": 3352
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2665148255007546,
      "learning_rate": 8.889372236796292e-06,
      "loss": 0.1998,
      "step": 3353
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.266448138690524,
      "learning_rate": 8.888644168269751e-06,
      "loss": 0.5737,
      "step": 3354
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.48207732269246,
      "learning_rate": 8.88791589101624e-06,
      "loss": 0.7817,
      "step": 3355
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3590775535492565,
      "learning_rate": 8.887187405074852e-06,
      "loss": 0.2068,
      "step": 3356
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3326477877855927,
      "learning_rate": 8.886458710484684e-06,
      "loss": 0.214,
      "step": 3357
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4733779770393356,
      "learning_rate": 8.885729807284855e-06,
      "loss": 0.2076,
      "step": 3358
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4531546610230743,
      "learning_rate": 8.885000695514486e-06,
      "loss": 0.2564,
      "step": 3359
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3412051609871583,
      "learning_rate": 8.884271375212714e-06,
      "loss": 0.2443,
      "step": 3360
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.472158654976976,
      "learning_rate": 8.883541846418687e-06,
      "loss": 0.2399,
      "step": 3361
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2940120326712834,
      "learning_rate": 8.882812109171561e-06,
      "loss": 0.2358,
      "step": 3362
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1251913401646718,
      "learning_rate": 8.882082163510507e-06,
      "loss": 0.1874,
      "step": 3363
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4993218690469448,
      "learning_rate": 8.881352009474704e-06,
      "loss": 0.2439,
      "step": 3364
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2372557034541531,
      "learning_rate": 8.880621647103346e-06,
      "loss": 0.2236,
      "step": 3365
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2943184727519619,
      "learning_rate": 8.879891076435636e-06,
      "loss": 0.193,
      "step": 3366
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4587366929940457,
      "learning_rate": 8.879160297510785e-06,
      "loss": 0.2473,
      "step": 3367
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.877472926600581,
      "learning_rate": 8.878429310368022e-06,
      "loss": 0.5652,
      "step": 3368
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.510777669951402,
      "learning_rate": 8.87769811504658e-06,
      "loss": 0.2325,
      "step": 3369
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2286085051030744,
      "learning_rate": 8.876966711585711e-06,
      "loss": 0.2018,
      "step": 3370
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.381130732552162,
      "learning_rate": 8.876235100024668e-06,
      "loss": 0.1819,
      "step": 3371
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3573986351631284,
      "learning_rate": 8.875503280402727e-06,
      "loss": 0.1972,
      "step": 3372
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.680697514470033,
      "learning_rate": 8.874771252759164e-06,
      "loss": 0.6843,
      "step": 3373
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3950776241986975,
      "learning_rate": 8.874039017133276e-06,
      "loss": 0.2052,
      "step": 3374
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.510031245656418,
      "learning_rate": 8.873306573564364e-06,
      "loss": 0.2211,
      "step": 3375
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5140672736192597,
      "learning_rate": 8.872573922091743e-06,
      "loss": 0.2164,
      "step": 3376
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4161333050129712,
      "learning_rate": 8.871841062754737e-06,
      "loss": 0.1943,
      "step": 3377
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3117065041013924,
      "learning_rate": 8.871107995592687e-06,
      "loss": 0.206,
      "step": 3378
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5516234389644101,
      "learning_rate": 8.870374720644936e-06,
      "loss": 0.2486,
      "step": 3379
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5271342069400944,
      "learning_rate": 8.86964123795085e-06,
      "loss": 0.2344,
      "step": 3380
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3655716960852164,
      "learning_rate": 8.868907547549793e-06,
      "loss": 0.2191,
      "step": 3381
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3224030483687998,
      "learning_rate": 8.86817364948115e-06,
      "loss": 0.2011,
      "step": 3382
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.459774043191652,
      "learning_rate": 8.867439543784313e-06,
      "loss": 0.2237,
      "step": 3383
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.430488206111277,
      "learning_rate": 8.866705230498686e-06,
      "loss": 0.2403,
      "step": 3384
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5137601063390582,
      "learning_rate": 8.865970709663682e-06,
      "loss": 0.2259,
      "step": 3385
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.487166428897814,
      "learning_rate": 8.865235981318732e-06,
      "loss": 0.5559,
      "step": 3386
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.373772738971246,
      "learning_rate": 8.864501045503268e-06,
      "loss": 0.1924,
      "step": 3387
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5039193835440003,
      "learning_rate": 8.863765902256742e-06,
      "loss": 0.2222,
      "step": 3388
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2738683093008236,
      "learning_rate": 8.863030551618613e-06,
      "loss": 0.1886,
      "step": 3389
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.367442924436874,
      "learning_rate": 8.862294993628348e-06,
      "loss": 0.215,
      "step": 3390
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.551663676117261,
      "learning_rate": 8.861559228325433e-06,
      "loss": 0.232,
      "step": 3391
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3999192675201675,
      "learning_rate": 8.860823255749362e-06,
      "loss": 0.2101,
      "step": 3392
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.632011962390878,
      "learning_rate": 8.860087075939636e-06,
      "loss": 0.5768,
      "step": 3393
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4214775226450687,
      "learning_rate": 8.859350688935769e-06,
      "loss": 0.2165,
      "step": 3394
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4747611725371281,
      "learning_rate": 8.858614094777292e-06,
      "loss": 0.1972,
      "step": 3395
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3661368245966141,
      "learning_rate": 8.857877293503739e-06,
      "loss": 0.2336,
      "step": 3396
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.531646155795514,
      "learning_rate": 8.857140285154658e-06,
      "loss": 0.6816,
      "step": 3397
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4566310621944527,
      "learning_rate": 8.85640306976961e-06,
      "loss": 0.2398,
      "step": 3398
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4758597182222741,
      "learning_rate": 8.855665647388167e-06,
      "loss": 0.2151,
      "step": 3399
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.372055257216578,
      "learning_rate": 8.85492801804991e-06,
      "loss": 0.2229,
      "step": 3400
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3958624833322675,
      "learning_rate": 8.854190181794433e-06,
      "loss": 0.1906,
      "step": 3401
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3913785077084155,
      "learning_rate": 8.853452138661338e-06,
      "loss": 0.2214,
      "step": 3402
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5901520433175405,
      "learning_rate": 8.852713888690239e-06,
      "loss": 0.2964,
      "step": 3403
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4859277916241667,
      "learning_rate": 8.851975431920768e-06,
      "loss": 0.2441,
      "step": 3404
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.446421761868939,
      "learning_rate": 8.851236768392556e-06,
      "loss": 0.1935,
      "step": 3405
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3639746948539013,
      "learning_rate": 8.850497898145256e-06,
      "loss": 0.2109,
      "step": 3406
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.398064511201901,
      "learning_rate": 8.849758821218525e-06,
      "loss": 0.7553,
      "step": 3407
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3380611537756595,
      "learning_rate": 8.849019537652035e-06,
      "loss": 0.1734,
      "step": 3408
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3633752856538814,
      "learning_rate": 8.848280047485468e-06,
      "loss": 0.2514,
      "step": 3409
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1902055027075862,
      "learning_rate": 8.847540350758516e-06,
      "loss": 0.1932,
      "step": 3410
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5142521791319392,
      "learning_rate": 8.846800447510884e-06,
      "loss": 0.2642,
      "step": 3411
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4314296013668832,
      "learning_rate": 8.846060337782288e-06,
      "loss": 0.2242,
      "step": 3412
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4353990399411127,
      "learning_rate": 8.845320021612448e-06,
      "loss": 0.2143,
      "step": 3413
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.592767234174408,
      "learning_rate": 8.84457949904111e-06,
      "loss": 0.1966,
      "step": 3414
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3498548696683323,
      "learning_rate": 8.843838770108018e-06,
      "loss": 0.185,
      "step": 3415
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6315400671663203,
      "learning_rate": 8.843097834852929e-06,
      "loss": 0.216,
      "step": 3416
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7189920891855792,
      "learning_rate": 8.842356693315619e-06,
      "loss": 0.2362,
      "step": 3417
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3331114834417002,
      "learning_rate": 8.841615345535865e-06,
      "loss": 0.2149,
      "step": 3418
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.083665803095692,
      "learning_rate": 8.840873791553461e-06,
      "loss": 0.5762,
      "step": 3419
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3692720174880115,
      "learning_rate": 8.84013203140821e-06,
      "loss": 0.1802,
      "step": 3420
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.388798759333109,
      "learning_rate": 8.83939006513993e-06,
      "loss": 0.2108,
      "step": 3421
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5061799058694654,
      "learning_rate": 8.838647892788443e-06,
      "loss": 0.2296,
      "step": 3422
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.348011868078559,
      "learning_rate": 8.837905514393587e-06,
      "loss": 0.1465,
      "step": 3423
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.341854065561041,
      "learning_rate": 8.837162929995212e-06,
      "loss": 0.1952,
      "step": 3424
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.424947633923621,
      "learning_rate": 8.836420139633173e-06,
      "loss": 0.2096,
      "step": 3425
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3189746719081146,
      "learning_rate": 8.835677143347343e-06,
      "loss": 0.2267,
      "step": 3426
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.289344867034444,
      "learning_rate": 8.834933941177604e-06,
      "loss": 0.2576,
      "step": 3427
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3352373183998387,
      "learning_rate": 8.834190533163844e-06,
      "loss": 0.2333,
      "step": 3428
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4239783625387485,
      "learning_rate": 8.83344691934597e-06,
      "loss": 0.1843,
      "step": 3429
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.490486275697981,
      "learning_rate": 8.832703099763894e-06,
      "loss": 0.2041,
      "step": 3430
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3367185460177247,
      "learning_rate": 8.83195907445754e-06,
      "loss": 0.2076,
      "step": 3431
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2489545011586032,
      "learning_rate": 8.83121484346685e-06,
      "loss": 0.1763,
      "step": 3432
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5202513292352233,
      "learning_rate": 8.830470406831767e-06,
      "loss": 0.208,
      "step": 3433
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3826391280267434,
      "learning_rate": 8.82972576459225e-06,
      "loss": 0.2348,
      "step": 3434
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.420107967857565,
      "learning_rate": 8.828980916788269e-06,
      "loss": 0.2403,
      "step": 3435
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4183016175140915,
      "learning_rate": 8.828235863459801e-06,
      "loss": 0.2097,
      "step": 3436
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4181374571751888,
      "learning_rate": 8.827490604646845e-06,
      "loss": 0.2666,
      "step": 3437
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.337114228016009,
      "learning_rate": 8.826745140389397e-06,
      "loss": 0.2364,
      "step": 3438
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4444529161124489,
      "learning_rate": 8.825999470727473e-06,
      "loss": 0.2402,
      "step": 3439
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.377432885848545,
      "learning_rate": 8.825253595701097e-06,
      "loss": 0.2087,
      "step": 3440
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.42755092706177,
      "learning_rate": 8.824507515350304e-06,
      "loss": 0.2286,
      "step": 3441
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3310633888039973,
      "learning_rate": 8.823761229715144e-06,
      "loss": 0.1924,
      "step": 3442
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3212561112899646,
      "learning_rate": 8.82301473883567e-06,
      "loss": 0.1703,
      "step": 3443
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.597161415635961,
      "learning_rate": 8.822268042751956e-06,
      "loss": 0.2608,
      "step": 3444
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3073723924360938,
      "learning_rate": 8.821521141504075e-06,
      "loss": 0.2312,
      "step": 3445
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4643042870741099,
      "learning_rate": 8.820774035132122e-06,
      "loss": 0.1982,
      "step": 3446
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.6647324206655885,
      "learning_rate": 8.8200267236762e-06,
      "loss": 0.5116,
      "step": 3447
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2681844338721362,
      "learning_rate": 8.819279207176418e-06,
      "loss": 0.1913,
      "step": 3448
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3590712286494593,
      "learning_rate": 8.818531485672902e-06,
      "loss": 0.2607,
      "step": 3449
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.269856942907231,
      "learning_rate": 8.817783559205784e-06,
      "loss": 0.6841,
      "step": 3450
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.537286679057048,
      "learning_rate": 8.817035427815214e-06,
      "loss": 0.1967,
      "step": 3451
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5691797491446957,
      "learning_rate": 8.816287091541346e-06,
      "loss": 0.2444,
      "step": 3452
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.245704647399468,
      "learning_rate": 8.815538550424347e-06,
      "loss": 0.1844,
      "step": 3453
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2022984775696763,
      "learning_rate": 8.8147898045044e-06,
      "loss": 0.1851,
      "step": 3454
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.7347282375961,
      "learning_rate": 8.814040853821688e-06,
      "loss": 0.5588,
      "step": 3455
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5063059529951168,
      "learning_rate": 8.813291698416417e-06,
      "loss": 0.2428,
      "step": 3456
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.431643797650711,
      "learning_rate": 8.812542338328797e-06,
      "loss": 0.2319,
      "step": 3457
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.330074686441079,
      "learning_rate": 8.81179277359905e-06,
      "loss": 0.2666,
      "step": 3458
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1856959380764205,
      "learning_rate": 8.811043004267411e-06,
      "loss": 0.1953,
      "step": 3459
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3265089760846611,
      "learning_rate": 8.810293030374126e-06,
      "loss": 0.2326,
      "step": 3460
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.401036447124268,
      "learning_rate": 8.809542851959448e-06,
      "loss": 0.2372,
      "step": 3461
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2740193197283487,
      "learning_rate": 8.80879246906364e-06,
      "loss": 0.2115,
      "step": 3462
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2834841775473325,
      "learning_rate": 8.808041881726988e-06,
      "loss": 0.2202,
      "step": 3463
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3812463078224733,
      "learning_rate": 8.807291089989776e-06,
      "loss": 0.2374,
      "step": 3464
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3341292725043825,
      "learning_rate": 8.806540093892302e-06,
      "loss": 0.1854,
      "step": 3465
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3143854799170787,
      "learning_rate": 8.80578889347488e-06,
      "loss": 0.2616,
      "step": 3466
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3732514738130441,
      "learning_rate": 8.805037488777827e-06,
      "loss": 0.2282,
      "step": 3467
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4522727476685582,
      "learning_rate": 8.804285879841481e-06,
      "loss": 0.2021,
      "step": 3468
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3822539442064339,
      "learning_rate": 8.803534066706183e-06,
      "loss": 0.1906,
      "step": 3469
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.097700997986065,
      "learning_rate": 8.802782049412286e-06,
      "loss": 0.7269,
      "step": 3470
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4175055475082383,
      "learning_rate": 8.802029828000157e-06,
      "loss": 0.237,
      "step": 3471
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4099038172743792,
      "learning_rate": 8.801277402510169e-06,
      "loss": 0.2475,
      "step": 3472
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.148999200392265,
      "learning_rate": 8.800524772982712e-06,
      "loss": 0.5292,
      "step": 3473
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5572007957258802,
      "learning_rate": 8.799771939458186e-06,
      "loss": 0.1956,
      "step": 3474
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3784975137607194,
      "learning_rate": 8.799018901976997e-06,
      "loss": 0.2341,
      "step": 3475
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3444255154790965,
      "learning_rate": 8.798265660579567e-06,
      "loss": 0.1815,
      "step": 3476
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.058409218546766,
      "learning_rate": 8.797512215306325e-06,
      "loss": 0.5875,
      "step": 3477
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3672000201131083,
      "learning_rate": 8.796758566197714e-06,
      "loss": 0.2122,
      "step": 3478
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4561403041767271,
      "learning_rate": 8.796004713294188e-06,
      "loss": 0.2684,
      "step": 3479
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4520918203577586,
      "learning_rate": 8.795250656636207e-06,
      "loss": 0.2735,
      "step": 3480
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4306970925535512,
      "learning_rate": 8.794496396264252e-06,
      "loss": 0.2121,
      "step": 3481
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5271108907038105,
      "learning_rate": 8.793741932218802e-06,
      "loss": 0.2709,
      "step": 3482
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4997989167028307,
      "learning_rate": 8.79298726454036e-06,
      "loss": 0.2511,
      "step": 3483
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.319760205957678,
      "learning_rate": 8.792232393269428e-06,
      "loss": 0.2119,
      "step": 3484
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5213111616613597,
      "learning_rate": 8.791477318446527e-06,
      "loss": 0.2202,
      "step": 3485
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5133731246783466,
      "learning_rate": 8.790722040112188e-06,
      "loss": 0.1967,
      "step": 3486
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.240615219142938,
      "learning_rate": 8.789966558306948e-06,
      "loss": 0.1795,
      "step": 3487
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4681425346219017,
      "learning_rate": 8.78921087307136e-06,
      "loss": 0.2428,
      "step": 3488
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.613626956732254,
      "learning_rate": 8.788454984445987e-06,
      "loss": 0.1999,
      "step": 3489
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.318662714152473,
      "learning_rate": 8.7876988924714e-06,
      "loss": 0.1884,
      "step": 3490
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1983018161825039,
      "learning_rate": 8.786942597188184e-06,
      "loss": 0.1824,
      "step": 3491
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.5697451605491075,
      "learning_rate": 8.786186098636935e-06,
      "loss": 0.6824,
      "step": 3492
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.416611687143626,
      "learning_rate": 8.785429396858258e-06,
      "loss": 0.2142,
      "step": 3493
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.191946226057742,
      "learning_rate": 8.784672491892769e-06,
      "loss": 0.1896,
      "step": 3494
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.831688127119905,
      "learning_rate": 8.7839153837811e-06,
      "loss": 0.7268,
      "step": 3495
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.957040585244163,
      "learning_rate": 8.78315807256388e-06,
      "loss": 0.66,
      "step": 3496
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4176036084956738,
      "learning_rate": 8.782400558281768e-06,
      "loss": 0.2006,
      "step": 3497
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3765226812644953,
      "learning_rate": 8.781642840975419e-06,
      "loss": 0.1959,
      "step": 3498
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3748192531774663,
      "learning_rate": 8.780884920685507e-06,
      "loss": 0.2404,
      "step": 3499
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.482517065468356,
      "learning_rate": 8.780126797452713e-06,
      "loss": 0.2586,
      "step": 3500
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4693765640052256,
      "learning_rate": 8.779368471317731e-06,
      "loss": 0.2431,
      "step": 3501
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.521710983542316,
      "learning_rate": 8.778609942321263e-06,
      "loss": 0.2367,
      "step": 3502
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2999981556714992,
      "learning_rate": 8.777851210504025e-06,
      "loss": 0.1912,
      "step": 3503
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.513404657769363,
      "learning_rate": 8.777092275906743e-06,
      "loss": 0.2516,
      "step": 3504
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4413557320617025,
      "learning_rate": 8.776333138570153e-06,
      "loss": 0.2428,
      "step": 3505
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.488152051116762,
      "learning_rate": 8.775573798535002e-06,
      "loss": 0.1921,
      "step": 3506
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5782741742482813,
      "learning_rate": 8.77481425584205e-06,
      "loss": 0.2566,
      "step": 3507
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4682532410223015,
      "learning_rate": 8.774054510532064e-06,
      "loss": 0.2286,
      "step": 3508
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.241297650818194,
      "learning_rate": 8.773294562645826e-06,
      "loss": 0.2226,
      "step": 3509
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.371851516830798,
      "learning_rate": 8.772534412224128e-06,
      "loss": 0.1948,
      "step": 3510
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.933212082311394,
      "learning_rate": 8.77177405930777e-06,
      "loss": 0.5866,
      "step": 3511
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4595417380909443,
      "learning_rate": 8.771013503937563e-06,
      "loss": 0.2091,
      "step": 3512
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3787810166214245,
      "learning_rate": 8.770252746154334e-06,
      "loss": 0.2297,
      "step": 3513
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.244046545621324,
      "learning_rate": 8.769491785998914e-06,
      "loss": 0.4104,
      "step": 3514
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.55061552776363,
      "learning_rate": 8.768730623512152e-06,
      "loss": 0.2498,
      "step": 3515
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2535863797730447,
      "learning_rate": 8.767969258734903e-06,
      "loss": 0.1893,
      "step": 3516
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4915819811911442,
      "learning_rate": 8.767207691708032e-06,
      "loss": 0.2239,
      "step": 3517
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7525037171060562,
      "learning_rate": 8.76644592247242e-06,
      "loss": 0.2258,
      "step": 3518
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4875652090202376,
      "learning_rate": 8.765683951068952e-06,
      "loss": 0.2525,
      "step": 3519
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.306116531299578,
      "learning_rate": 8.764921777538533e-06,
      "loss": 0.2255,
      "step": 3520
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3299102580440942,
      "learning_rate": 8.764159401922068e-06,
      "loss": 0.2067,
      "step": 3521
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.531280167049909,
      "learning_rate": 8.763396824260482e-06,
      "loss": 0.2168,
      "step": 3522
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4439434464232521,
      "learning_rate": 8.762634044594704e-06,
      "loss": 0.2309,
      "step": 3523
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2875234620153186,
      "learning_rate": 8.761871062965679e-06,
      "loss": 0.2047,
      "step": 3524
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5089507667914641,
      "learning_rate": 8.761107879414362e-06,
      "loss": 0.2599,
      "step": 3525
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4437356679462925,
      "learning_rate": 8.760344493981714e-06,
      "loss": 0.2205,
      "step": 3526
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5392915086275918,
      "learning_rate": 8.759580906708714e-06,
      "loss": 0.2419,
      "step": 3527
    },
    {
      "epoch": 0.25,
      "grad_norm": 38.905439270924006,
      "learning_rate": 8.758817117636345e-06,
      "loss": 0.5083,
      "step": 3528
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5192325288704667,
      "learning_rate": 8.758053126805608e-06,
      "loss": 0.2059,
      "step": 3529
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4430968894093856,
      "learning_rate": 8.757288934257509e-06,
      "loss": 0.2689,
      "step": 3530
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3798408298197458,
      "learning_rate": 8.756524540033066e-06,
      "loss": 0.2031,
      "step": 3531
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3896759524097035,
      "learning_rate": 8.75575994417331e-06,
      "loss": 0.2395,
      "step": 3532
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5436255189263635,
      "learning_rate": 8.754995146719281e-06,
      "loss": 0.2554,
      "step": 3533
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4341038003876296,
      "learning_rate": 8.754230147712031e-06,
      "loss": 0.2222,
      "step": 3534
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.319592719778914,
      "learning_rate": 8.75346494719262e-06,
      "loss": 0.2154,
      "step": 3535
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.316911936261718,
      "learning_rate": 8.752699545202124e-06,
      "loss": 0.2106,
      "step": 3536
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.338061016540739,
      "learning_rate": 8.751933941781624e-06,
      "loss": 0.1823,
      "step": 3537
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4650919989159568,
      "learning_rate": 8.751168136972217e-06,
      "loss": 0.2157,
      "step": 3538
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5580500699771689,
      "learning_rate": 8.750402130815005e-06,
      "loss": 0.2473,
      "step": 3539
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4852966582571063,
      "learning_rate": 8.749635923351108e-06,
      "loss": 0.242,
      "step": 3540
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.632942814709578,
      "learning_rate": 8.748869514621649e-06,
      "loss": 0.1588,
      "step": 3541
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2845291303522697,
      "learning_rate": 8.74810290466777e-06,
      "loss": 0.1745,
      "step": 3542
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2507709830571156,
      "learning_rate": 8.747336093530617e-06,
      "loss": 0.1844,
      "step": 3543
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5686949868723838,
      "learning_rate": 8.74656908125135e-06,
      "loss": 0.2736,
      "step": 3544
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5002574777247166,
      "learning_rate": 8.745801867871138e-06,
      "loss": 0.2052,
      "step": 3545
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6505551768471391,
      "learning_rate": 8.745034453431165e-06,
      "loss": 0.2729,
      "step": 3546
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.1081811186792425,
      "learning_rate": 8.74426683797262e-06,
      "loss": 0.5648,
      "step": 3547
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.609825386496396,
      "learning_rate": 8.743499021536705e-06,
      "loss": 0.6667,
      "step": 3548
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3764682796928078,
      "learning_rate": 8.742731004164636e-06,
      "loss": 0.2186,
      "step": 3549
    },
    {
      "epoch": 0.25,
      "grad_norm": 12.094134099366917,
      "learning_rate": 8.741962785897634e-06,
      "loss": 0.6248,
      "step": 3550
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.602519796392337,
      "learning_rate": 8.741194366776937e-06,
      "loss": 0.6259,
      "step": 3551
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3250663290820373,
      "learning_rate": 8.74042574684379e-06,
      "loss": 0.2046,
      "step": 3552
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4679756514498974,
      "learning_rate": 8.739656926139448e-06,
      "loss": 0.1866,
      "step": 3553
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6160587197491099,
      "learning_rate": 8.738887904705179e-06,
      "loss": 0.2818,
      "step": 3554
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.262920849537033,
      "learning_rate": 8.738118682582262e-06,
      "loss": 0.1805,
      "step": 3555
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2790353290893421,
      "learning_rate": 8.737349259811984e-06,
      "loss": 0.1693,
      "step": 3556
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3421431264287502,
      "learning_rate": 8.736579636435645e-06,
      "loss": 0.1964,
      "step": 3557
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5073213647446664,
      "learning_rate": 8.735809812494557e-06,
      "loss": 0.234,
      "step": 3558
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4391649603308385,
      "learning_rate": 8.73503978803004e-06,
      "loss": 0.2187,
      "step": 3559
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.9150450311019975,
      "learning_rate": 8.734269563083424e-06,
      "loss": 0.5778,
      "step": 3560
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.363786761957576,
      "learning_rate": 8.733499137696054e-06,
      "loss": 0.2208,
      "step": 3561
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4334337370598316,
      "learning_rate": 8.732728511909283e-06,
      "loss": 0.2136,
      "step": 3562
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.358821096864716,
      "learning_rate": 8.731957685764474e-06,
      "loss": 0.21,
      "step": 3563
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3011599558892513,
      "learning_rate": 8.731186659303004e-06,
      "loss": 0.2087,
      "step": 3564
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4095166472010363,
      "learning_rate": 8.730415432566256e-06,
      "loss": 0.2044,
      "step": 3565
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5020613358153465,
      "learning_rate": 8.72964400559563e-06,
      "loss": 0.2468,
      "step": 3566
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3211988481952324,
      "learning_rate": 8.728872378432529e-06,
      "loss": 0.1804,
      "step": 3567
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.783224591029421,
      "learning_rate": 8.728100551118372e-06,
      "loss": 0.5846,
      "step": 3568
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3736052175835756,
      "learning_rate": 8.727328523694591e-06,
      "loss": 0.1927,
      "step": 3569
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6062623185466405,
      "learning_rate": 8.726556296202622e-06,
      "loss": 0.2628,
      "step": 3570
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4839620595685727,
      "learning_rate": 8.725783868683915e-06,
      "loss": 0.2434,
      "step": 3571
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3043290153160276,
      "learning_rate": 8.72501124117993e-06,
      "loss": 0.1823,
      "step": 3572
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4121528772844587,
      "learning_rate": 8.724238413732143e-06,
      "loss": 0.2251,
      "step": 3573
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3927881020252728,
      "learning_rate": 8.723465386382032e-06,
      "loss": 0.2016,
      "step": 3574
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2834058695364785,
      "learning_rate": 8.722692159171093e-06,
      "loss": 0.2115,
      "step": 3575
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5176101395336725,
      "learning_rate": 8.721918732140829e-06,
      "loss": 0.2324,
      "step": 3576
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4058202994508628,
      "learning_rate": 8.721145105332752e-06,
      "loss": 0.2002,
      "step": 3577
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.513263119324629,
      "learning_rate": 8.720371278788388e-06,
      "loss": 0.208,
      "step": 3578
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4361083208445464,
      "learning_rate": 8.719597252549277e-06,
      "loss": 0.2137,
      "step": 3579
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5770053410631122,
      "learning_rate": 8.71882302665696e-06,
      "loss": 0.2305,
      "step": 3580
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3455963477308486,
      "learning_rate": 8.718048601153e-06,
      "loss": 0.1903,
      "step": 3581
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5694022795297426,
      "learning_rate": 8.717273976078959e-06,
      "loss": 0.269,
      "step": 3582
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4442507099849007,
      "learning_rate": 8.71649915147642e-06,
      "loss": 0.23,
      "step": 3583
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2673091790275623,
      "learning_rate": 8.715724127386971e-06,
      "loss": 0.2193,
      "step": 3584
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.312924200354133,
      "learning_rate": 8.714948903852214e-06,
      "loss": 0.1711,
      "step": 3585
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3324604288521895,
      "learning_rate": 8.714173480913759e-06,
      "loss": 0.1964,
      "step": 3586
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.520153053960255,
      "learning_rate": 8.713397858613225e-06,
      "loss": 0.1928,
      "step": 3587
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4012050594272891,
      "learning_rate": 8.712622036992248e-06,
      "loss": 0.2133,
      "step": 3588
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4365188702349088,
      "learning_rate": 8.711846016092468e-06,
      "loss": 0.1973,
      "step": 3589
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5822814622006793,
      "learning_rate": 8.711069795955543e-06,
      "loss": 0.2402,
      "step": 3590
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.6111822559081395,
      "learning_rate": 8.710293376623132e-06,
      "loss": 0.7092,
      "step": 3591
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2479384625692431,
      "learning_rate": 8.709516758136913e-06,
      "loss": 0.1959,
      "step": 3592
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.448073854697756,
      "learning_rate": 8.708739940538573e-06,
      "loss": 0.1782,
      "step": 3593
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3371933544543355,
      "learning_rate": 8.707962923869806e-06,
      "loss": 0.1807,
      "step": 3594
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2765175158409978,
      "learning_rate": 8.707185708172322e-06,
      "loss": 0.2316,
      "step": 3595
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3677570962487962,
      "learning_rate": 8.706408293487833e-06,
      "loss": 0.221,
      "step": 3596
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4546737844569366,
      "learning_rate": 8.705630679858075e-06,
      "loss": 0.2258,
      "step": 3597
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5599792406944677,
      "learning_rate": 8.704852867324783e-06,
      "loss": 0.2349,
      "step": 3598
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.471731278074859,
      "learning_rate": 8.704074855929708e-06,
      "loss": 0.2492,
      "step": 3599
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.110888301431983,
      "learning_rate": 8.70329664571461e-06,
      "loss": 0.6398,
      "step": 3600
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.504036494668173,
      "learning_rate": 8.70251823672126e-06,
      "loss": 0.2173,
      "step": 3601
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2426616378513164,
      "learning_rate": 8.701739628991442e-06,
      "loss": 0.2061,
      "step": 3602
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5465672950536882,
      "learning_rate": 8.700960822566943e-06,
      "loss": 0.2459,
      "step": 3603
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.355312231546318,
      "learning_rate": 8.700181817489575e-06,
      "loss": 0.2552,
      "step": 3604
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.078648726272848,
      "learning_rate": 8.699402613801145e-06,
      "loss": 0.6406,
      "step": 3605
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2930906564775058,
      "learning_rate": 8.698623211543478e-06,
      "loss": 0.2034,
      "step": 3606
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3808716774983836,
      "learning_rate": 8.697843610758413e-06,
      "loss": 0.2015,
      "step": 3607
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4714266271669127,
      "learning_rate": 8.697063811487793e-06,
      "loss": 0.2068,
      "step": 3608
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3882492151808599,
      "learning_rate": 8.696283813773475e-06,
      "loss": 0.1994,
      "step": 3609
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3311433421636816,
      "learning_rate": 8.695503617657328e-06,
      "loss": 0.2012,
      "step": 3610
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2732421567547827,
      "learning_rate": 8.694723223181226e-06,
      "loss": 0.1993,
      "step": 3611
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2649671765374781,
      "learning_rate": 8.69394263038706e-06,
      "loss": 0.1857,
      "step": 3612
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3614550487609511,
      "learning_rate": 8.693161839316731e-06,
      "loss": 0.2375,
      "step": 3613
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2190247167958133,
      "learning_rate": 8.692380850012145e-06,
      "loss": 0.1748,
      "step": 3614
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5641275002238877,
      "learning_rate": 8.691599662515224e-06,
      "loss": 0.2282,
      "step": 3615
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3916844818149299,
      "learning_rate": 8.6908182768679e-06,
      "loss": 0.2516,
      "step": 3616
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.18483380444172,
      "learning_rate": 8.690036693112112e-06,
      "loss": 0.5917,
      "step": 3617
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.609677420408956,
      "learning_rate": 8.689254911289816e-06,
      "loss": 0.5138,
      "step": 3618
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4479070490845398,
      "learning_rate": 8.688472931442972e-06,
      "loss": 0.2426,
      "step": 3619
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5123666783855032,
      "learning_rate": 8.687690753613554e-06,
      "loss": 0.2158,
      "step": 3620
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4300847273683959,
      "learning_rate": 8.686908377843547e-06,
      "loss": 0.2301,
      "step": 3621
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3080294668417833,
      "learning_rate": 8.686125804174947e-06,
      "loss": 0.1772,
      "step": 3622
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4053402814324312,
      "learning_rate": 8.685343032649758e-06,
      "loss": 0.2181,
      "step": 3623
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.41175785513272,
      "learning_rate": 8.684560063309994e-06,
      "loss": 0.2017,
      "step": 3624
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3821512422026803,
      "learning_rate": 8.683776896197686e-06,
      "loss": 0.2071,
      "step": 3625
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3406631132475026,
      "learning_rate": 8.682993531354868e-06,
      "loss": 0.1721,
      "step": 3626
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4294988814561305,
      "learning_rate": 8.682209968823589e-06,
      "loss": 0.1923,
      "step": 3627
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4578902780291605,
      "learning_rate": 8.681426208645906e-06,
      "loss": 0.2414,
      "step": 3628
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.246032363260343,
      "learning_rate": 8.680642250863892e-06,
      "loss": 0.1851,
      "step": 3629
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.015773133168359,
      "learning_rate": 8.679858095519624e-06,
      "loss": 0.5859,
      "step": 3630
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4557244153862854,
      "learning_rate": 8.679073742655192e-06,
      "loss": 0.1992,
      "step": 3631
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.180161468560182,
      "learning_rate": 8.6782891923127e-06,
      "loss": 0.1967,
      "step": 3632
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4885763535034882,
      "learning_rate": 8.677504444534253e-06,
      "loss": 0.2484,
      "step": 3633
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3773598582017972,
      "learning_rate": 8.67671949936198e-06,
      "loss": 0.235,
      "step": 3634
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3469819475814317,
      "learning_rate": 8.675934356838012e-06,
      "loss": 0.2165,
      "step": 3635
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5024048532435206,
      "learning_rate": 8.67514901700449e-06,
      "loss": 0.2716,
      "step": 3636
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3458938941602556,
      "learning_rate": 8.674363479903569e-06,
      "loss": 0.2058,
      "step": 3637
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4001320233741732,
      "learning_rate": 8.673577745577414e-06,
      "loss": 0.1935,
      "step": 3638
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4269731070102218,
      "learning_rate": 8.6727918140682e-06,
      "loss": 0.2306,
      "step": 3639
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3699289145976898,
      "learning_rate": 8.672005685418115e-06,
      "loss": 0.1926,
      "step": 3640
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4512164382661474,
      "learning_rate": 8.671219359669349e-06,
      "loss": 0.2386,
      "step": 3641
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2605249228203246,
      "learning_rate": 8.670432836864115e-06,
      "loss": 0.1791,
      "step": 3642
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5438693182045768,
      "learning_rate": 8.669646117044627e-06,
      "loss": 0.2365,
      "step": 3643
    },
    {
      "epoch": 0.26,
      "grad_norm": 10.170209911578782,
      "learning_rate": 8.668859200253116e-06,
      "loss": 0.6741,
      "step": 3644
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.599932155038986,
      "learning_rate": 8.668072086531818e-06,
      "loss": 0.67,
      "step": 3645
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.410254857501086,
      "learning_rate": 8.667284775922982e-06,
      "loss": 0.2435,
      "step": 3646
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3536109860503263,
      "learning_rate": 8.666497268468869e-06,
      "loss": 0.2142,
      "step": 3647
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2680493215714415,
      "learning_rate": 8.66570956421175e-06,
      "loss": 0.2125,
      "step": 3648
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5084335267763196,
      "learning_rate": 8.664921663193904e-06,
      "loss": 0.2126,
      "step": 3649
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.528444591957655,
      "learning_rate": 8.664133565457623e-06,
      "loss": 0.2436,
      "step": 3650
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3133912476408447,
      "learning_rate": 8.663345271045207e-06,
      "loss": 0.1905,
      "step": 3651
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4326530666856625,
      "learning_rate": 8.662556779998974e-06,
      "loss": 0.1796,
      "step": 3652
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.552917381106129,
      "learning_rate": 8.661768092361241e-06,
      "loss": 0.7356,
      "step": 3653
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6112582334380086,
      "learning_rate": 8.660979208174345e-06,
      "loss": 0.2218,
      "step": 3654
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5935572663030553,
      "learning_rate": 8.66019012748063e-06,
      "loss": 0.2762,
      "step": 3655
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2720226479046026,
      "learning_rate": 8.659400850322451e-06,
      "loss": 0.1911,
      "step": 3656
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1992858958944708,
      "learning_rate": 8.65861137674217e-06,
      "loss": 0.1841,
      "step": 3657
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3055061879240084,
      "learning_rate": 8.657821706782166e-06,
      "loss": 0.1751,
      "step": 3658
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4092479659674235,
      "learning_rate": 8.657031840484825e-06,
      "loss": 0.2339,
      "step": 3659
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.620787342646206,
      "learning_rate": 8.656241777892544e-06,
      "loss": 0.5402,
      "step": 3660
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2306326391241493,
      "learning_rate": 8.655451519047729e-06,
      "loss": 0.2107,
      "step": 3661
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3136535876307098,
      "learning_rate": 8.654661063992799e-06,
      "loss": 0.2341,
      "step": 3662
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3190342215891908,
      "learning_rate": 8.653870412770182e-06,
      "loss": 0.2082,
      "step": 3663
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4399647918708154,
      "learning_rate": 8.653079565422318e-06,
      "loss": 0.2067,
      "step": 3664
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5822197544676122,
      "learning_rate": 8.652288521991656e-06,
      "loss": 0.2411,
      "step": 3665
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3809179381008752,
      "learning_rate": 8.651497282520654e-06,
      "loss": 0.2495,
      "step": 3666
    },
    {
      "epoch": 0.26,
      "grad_norm": 25.83720614132753,
      "learning_rate": 8.650705847051786e-06,
      "loss": 0.5793,
      "step": 3667
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6028630306315217,
      "learning_rate": 8.64991421562753e-06,
      "loss": 0.2486,
      "step": 3668
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.174026891905283,
      "learning_rate": 8.649122388290383e-06,
      "loss": 0.1731,
      "step": 3669
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3008370950526544,
      "learning_rate": 8.64833036508284e-06,
      "loss": 0.1766,
      "step": 3670
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4591698283135168,
      "learning_rate": 8.647538146047418e-06,
      "loss": 0.2722,
      "step": 3671
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.37903803953325,
      "learning_rate": 8.64674573122664e-06,
      "loss": 0.2429,
      "step": 3672
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4321179853388204,
      "learning_rate": 8.645953120663038e-06,
      "loss": 0.2259,
      "step": 3673
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3084897474827066,
      "learning_rate": 8.645160314399157e-06,
      "loss": 0.1617,
      "step": 3674
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2886844080760904,
      "learning_rate": 8.644367312477552e-06,
      "loss": 0.1928,
      "step": 3675
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4204960197469758,
      "learning_rate": 8.643574114940789e-06,
      "loss": 0.2532,
      "step": 3676
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4537882977779275,
      "learning_rate": 8.64278072183144e-06,
      "loss": 0.2447,
      "step": 3677
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2579570039529764,
      "learning_rate": 8.641987133192096e-06,
      "loss": 0.2121,
      "step": 3678
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2745975737245494,
      "learning_rate": 8.641193349065351e-06,
      "loss": 0.1722,
      "step": 3679
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2725434275095413,
      "learning_rate": 8.640399369493813e-06,
      "loss": 0.1685,
      "step": 3680
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3696478668492316,
      "learning_rate": 8.6396051945201e-06,
      "loss": 0.2366,
      "step": 3681
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4125743610898953,
      "learning_rate": 8.638810824186839e-06,
      "loss": 0.215,
      "step": 3682
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.264651651921664,
      "learning_rate": 8.638016258536668e-06,
      "loss": 0.157,
      "step": 3683
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3164292254863632,
      "learning_rate": 8.637221497612238e-06,
      "loss": 0.231,
      "step": 3684
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4171474106663915,
      "learning_rate": 8.636426541456208e-06,
      "loss": 0.2242,
      "step": 3685
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.25277934153627,
      "learning_rate": 8.635631390111248e-06,
      "loss": 0.2237,
      "step": 3686
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.181973643974844,
      "learning_rate": 8.634836043620038e-06,
      "loss": 0.1661,
      "step": 3687
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6979088828056528,
      "learning_rate": 8.63404050202527e-06,
      "loss": 0.2863,
      "step": 3688
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.39466882472605,
      "learning_rate": 8.633244765369648e-06,
      "loss": 0.2102,
      "step": 3689
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.755946383717729,
      "learning_rate": 8.632448833695878e-06,
      "loss": 0.6949,
      "step": 3690
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.641083154816055,
      "learning_rate": 8.631652707046686e-06,
      "loss": 0.5644,
      "step": 3691
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3395622409000634,
      "learning_rate": 8.630856385464805e-06,
      "loss": 0.2019,
      "step": 3692
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2762870830118762,
      "learning_rate": 8.630059868992978e-06,
      "loss": 0.2112,
      "step": 3693
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1789605845016022,
      "learning_rate": 8.629263157673958e-06,
      "loss": 0.2029,
      "step": 3694
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4282583141589658,
      "learning_rate": 8.628466251550509e-06,
      "loss": 0.2056,
      "step": 3695
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.434177671039088,
      "learning_rate": 8.627669150665407e-06,
      "loss": 0.2407,
      "step": 3696
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4519080142263323,
      "learning_rate": 8.626871855061438e-06,
      "loss": 0.19,
      "step": 3697
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.276463931167107,
      "learning_rate": 8.626074364781399e-06,
      "loss": 0.187,
      "step": 3698
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.243157383330987,
      "learning_rate": 8.62527667986809e-06,
      "loss": 0.1821,
      "step": 3699
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.447528236571992,
      "learning_rate": 8.624478800364332e-06,
      "loss": 0.6182,
      "step": 3700
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1717162837881208,
      "learning_rate": 8.623680726312953e-06,
      "loss": 0.1882,
      "step": 3701
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2427571157597763,
      "learning_rate": 8.622882457756787e-06,
      "loss": 0.199,
      "step": 3702
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3857724677518324,
      "learning_rate": 8.622083994738683e-06,
      "loss": 0.2161,
      "step": 3703
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3714289612765016,
      "learning_rate": 8.621285337301502e-06,
      "loss": 0.2465,
      "step": 3704
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.448574625903545,
      "learning_rate": 8.62048648548811e-06,
      "loss": 0.2557,
      "step": 3705
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.551615220918818,
      "learning_rate": 8.619687439341386e-06,
      "loss": 0.5106,
      "step": 3706
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1928727255980196,
      "learning_rate": 8.61888819890422e-06,
      "loss": 0.1928,
      "step": 3707
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.138206198845635,
      "learning_rate": 8.618088764219514e-06,
      "loss": 0.6126,
      "step": 3708
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4238892807408703,
      "learning_rate": 8.617289135330177e-06,
      "loss": 0.2019,
      "step": 3709
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.572065335113266,
      "learning_rate": 8.616489312279131e-06,
      "loss": 0.6101,
      "step": 3710
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.386879721499365,
      "learning_rate": 8.615689295109304e-06,
      "loss": 0.2759,
      "step": 3711
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.35698005446649,
      "learning_rate": 8.614889083863642e-06,
      "loss": 0.1763,
      "step": 3712
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6688673039010682,
      "learning_rate": 8.614088678585092e-06,
      "loss": 0.2658,
      "step": 3713
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3855174880737342,
      "learning_rate": 8.613288079316624e-06,
      "loss": 0.1845,
      "step": 3714
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3408652673693586,
      "learning_rate": 8.612487286101204e-06,
      "loss": 0.2083,
      "step": 3715
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3903959511207398,
      "learning_rate": 8.61168629898182e-06,
      "loss": 0.2563,
      "step": 3716
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4767600595492247,
      "learning_rate": 8.610885118001462e-06,
      "loss": 0.2308,
      "step": 3717
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4589461411253029,
      "learning_rate": 8.610083743203136e-06,
      "loss": 0.2081,
      "step": 3718
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.373707232171107,
      "learning_rate": 8.609282174629859e-06,
      "loss": 0.2257,
      "step": 3719
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3736717254822854,
      "learning_rate": 8.608480412324652e-06,
      "loss": 0.2303,
      "step": 3720
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.405861347386478,
      "learning_rate": 8.607678456330552e-06,
      "loss": 0.2539,
      "step": 3721
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4048214408132158,
      "learning_rate": 8.606876306690606e-06,
      "loss": 0.2004,
      "step": 3722
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.402885719250304,
      "learning_rate": 8.606073963447868e-06,
      "loss": 0.2296,
      "step": 3723
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.8891248047445655,
      "learning_rate": 8.605271426645408e-06,
      "loss": 0.2609,
      "step": 3724
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5029992380872932,
      "learning_rate": 8.604468696326299e-06,
      "loss": 0.291,
      "step": 3725
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2944434907898885,
      "learning_rate": 8.603665772533631e-06,
      "loss": 0.2098,
      "step": 3726
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7292187874233913,
      "learning_rate": 8.6028626553105e-06,
      "loss": 0.2723,
      "step": 3727
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3332799470322534,
      "learning_rate": 8.602059344700015e-06,
      "loss": 0.2403,
      "step": 3728
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2545676697316803,
      "learning_rate": 8.601255840745297e-06,
      "loss": 0.1767,
      "step": 3729
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.362349085751693,
      "learning_rate": 8.60045214348947e-06,
      "loss": 0.5523,
      "step": 3730
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2701766759145263,
      "learning_rate": 8.599648252975676e-06,
      "loss": 0.1965,
      "step": 3731
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3268017744971359,
      "learning_rate": 8.598844169247064e-06,
      "loss": 0.216,
      "step": 3732
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0730807262687505,
      "learning_rate": 8.598039892346797e-06,
      "loss": 0.2062,
      "step": 3733
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.319315926696157,
      "learning_rate": 8.597235422318041e-06,
      "loss": 0.2005,
      "step": 3734
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3362468249150978,
      "learning_rate": 8.596430759203979e-06,
      "loss": 0.1597,
      "step": 3735
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3671083974152785,
      "learning_rate": 8.595625903047801e-06,
      "loss": 0.2248,
      "step": 3736
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.311381333957636,
      "learning_rate": 8.59482085389271e-06,
      "loss": 0.2341,
      "step": 3737
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4146299287788768,
      "learning_rate": 8.594015611781917e-06,
      "loss": 0.2287,
      "step": 3738
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.608191484225012,
      "learning_rate": 8.593210176758644e-06,
      "loss": 0.5675,
      "step": 3739
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.231463584323397,
      "learning_rate": 8.592404548866123e-06,
      "loss": 0.8225,
      "step": 3740
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3022457972067576,
      "learning_rate": 8.5915987281476e-06,
      "loss": 0.183,
      "step": 3741
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.3679706087039785,
      "learning_rate": 8.590792714646323e-06,
      "loss": 0.5342,
      "step": 3742
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.508856184978745,
      "learning_rate": 8.58998650840556e-06,
      "loss": 0.1893,
      "step": 3743
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.319264800140163,
      "learning_rate": 8.589180109468584e-06,
      "loss": 0.1988,
      "step": 3744
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3207946987624866,
      "learning_rate": 8.588373517878678e-06,
      "loss": 0.1889,
      "step": 3745
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4605806029169857,
      "learning_rate": 8.587566733679137e-06,
      "loss": 0.1976,
      "step": 3746
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3846173005812834,
      "learning_rate": 8.586759756913268e-06,
      "loss": 0.2017,
      "step": 3747
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.351304120329717,
      "learning_rate": 8.585952587624382e-06,
      "loss": 0.2075,
      "step": 3748
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4661103954883927,
      "learning_rate": 8.58514522585581e-06,
      "loss": 0.2267,
      "step": 3749
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3720702358694754,
      "learning_rate": 8.584337671650884e-06,
      "loss": 0.258,
      "step": 3750
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3758286068019823,
      "learning_rate": 8.583529925052952e-06,
      "loss": 0.2424,
      "step": 3751
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5776159633652178,
      "learning_rate": 8.58272198610537e-06,
      "loss": 0.2213,
      "step": 3752
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3607929834835275,
      "learning_rate": 8.581913854851506e-06,
      "loss": 0.2355,
      "step": 3753
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4698516044360939,
      "learning_rate": 8.581105531334735e-06,
      "loss": 0.2059,
      "step": 3754
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4508342304163457,
      "learning_rate": 8.580297015598447e-06,
      "loss": 0.2491,
      "step": 3755
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4339251429230815,
      "learning_rate": 8.57948830768604e-06,
      "loss": 0.2089,
      "step": 3756
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.415239555826219,
      "learning_rate": 8.578679407640919e-06,
      "loss": 0.1714,
      "step": 3757
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2813124770634041,
      "learning_rate": 8.577870315506506e-06,
      "loss": 0.1666,
      "step": 3758
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4988480636735053,
      "learning_rate": 8.577061031326227e-06,
      "loss": 0.2331,
      "step": 3759
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.773984018679814,
      "learning_rate": 8.576251555143524e-06,
      "loss": 0.5543,
      "step": 3760
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4032698870309013,
      "learning_rate": 8.575441887001847e-06,
      "loss": 0.2414,
      "step": 3761
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5419698548067005,
      "learning_rate": 8.574632026944652e-06,
      "loss": 0.2585,
      "step": 3762
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.319388853697744,
      "learning_rate": 8.57382197501541e-06,
      "loss": 0.2225,
      "step": 3763
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.344821102616439,
      "learning_rate": 8.573011731257605e-06,
      "loss": 0.1944,
      "step": 3764
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3392339595516822,
      "learning_rate": 8.572201295714723e-06,
      "loss": 0.1805,
      "step": 3765
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5048278098849377,
      "learning_rate": 8.571390668430268e-06,
      "loss": 0.2611,
      "step": 3766
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.255493267859696,
      "learning_rate": 8.570579849447752e-06,
      "loss": 0.1779,
      "step": 3767
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5282593137374865,
      "learning_rate": 8.569768838810691e-06,
      "loss": 0.183,
      "step": 3768
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.622052437571073,
      "learning_rate": 8.568957636562622e-06,
      "loss": 0.2793,
      "step": 3769
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3267230791955964,
      "learning_rate": 8.568146242747089e-06,
      "loss": 0.2423,
      "step": 3770
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2599102267646691,
      "learning_rate": 8.567334657407637e-06,
      "loss": 0.1887,
      "step": 3771
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6804956079989428,
      "learning_rate": 8.566522880587836e-06,
      "loss": 0.2724,
      "step": 3772
    },
    {
      "epoch": 0.27,
      "grad_norm": 21.36524085662909,
      "learning_rate": 8.565710912331253e-06,
      "loss": 0.5501,
      "step": 3773
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2720655163227121,
      "learning_rate": 8.564898752681476e-06,
      "loss": 0.2139,
      "step": 3774
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3441824566373441,
      "learning_rate": 8.564086401682095e-06,
      "loss": 0.1782,
      "step": 3775
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4209196093178265,
      "learning_rate": 8.563273859376717e-06,
      "loss": 0.2276,
      "step": 3776
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3154652703848242,
      "learning_rate": 8.562461125808955e-06,
      "loss": 0.1879,
      "step": 3777
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3878768452410848,
      "learning_rate": 8.56164820102243e-06,
      "loss": 0.2146,
      "step": 3778
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.8679886368950696,
      "learning_rate": 8.56083508506078e-06,
      "loss": 0.4543,
      "step": 3779
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5101079599711065,
      "learning_rate": 8.56002177796765e-06,
      "loss": 0.253,
      "step": 3780
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.352043590167883,
      "learning_rate": 8.559208279786693e-06,
      "loss": 0.1802,
      "step": 3781
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5023990757872405,
      "learning_rate": 8.558394590561575e-06,
      "loss": 0.2388,
      "step": 3782
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4019970606974428,
      "learning_rate": 8.557580710335976e-06,
      "loss": 0.2222,
      "step": 3783
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4235292783458622,
      "learning_rate": 8.556766639153575e-06,
      "loss": 0.2383,
      "step": 3784
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.622177677072024,
      "learning_rate": 8.555952377058071e-06,
      "loss": 0.6352,
      "step": 3785
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.255773833922842,
      "learning_rate": 8.55513792409317e-06,
      "loss": 0.1712,
      "step": 3786
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4176935537783162,
      "learning_rate": 8.55432328030259e-06,
      "loss": 0.2251,
      "step": 3787
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2872261987272458,
      "learning_rate": 8.553508445730059e-06,
      "loss": 0.2025,
      "step": 3788
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2901634942000337,
      "learning_rate": 8.552693420419309e-06,
      "loss": 0.2289,
      "step": 3789
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.46746626514602,
      "learning_rate": 8.551878204414092e-06,
      "loss": 0.181,
      "step": 3790
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.296549302443851,
      "learning_rate": 8.551062797758162e-06,
      "loss": 0.228,
      "step": 3791
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.214529774209249,
      "learning_rate": 8.550247200495289e-06,
      "loss": 0.1886,
      "step": 3792
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3398654654605422,
      "learning_rate": 8.54943141266925e-06,
      "loss": 0.2401,
      "step": 3793
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4301536041908605,
      "learning_rate": 8.548615434323837e-06,
      "loss": 0.2369,
      "step": 3794
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4160362103673751,
      "learning_rate": 8.547799265502844e-06,
      "loss": 0.1808,
      "step": 3795
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6080012519209934,
      "learning_rate": 8.546982906250082e-06,
      "loss": 0.2171,
      "step": 3796
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5789634131715335,
      "learning_rate": 8.546166356609366e-06,
      "loss": 0.1684,
      "step": 3797
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3864888188164757,
      "learning_rate": 8.545349616624529e-06,
      "loss": 0.2195,
      "step": 3798
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2981800397987413,
      "learning_rate": 8.544532686339411e-06,
      "loss": 0.1681,
      "step": 3799
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3493144257686176,
      "learning_rate": 8.543715565797861e-06,
      "loss": 0.2276,
      "step": 3800
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.175495981955503,
      "learning_rate": 8.542898255043735e-06,
      "loss": 0.1414,
      "step": 3801
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.284617242951081,
      "learning_rate": 8.542080754120907e-06,
      "loss": 0.2513,
      "step": 3802
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4711552359422857,
      "learning_rate": 8.541263063073256e-06,
      "loss": 0.2724,
      "step": 3803
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.712727661415676,
      "learning_rate": 8.540445181944673e-06,
      "loss": 0.6565,
      "step": 3804
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2021653482011387,
      "learning_rate": 8.53962711077906e-06,
      "loss": 0.2064,
      "step": 3805
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.742796825476432,
      "learning_rate": 8.538808849620322e-06,
      "loss": 0.2273,
      "step": 3806
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2357270266781555,
      "learning_rate": 8.537990398512386e-06,
      "loss": 0.1718,
      "step": 3807
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3326391721760527,
      "learning_rate": 8.53717175749918e-06,
      "loss": 0.2318,
      "step": 3808
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.421780249659735,
      "learning_rate": 8.536352926624648e-06,
      "loss": 0.2467,
      "step": 3809
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4640046874206998,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.2336,
      "step": 3810
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4162658461869917,
      "learning_rate": 8.534714695467416e-06,
      "loss": 0.1892,
      "step": 3811
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2946165276881276,
      "learning_rate": 8.53389529527265e-06,
      "loss": 0.1864,
      "step": 3812
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3520973820619984,
      "learning_rate": 8.533075705392426e-06,
      "loss": 0.2101,
      "step": 3813
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.404255472017041,
      "learning_rate": 8.532255925870733e-06,
      "loss": 0.2708,
      "step": 3814
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5344333505848957,
      "learning_rate": 8.531435956751575e-06,
      "loss": 0.2146,
      "step": 3815
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5185512805731054,
      "learning_rate": 8.530615798078965e-06,
      "loss": 0.2285,
      "step": 3816
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2809346535080042,
      "learning_rate": 8.529795449896926e-06,
      "loss": 0.1944,
      "step": 3817
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2635046554670997,
      "learning_rate": 8.528974912249486e-06,
      "loss": 0.2213,
      "step": 3818
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5520902591388555,
      "learning_rate": 8.528154185180698e-06,
      "loss": 0.2765,
      "step": 3819
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4399614897339739,
      "learning_rate": 8.527333268734607e-06,
      "loss": 0.1828,
      "step": 3820
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.282272758232697,
      "learning_rate": 8.52651216295528e-06,
      "loss": 0.5017,
      "step": 3821
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5083858400591068,
      "learning_rate": 8.525690867886791e-06,
      "loss": 0.1753,
      "step": 3822
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4134577679998788,
      "learning_rate": 8.524869383573223e-06,
      "loss": 0.1769,
      "step": 3823
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3260585276355787,
      "learning_rate": 8.524047710058669e-06,
      "loss": 0.1849,
      "step": 3824
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3412548012624133,
      "learning_rate": 8.523225847387234e-06,
      "loss": 0.1809,
      "step": 3825
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.464972920229026,
      "learning_rate": 8.522403795603034e-06,
      "loss": 0.2464,
      "step": 3826
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4615100189581633,
      "learning_rate": 8.52158155475019e-06,
      "loss": 0.2194,
      "step": 3827
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4330635049973262,
      "learning_rate": 8.52075912487284e-06,
      "loss": 0.2399,
      "step": 3828
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4276667147227575,
      "learning_rate": 8.519936506015128e-06,
      "loss": 0.2479,
      "step": 3829
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5392773679120029,
      "learning_rate": 8.519113698221207e-06,
      "loss": 0.2757,
      "step": 3830
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.888178509053642,
      "learning_rate": 8.518290701535244e-06,
      "loss": 0.53,
      "step": 3831
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2449269402585341,
      "learning_rate": 8.517467516001413e-06,
      "loss": 0.1912,
      "step": 3832
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.464634182315767,
      "learning_rate": 8.516644141663902e-06,
      "loss": 0.233,
      "step": 3833
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.420062555196312,
      "learning_rate": 8.515820578566902e-06,
      "loss": 0.1994,
      "step": 3834
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3744203931568093,
      "learning_rate": 8.514996826754622e-06,
      "loss": 0.2382,
      "step": 3835
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2427035375392246,
      "learning_rate": 8.514172886271277e-06,
      "loss": 0.2193,
      "step": 3836
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1799765071536565,
      "learning_rate": 8.513348757161091e-06,
      "loss": 0.175,
      "step": 3837
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4261323105009915,
      "learning_rate": 8.512524439468304e-06,
      "loss": 0.239,
      "step": 3838
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5825680546863101,
      "learning_rate": 8.511699933237158e-06,
      "loss": 0.2139,
      "step": 3839
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3602769139965654,
      "learning_rate": 8.510875238511911e-06,
      "loss": 0.2345,
      "step": 3840
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.7587656928016475,
      "learning_rate": 8.510050355336828e-06,
      "loss": 0.5329,
      "step": 3841
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2989006186480576,
      "learning_rate": 8.509225283756188e-06,
      "loss": 0.2343,
      "step": 3842
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.399484368227784,
      "learning_rate": 8.508400023814277e-06,
      "loss": 0.1919,
      "step": 3843
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5173475415398716,
      "learning_rate": 8.50757457555539e-06,
      "loss": 0.2408,
      "step": 3844
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3936609217596883,
      "learning_rate": 8.506748939023835e-06,
      "loss": 0.2065,
      "step": 3845
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3999993171933802,
      "learning_rate": 8.505923114263927e-06,
      "loss": 0.2354,
      "step": 3846
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2728624786418958,
      "learning_rate": 8.505097101319995e-06,
      "loss": 0.2071,
      "step": 3847
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4683537834395277,
      "learning_rate": 8.504270900236376e-06,
      "loss": 0.245,
      "step": 3848
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.464388523583066,
      "learning_rate": 8.503444511057418e-06,
      "loss": 0.2549,
      "step": 3849
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2209656308717327,
      "learning_rate": 8.502617933827475e-06,
      "loss": 0.21,
      "step": 3850
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2676298409385023,
      "learning_rate": 8.501791168590918e-06,
      "loss": 0.282,
      "step": 3851
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3578261812017063,
      "learning_rate": 8.500964215392122e-06,
      "loss": 0.2441,
      "step": 3852
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3049482188584087,
      "learning_rate": 8.500137074275477e-06,
      "loss": 0.1885,
      "step": 3853
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2871134037657361,
      "learning_rate": 8.499309745285377e-06,
      "loss": 0.1702,
      "step": 3854
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2872539540030306,
      "learning_rate": 8.498482228466234e-06,
      "loss": 0.2068,
      "step": 3855
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.429545612292519,
      "learning_rate": 8.497654523862462e-06,
      "loss": 0.6066,
      "step": 3856
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3587465498432072,
      "learning_rate": 8.496826631518493e-06,
      "loss": 0.221,
      "step": 3857
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.29942182571122,
      "learning_rate": 8.495998551478758e-06,
      "loss": 0.2487,
      "step": 3858
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2344268686084001,
      "learning_rate": 8.495170283787714e-06,
      "loss": 0.2086,
      "step": 3859
    },
    {
      "epoch": 0.28,
      "grad_norm": 9.696172534643765,
      "learning_rate": 8.494341828489812e-06,
      "loss": 0.7293,
      "step": 3860
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4480304811526563,
      "learning_rate": 8.493513185629525e-06,
      "loss": 0.2508,
      "step": 3861
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.061888987038116,
      "learning_rate": 8.492684355251328e-06,
      "loss": 0.722,
      "step": 3862
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3827564700987485,
      "learning_rate": 8.49185533739971e-06,
      "loss": 0.2013,
      "step": 3863
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4893528144147292,
      "learning_rate": 8.491026132119172e-06,
      "loss": 0.2027,
      "step": 3864
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4478083118091474,
      "learning_rate": 8.49019673945422e-06,
      "loss": 0.2181,
      "step": 3865
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4115218975626034,
      "learning_rate": 8.489367159449372e-06,
      "loss": 0.1628,
      "step": 3866
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4403719132612645,
      "learning_rate": 8.488537392149156e-06,
      "loss": 0.2033,
      "step": 3867
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2623247936867024,
      "learning_rate": 8.487707437598114e-06,
      "loss": 0.1924,
      "step": 3868
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3373492947847219,
      "learning_rate": 8.486877295840792e-06,
      "loss": 0.2243,
      "step": 3869
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3100192214626967,
      "learning_rate": 8.48604696692175e-06,
      "loss": 0.2112,
      "step": 3870
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.014827214460886,
      "learning_rate": 8.485216450885555e-06,
      "loss": 0.641,
      "step": 3871
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3585463397851476,
      "learning_rate": 8.484385747776788e-06,
      "loss": 0.2012,
      "step": 3872
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.0348356668256145,
      "learning_rate": 8.483554857640036e-06,
      "loss": 0.6327,
      "step": 3873
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.45168813633487,
      "learning_rate": 8.482723780519899e-06,
      "loss": 0.7452,
      "step": 3874
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4578224311653685,
      "learning_rate": 8.481892516460986e-06,
      "loss": 0.2459,
      "step": 3875
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.472959944912008,
      "learning_rate": 8.481061065507915e-06,
      "loss": 0.2345,
      "step": 3876
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3325024143997988,
      "learning_rate": 8.480229427705316e-06,
      "loss": 0.2297,
      "step": 3877
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.331005411627358,
      "learning_rate": 8.479397603097827e-06,
      "loss": 0.2099,
      "step": 3878
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5341739560655858,
      "learning_rate": 8.478565591730096e-06,
      "loss": 0.237,
      "step": 3879
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.729469243657141,
      "learning_rate": 8.477733393646787e-06,
      "loss": 0.6163,
      "step": 3880
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3142664877759358,
      "learning_rate": 8.476901008892565e-06,
      "loss": 0.2142,
      "step": 3881
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2290010762016657,
      "learning_rate": 8.476068437512108e-06,
      "loss": 0.2068,
      "step": 3882
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.283988312428219,
      "learning_rate": 8.475235679550107e-06,
      "loss": 0.201,
      "step": 3883
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.690341955628392,
      "learning_rate": 8.474402735051263e-06,
      "loss": 0.5353,
      "step": 3884
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2425303793971396,
      "learning_rate": 8.473569604060282e-06,
      "loss": 0.1889,
      "step": 3885
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.772102796295855,
      "learning_rate": 8.472736286621885e-06,
      "loss": 0.6061,
      "step": 3886
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3859004863526245,
      "learning_rate": 8.4719027827808e-06,
      "loss": 0.1711,
      "step": 3887
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3036121916582706,
      "learning_rate": 8.471069092581768e-06,
      "loss": 0.1641,
      "step": 3888
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4343413663823026,
      "learning_rate": 8.470235216069536e-06,
      "loss": 0.2033,
      "step": 3889
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.400341220731093,
      "learning_rate": 8.469401153288864e-06,
      "loss": 0.1985,
      "step": 3890
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4007034078501162,
      "learning_rate": 8.468566904284522e-06,
      "loss": 0.1698,
      "step": 3891
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.841480205283238,
      "learning_rate": 8.467732469101286e-06,
      "loss": 0.637,
      "step": 3892
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3687822350774363,
      "learning_rate": 8.46689784778395e-06,
      "loss": 0.1963,
      "step": 3893
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2347683492206658,
      "learning_rate": 8.46606304037731e-06,
      "loss": 0.1626,
      "step": 3894
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3249510926380665,
      "learning_rate": 8.465228046926176e-06,
      "loss": 0.1445,
      "step": 3895
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5296602802178285,
      "learning_rate": 8.464392867475367e-06,
      "loss": 0.2097,
      "step": 3896
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2198063049719832,
      "learning_rate": 8.463557502069713e-06,
      "loss": 0.1823,
      "step": 3897
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5790548855219289,
      "learning_rate": 8.462721950754051e-06,
      "loss": 0.237,
      "step": 3898
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.429830998163152,
      "learning_rate": 8.461886213573233e-06,
      "loss": 0.2051,
      "step": 3899
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3713118195382008,
      "learning_rate": 8.461050290572114e-06,
      "loss": 0.1962,
      "step": 3900
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4333873339597056,
      "learning_rate": 8.460214181795567e-06,
      "loss": 0.2371,
      "step": 3901
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.445855041528906,
      "learning_rate": 8.45937788728847e-06,
      "loss": 0.202,
      "step": 3902
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.373078371649321,
      "learning_rate": 8.458541407095711e-06,
      "loss": 0.215,
      "step": 3903
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.448175548528807,
      "learning_rate": 8.45770474126219e-06,
      "loss": 0.2797,
      "step": 3904
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1102170609217346,
      "learning_rate": 8.456867889832814e-06,
      "loss": 0.1593,
      "step": 3905
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.259627632966343,
      "learning_rate": 8.456030852852503e-06,
      "loss": 0.6566,
      "step": 3906
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3658408605205474,
      "learning_rate": 8.455193630366186e-06,
      "loss": 0.2265,
      "step": 3907
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2570739895655239,
      "learning_rate": 8.454356222418802e-06,
      "loss": 0.1879,
      "step": 3908
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1864070260657922,
      "learning_rate": 8.453518629055299e-06,
      "loss": 0.1758,
      "step": 3909
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3751739745230591,
      "learning_rate": 8.452680850320636e-06,
      "loss": 0.1963,
      "step": 3910
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3987720685250922,
      "learning_rate": 8.451842886259783e-06,
      "loss": 0.2156,
      "step": 3911
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5015606420035643,
      "learning_rate": 8.451004736917716e-06,
      "loss": 0.2151,
      "step": 3912
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3559298919557874,
      "learning_rate": 8.450166402339426e-06,
      "loss": 0.1961,
      "step": 3913
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4349465025694126,
      "learning_rate": 8.449327882569909e-06,
      "loss": 0.2584,
      "step": 3914
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6226511532980332,
      "learning_rate": 8.448489177654175e-06,
      "loss": 0.2284,
      "step": 3915
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.326578529274266,
      "learning_rate": 8.447650287637244e-06,
      "loss": 0.1839,
      "step": 3916
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2379393649266521,
      "learning_rate": 8.446811212564138e-06,
      "loss": 0.1666,
      "step": 3917
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2820510903813804,
      "learning_rate": 8.445971952479903e-06,
      "loss": 0.2055,
      "step": 3918
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2197137820686106,
      "learning_rate": 8.445132507429584e-06,
      "loss": 0.1793,
      "step": 3919
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3335263545967568,
      "learning_rate": 8.444292877458238e-06,
      "loss": 0.2609,
      "step": 3920
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3724820740311976,
      "learning_rate": 8.443453062610933e-06,
      "loss": 0.2073,
      "step": 3921
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.507924337360983,
      "learning_rate": 8.442613062932747e-06,
      "loss": 0.2397,
      "step": 3922
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3178126037265236,
      "learning_rate": 8.44177287846877e-06,
      "loss": 0.2052,
      "step": 3923
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.128395941676882,
      "learning_rate": 8.440932509264099e-06,
      "loss": 0.5071,
      "step": 3924
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3306959169845087,
      "learning_rate": 8.440091955363841e-06,
      "loss": 0.1978,
      "step": 3925
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5267136285115483,
      "learning_rate": 8.439251216813111e-06,
      "loss": 0.2397,
      "step": 3926
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4499652822798768,
      "learning_rate": 8.438410293657042e-06,
      "loss": 0.2243,
      "step": 3927
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3330598220414203,
      "learning_rate": 8.437569185940766e-06,
      "loss": 0.2292,
      "step": 3928
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1362551629083406,
      "learning_rate": 8.436727893709435e-06,
      "loss": 0.182,
      "step": 3929
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.403654715046327,
      "learning_rate": 8.435886417008203e-06,
      "loss": 0.2238,
      "step": 3930
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3015050504774932,
      "learning_rate": 8.43504475588224e-06,
      "loss": 0.209,
      "step": 3931
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3913390229002132,
      "learning_rate": 8.434202910376719e-06,
      "loss": 0.1967,
      "step": 3932
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3760598461389233,
      "learning_rate": 8.433360880536829e-06,
      "loss": 0.2246,
      "step": 3933
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.535414410452314,
      "learning_rate": 8.432518666407767e-06,
      "loss": 0.5979,
      "step": 3934
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.589676027287786,
      "learning_rate": 8.43167626803474e-06,
      "loss": 0.2068,
      "step": 3935
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.549994991659285,
      "learning_rate": 8.430833685462965e-06,
      "loss": 0.5945,
      "step": 3936
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4517409351147652,
      "learning_rate": 8.42999091873767e-06,
      "loss": 0.204,
      "step": 3937
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.426627335584929,
      "learning_rate": 8.429147967904085e-06,
      "loss": 0.5697,
      "step": 3938
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5473129845939642,
      "learning_rate": 8.428304833007462e-06,
      "loss": 0.2538,
      "step": 3939
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3095140515001482,
      "learning_rate": 8.427461514093056e-06,
      "loss": 0.2514,
      "step": 3940
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4118972722141407,
      "learning_rate": 8.426618011206132e-06,
      "loss": 0.2364,
      "step": 3941
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.254087021132135,
      "learning_rate": 8.425774324391966e-06,
      "loss": 0.1934,
      "step": 3942
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3343530492923716,
      "learning_rate": 8.424930453695844e-06,
      "loss": 0.2093,
      "step": 3943
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4404064392691347,
      "learning_rate": 8.424086399163063e-06,
      "loss": 0.2047,
      "step": 3944
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.103972321061653,
      "learning_rate": 8.423242160838927e-06,
      "loss": 0.6356,
      "step": 3945
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4499695427441652,
      "learning_rate": 8.42239773876875e-06,
      "loss": 0.2039,
      "step": 3946
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2965724784268746,
      "learning_rate": 8.42155313299786e-06,
      "loss": 0.2264,
      "step": 3947
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.358675380966728,
      "learning_rate": 8.42070834357159e-06,
      "loss": 0.2468,
      "step": 3948
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.417970703054247,
      "learning_rate": 8.419863370535287e-06,
      "loss": 0.1817,
      "step": 3949
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4114583450157883,
      "learning_rate": 8.419018213934305e-06,
      "loss": 0.1947,
      "step": 3950
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4615083198859138,
      "learning_rate": 8.418172873814008e-06,
      "loss": 0.2651,
      "step": 3951
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3036228315040475,
      "learning_rate": 8.41732735021977e-06,
      "loss": 0.2003,
      "step": 3952
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4132115684762423,
      "learning_rate": 8.416481643196977e-06,
      "loss": 0.2499,
      "step": 3953
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3731397025275016,
      "learning_rate": 8.415635752791022e-06,
      "loss": 0.191,
      "step": 3954
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3145367733790236,
      "learning_rate": 8.414789679047307e-06,
      "loss": 0.221,
      "step": 3955
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4392843679164615,
      "learning_rate": 8.413943422011252e-06,
      "loss": 0.2514,
      "step": 3956
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3917665737365188,
      "learning_rate": 8.413096981728275e-06,
      "loss": 0.2377,
      "step": 3957
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.426169340192788,
      "learning_rate": 8.412250358243814e-06,
      "loss": 0.1959,
      "step": 3958
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3286828057417015,
      "learning_rate": 8.411403551603307e-06,
      "loss": 0.1512,
      "step": 3959
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.969693749666775,
      "learning_rate": 8.410556561852212e-06,
      "loss": 0.5898,
      "step": 3960
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5039687568481517,
      "learning_rate": 8.40970938903599e-06,
      "loss": 0.2345,
      "step": 3961
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.000501851715352,
      "learning_rate": 8.408862033200113e-06,
      "loss": 0.6636,
      "step": 3962
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.369195458880927,
      "learning_rate": 8.408014494390065e-06,
      "loss": 0.2314,
      "step": 3963
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2710644482352162,
      "learning_rate": 8.407166772651339e-06,
      "loss": 0.1675,
      "step": 3964
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3830886185178681,
      "learning_rate": 8.406318868029436e-06,
      "loss": 0.2041,
      "step": 3965
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3826428304776857,
      "learning_rate": 8.405470780569869e-06,
      "loss": 0.2607,
      "step": 3966
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.0361764970270695,
      "learning_rate": 8.40462251031816e-06,
      "loss": 0.5131,
      "step": 3967
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.308915712136961,
      "learning_rate": 8.40377405731984e-06,
      "loss": 0.2263,
      "step": 3968
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3666572708447056,
      "learning_rate": 8.402925421620452e-06,
      "loss": 0.2252,
      "step": 3969
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3411335296611422,
      "learning_rate": 8.402076603265543e-06,
      "loss": 0.2613,
      "step": 3970
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3195904899125672,
      "learning_rate": 8.40122760230068e-06,
      "loss": 0.2104,
      "step": 3971
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4385890010815705,
      "learning_rate": 8.400378418771434e-06,
      "loss": 0.2371,
      "step": 3972
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.533849862908287,
      "learning_rate": 8.39952905272338e-06,
      "loss": 0.6887,
      "step": 3973
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4129014984986605,
      "learning_rate": 8.398679504202114e-06,
      "loss": 0.2268,
      "step": 3974
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2877143380856004,
      "learning_rate": 8.397829773253234e-06,
      "loss": 0.2112,
      "step": 3975
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5128662918305495,
      "learning_rate": 8.396979859922351e-06,
      "loss": 0.2451,
      "step": 3976
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3974072800413495,
      "learning_rate": 8.396129764255086e-06,
      "loss": 0.2172,
      "step": 3977
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2761549189162609,
      "learning_rate": 8.395279486297065e-06,
      "loss": 0.1825,
      "step": 3978
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4451198547929118,
      "learning_rate": 8.39442902609393e-06,
      "loss": 0.2595,
      "step": 3979
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4259788487331106,
      "learning_rate": 8.39357838369133e-06,
      "loss": 0.2161,
      "step": 3980
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4916417807469637,
      "learning_rate": 8.392727559134926e-06,
      "loss": 0.2029,
      "step": 3981
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3259760193844965,
      "learning_rate": 8.391876552470383e-06,
      "loss": 0.2248,
      "step": 3982
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.331501075800485,
      "learning_rate": 8.391025363743383e-06,
      "loss": 0.1621,
      "step": 3983
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3001003618940754,
      "learning_rate": 8.390173992999613e-06,
      "loss": 0.2164,
      "step": 3984
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.449375766916586,
      "learning_rate": 8.389322440284773e-06,
      "loss": 0.1974,
      "step": 3985
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5474298612342725,
      "learning_rate": 8.388470705644569e-06,
      "loss": 0.2091,
      "step": 3986
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5474226228068124,
      "learning_rate": 8.387618789124718e-06,
      "loss": 0.2486,
      "step": 3987
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3917040139502261,
      "learning_rate": 8.386766690770948e-06,
      "loss": 0.2056,
      "step": 3988
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4054707048144732,
      "learning_rate": 8.385914410629e-06,
      "loss": 0.216,
      "step": 3989
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.2084614527088675,
      "learning_rate": 8.385061948744615e-06,
      "loss": 0.5791,
      "step": 3990
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3425730483106462,
      "learning_rate": 8.384209305163555e-06,
      "loss": 0.2164,
      "step": 3991
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5519198070654316,
      "learning_rate": 8.383356479931583e-06,
      "loss": 0.2659,
      "step": 3992
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.553094479680816,
      "learning_rate": 8.382503473094477e-06,
      "loss": 0.6156,
      "step": 3993
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4812023814302289,
      "learning_rate": 8.381650284698022e-06,
      "loss": 0.2035,
      "step": 3994
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3103359528506466,
      "learning_rate": 8.380796914788014e-06,
      "loss": 0.2329,
      "step": 3995
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3242123138762312,
      "learning_rate": 8.379943363410259e-06,
      "loss": 0.217,
      "step": 3996
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5208913474887478,
      "learning_rate": 8.379089630610573e-06,
      "loss": 0.2829,
      "step": 3997
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.950024749502722,
      "learning_rate": 8.37823571643478e-06,
      "loss": 0.5987,
      "step": 3998
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6180169697676572,
      "learning_rate": 8.377381620928713e-06,
      "loss": 0.2772,
      "step": 3999
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.959187466917736,
      "learning_rate": 8.376527344138222e-06,
      "loss": 0.6952,
      "step": 4000
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4346031586590502,
      "learning_rate": 8.375672886109152e-06,
      "loss": 0.2078,
      "step": 4001
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2049168639495935,
      "learning_rate": 8.374818246887376e-06,
      "loss": 0.1918,
      "step": 4002
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2708455010806656,
      "learning_rate": 8.373963426518764e-06,
      "loss": 0.2081,
      "step": 4003
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3758884523250772,
      "learning_rate": 8.373108425049198e-06,
      "loss": 0.1992,
      "step": 4004
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5098540920827337,
      "learning_rate": 8.372253242524573e-06,
      "loss": 0.2115,
      "step": 4005
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4306602309199932,
      "learning_rate": 8.37139787899079e-06,
      "loss": 0.2435,
      "step": 4006
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2386801952455295,
      "learning_rate": 8.370542334493764e-06,
      "loss": 0.1873,
      "step": 4007
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4603135799433573,
      "learning_rate": 8.369686609079416e-06,
      "loss": 0.2319,
      "step": 4008
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4519361273828328,
      "learning_rate": 8.368830702793678e-06,
      "loss": 0.1775,
      "step": 4009
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.51288341908173,
      "learning_rate": 8.36797461568249e-06,
      "loss": 0.48,
      "step": 4010
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.411305600834009,
      "learning_rate": 8.367118347791807e-06,
      "loss": 0.2236,
      "step": 4011
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.388958296412984,
      "learning_rate": 8.366261899167588e-06,
      "loss": 0.1782,
      "step": 4012
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3416384476774192,
      "learning_rate": 8.365405269855801e-06,
      "loss": 0.1475,
      "step": 4013
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3328969301100053,
      "learning_rate": 8.364548459902431e-06,
      "loss": 0.2147,
      "step": 4014
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4370870661313566,
      "learning_rate": 8.363691469353465e-06,
      "loss": 0.2139,
      "step": 4015
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2281535786795532,
      "learning_rate": 8.362834298254906e-06,
      "loss": 0.1752,
      "step": 4016
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3232472634924308,
      "learning_rate": 8.361976946652761e-06,
      "loss": 0.1694,
      "step": 4017
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4420436049989904,
      "learning_rate": 8.361119414593048e-06,
      "loss": 0.1826,
      "step": 4018
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2933491624861813,
      "learning_rate": 8.360261702121802e-06,
      "loss": 0.1806,
      "step": 4019
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4301124568227406,
      "learning_rate": 8.359403809285054e-06,
      "loss": 0.2127,
      "step": 4020
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2892218668830984,
      "learning_rate": 8.358545736128856e-06,
      "loss": 0.1963,
      "step": 4021
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2638291220447115,
      "learning_rate": 8.357687482699268e-06,
      "loss": 0.191,
      "step": 4022
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.281366995641839,
      "learning_rate": 8.356829049042352e-06,
      "loss": 0.2055,
      "step": 4023
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4527996120495386,
      "learning_rate": 8.355970435204192e-06,
      "loss": 0.2383,
      "step": 4024
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2730141165369886,
      "learning_rate": 8.355111641230869e-06,
      "loss": 0.1957,
      "step": 4025
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.5983967912750545,
      "learning_rate": 8.354252667168483e-06,
      "loss": 0.4839,
      "step": 4026
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2326326377365737,
      "learning_rate": 8.353393513063141e-06,
      "loss": 0.1788,
      "step": 4027
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.177843654907134,
      "learning_rate": 8.352534178960957e-06,
      "loss": 0.5307,
      "step": 4028
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.550671996642033,
      "learning_rate": 8.351674664908055e-06,
      "loss": 0.2046,
      "step": 4029
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.501404411655909,
      "learning_rate": 8.350814970950576e-06,
      "loss": 0.226,
      "step": 4030
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4630369680517794,
      "learning_rate": 8.349955097134658e-06,
      "loss": 0.2063,
      "step": 4031
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.32373364183077,
      "learning_rate": 8.34909504350646e-06,
      "loss": 0.2391,
      "step": 4032
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2710467936835261,
      "learning_rate": 8.348234810112147e-06,
      "loss": 0.1915,
      "step": 4033
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.403759358507858,
      "learning_rate": 8.347374396997891e-06,
      "loss": 0.2808,
      "step": 4034
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3486054279354456,
      "learning_rate": 8.346513804209873e-06,
      "loss": 0.2261,
      "step": 4035
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3535582327268165,
      "learning_rate": 8.345653031794292e-06,
      "loss": 0.2399,
      "step": 4036
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.31878277785827,
      "learning_rate": 8.344792079797346e-06,
      "loss": 0.6265,
      "step": 4037
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4701425065115221,
      "learning_rate": 8.343930948265252e-06,
      "loss": 0.2574,
      "step": 4038
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5054124948717251,
      "learning_rate": 8.343069637244225e-06,
      "loss": 0.2032,
      "step": 4039
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3014033636689692,
      "learning_rate": 8.342208146780504e-06,
      "loss": 0.1629,
      "step": 4040
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4758254822502992,
      "learning_rate": 8.341346476920327e-06,
      "loss": 0.2242,
      "step": 4041
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4187252546423248,
      "learning_rate": 8.340484627709946e-06,
      "loss": 0.2029,
      "step": 4042
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5190637174147896,
      "learning_rate": 8.33962259919562e-06,
      "loss": 0.2247,
      "step": 4043
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3729387479960766,
      "learning_rate": 8.338760391423623e-06,
      "loss": 0.2412,
      "step": 4044
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4997378865254147,
      "learning_rate": 8.337898004440231e-06,
      "loss": 0.265,
      "step": 4045
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.380536199730992,
      "learning_rate": 8.337035438291736e-06,
      "loss": 0.179,
      "step": 4046
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1338894273050752,
      "learning_rate": 8.336172693024434e-06,
      "loss": 0.1637,
      "step": 4047
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4326135358289709,
      "learning_rate": 8.335309768684637e-06,
      "loss": 0.2251,
      "step": 4048
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.932507727142624,
      "learning_rate": 8.334446665318661e-06,
      "loss": 0.4825,
      "step": 4049
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.233574299792511,
      "learning_rate": 8.333583382972834e-06,
      "loss": 0.1734,
      "step": 4050
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5577222441950687,
      "learning_rate": 8.332719921693497e-06,
      "loss": 0.2392,
      "step": 4051
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.342977682785485,
      "learning_rate": 8.331856281526994e-06,
      "loss": 0.1966,
      "step": 4052
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5102325061533506,
      "learning_rate": 8.330992462519683e-06,
      "loss": 0.17,
      "step": 4053
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3776971260277344,
      "learning_rate": 8.33012846471793e-06,
      "loss": 0.1949,
      "step": 4054
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3187182366377714,
      "learning_rate": 8.329264288168112e-06,
      "loss": 0.2348,
      "step": 4055
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4538909032723915,
      "learning_rate": 8.328399932916612e-06,
      "loss": 0.1979,
      "step": 4056
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5400064914165799,
      "learning_rate": 8.327535399009825e-06,
      "loss": 0.228,
      "step": 4057
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3382103896943622,
      "learning_rate": 8.326670686494158e-06,
      "loss": 0.2037,
      "step": 4058
    },
    {
      "epoch": 0.29,
      "grad_norm": 19.52068815925955,
      "learning_rate": 8.325805795416026e-06,
      "loss": 0.5157,
      "step": 4059
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3112256934612216,
      "learning_rate": 8.324940725821853e-06,
      "loss": 0.1944,
      "step": 4060
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3049338751172164,
      "learning_rate": 8.324075477758069e-06,
      "loss": 0.1836,
      "step": 4061
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4500275111610934,
      "learning_rate": 8.32321005127112e-06,
      "loss": 0.2391,
      "step": 4062
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.419033641146294,
      "learning_rate": 8.322344446407458e-06,
      "loss": 0.22,
      "step": 4063
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3499290967145556,
      "learning_rate": 8.321478663213545e-06,
      "loss": 0.234,
      "step": 4064
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.753539244912161,
      "learning_rate": 8.32061270173585e-06,
      "loss": 0.6655,
      "step": 4065
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2452613712248535,
      "learning_rate": 8.31974656202086e-06,
      "loss": 0.1633,
      "step": 4066
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4274079751012592,
      "learning_rate": 8.318880244115064e-06,
      "loss": 0.2086,
      "step": 4067
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4469898126182499,
      "learning_rate": 8.318013748064962e-06,
      "loss": 0.2201,
      "step": 4068
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4065476551753204,
      "learning_rate": 8.317147073917061e-06,
      "loss": 0.2172,
      "step": 4069
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3487391492003584,
      "learning_rate": 8.316280221717885e-06,
      "loss": 0.2126,
      "step": 4070
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3739442032285805,
      "learning_rate": 8.315413191513962e-06,
      "loss": 0.1873,
      "step": 4071
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3566982024593732,
      "learning_rate": 8.314545983351829e-06,
      "loss": 0.2468,
      "step": 4072
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.356501681831554,
      "learning_rate": 8.313678597278037e-06,
      "loss": 0.2224,
      "step": 4073
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2210012286000975,
      "learning_rate": 8.312811033339142e-06,
      "loss": 0.201,
      "step": 4074
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.1509725616504305,
      "learning_rate": 8.311943291581712e-06,
      "loss": 0.7771,
      "step": 4075
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.291329544504811,
      "learning_rate": 8.311075372052323e-06,
      "loss": 0.1863,
      "step": 4076
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1936210728746244,
      "learning_rate": 8.310207274797562e-06,
      "loss": 0.2061,
      "step": 4077
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.579251319348325,
      "learning_rate": 8.309338999864028e-06,
      "loss": 0.249,
      "step": 4078
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3274821216987136,
      "learning_rate": 8.30847054729832e-06,
      "loss": 0.2101,
      "step": 4079
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3685723584451246,
      "learning_rate": 8.30760191714706e-06,
      "loss": 0.2118,
      "step": 4080
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4376530406426933,
      "learning_rate": 8.306733109456867e-06,
      "loss": 0.204,
      "step": 4081
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2200599052393177,
      "learning_rate": 8.305864124274378e-06,
      "loss": 0.2041,
      "step": 4082
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.222004212985217,
      "learning_rate": 8.304994961646237e-06,
      "loss": 0.6477,
      "step": 4083
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3348909077177873,
      "learning_rate": 8.304125621619097e-06,
      "loss": 0.2147,
      "step": 4084
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4905274395269266,
      "learning_rate": 8.303256104239619e-06,
      "loss": 0.2135,
      "step": 4085
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.22422801618588,
      "learning_rate": 8.302386409554478e-06,
      "loss": 0.5331,
      "step": 4086
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.184461587209587,
      "learning_rate": 8.301516537610352e-06,
      "loss": 0.1987,
      "step": 4087
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.407086091293792,
      "learning_rate": 8.300646488453937e-06,
      "loss": 0.4933,
      "step": 4088
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3883815705156846,
      "learning_rate": 8.29977626213193e-06,
      "loss": 0.2346,
      "step": 4089
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.431575290629615,
      "learning_rate": 8.298905858691043e-06,
      "loss": 0.2121,
      "step": 4090
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2854725192766772,
      "learning_rate": 8.298035278177995e-06,
      "loss": 0.2244,
      "step": 4091
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.554042431356747,
      "learning_rate": 8.297164520639515e-06,
      "loss": 0.5459,
      "step": 4092
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.290859982180356,
      "learning_rate": 8.296293586122344e-06,
      "loss": 0.2481,
      "step": 4093
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.405049609277019,
      "learning_rate": 8.295422474673229e-06,
      "loss": 0.2254,
      "step": 4094
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4252413609951429,
      "learning_rate": 8.294551186338927e-06,
      "loss": 0.2677,
      "step": 4095
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.591828484511815,
      "learning_rate": 8.293679721166203e-06,
      "loss": 0.2345,
      "step": 4096
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3012073390371965,
      "learning_rate": 8.292808079201841e-06,
      "loss": 0.2349,
      "step": 4097
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.354907016075446,
      "learning_rate": 8.29193626049262e-06,
      "loss": 0.2316,
      "step": 4098
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.344608516474486,
      "learning_rate": 8.29106426508534e-06,
      "loss": 0.2126,
      "step": 4099
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.51590703523331,
      "learning_rate": 8.290192093026805e-06,
      "loss": 0.2012,
      "step": 4100
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4320061458719695,
      "learning_rate": 8.28931974436383e-06,
      "loss": 0.2346,
      "step": 4101
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.255925565059469,
      "learning_rate": 8.288447219143238e-06,
      "loss": 0.7,
      "step": 4102
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.314163445815323,
      "learning_rate": 8.287574517411864e-06,
      "loss": 0.1997,
      "step": 4103
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.83466197437151,
      "learning_rate": 8.28670163921655e-06,
      "loss": 0.5624,
      "step": 4104
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4770252516989408,
      "learning_rate": 8.28582858460415e-06,
      "loss": 0.2237,
      "step": 4105
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1830303643747915,
      "learning_rate": 8.284955353621527e-06,
      "loss": 0.2199,
      "step": 4106
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2878836431173761,
      "learning_rate": 8.284081946315548e-06,
      "loss": 0.2009,
      "step": 4107
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.311095538461595,
      "learning_rate": 8.283208362733099e-06,
      "loss": 0.1736,
      "step": 4108
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.29895157642666,
      "learning_rate": 8.282334602921069e-06,
      "loss": 0.1482,
      "step": 4109
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4466117098491575,
      "learning_rate": 8.281460666926357e-06,
      "loss": 0.2376,
      "step": 4110
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4990334342385714,
      "learning_rate": 8.280586554795872e-06,
      "loss": 0.2249,
      "step": 4111
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.385249887752725,
      "learning_rate": 8.279712266576535e-06,
      "loss": 0.244,
      "step": 4112
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2791768468954543,
      "learning_rate": 8.278837802315273e-06,
      "loss": 0.2307,
      "step": 4113
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.514441783731017,
      "learning_rate": 8.277963162059022e-06,
      "loss": 0.2851,
      "step": 4114
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3623507782184394,
      "learning_rate": 8.277088345854733e-06,
      "loss": 0.2606,
      "step": 4115
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5372309526673857,
      "learning_rate": 8.27621335374936e-06,
      "loss": 0.2487,
      "step": 4116
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4027257833471676,
      "learning_rate": 8.275338185789871e-06,
      "loss": 0.2432,
      "step": 4117
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2316506492789152,
      "learning_rate": 8.274462842023242e-06,
      "loss": 0.1781,
      "step": 4118
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.374250892893682,
      "learning_rate": 8.273587322496452e-06,
      "loss": 0.2012,
      "step": 4119
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.22973823962307,
      "learning_rate": 8.272711627256501e-06,
      "loss": 0.1926,
      "step": 4120
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.31771661945902,
      "learning_rate": 8.271835756350394e-06,
      "loss": 0.217,
      "step": 4121
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3528779786346508,
      "learning_rate": 8.27095970982514e-06,
      "loss": 0.2179,
      "step": 4122
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3347701668028291,
      "learning_rate": 8.270083487727765e-06,
      "loss": 0.2385,
      "step": 4123
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3508483193699328,
      "learning_rate": 8.269207090105298e-06,
      "loss": 0.2441,
      "step": 4124
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.331376672002768,
      "learning_rate": 8.268330517004783e-06,
      "loss": 0.2366,
      "step": 4125
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.227092700701747,
      "learning_rate": 8.26745376847327e-06,
      "loss": 0.1749,
      "step": 4126
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3760598933805466,
      "learning_rate": 8.266576844557821e-06,
      "loss": 0.2193,
      "step": 4127
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2654406922659633,
      "learning_rate": 8.265699745305503e-06,
      "loss": 0.18,
      "step": 4128
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3369645934186092,
      "learning_rate": 8.264822470763398e-06,
      "loss": 0.2395,
      "step": 4129
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2639163215814413,
      "learning_rate": 8.263945020978592e-06,
      "loss": 0.1928,
      "step": 4130
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.374377401477458,
      "learning_rate": 8.263067395998184e-06,
      "loss": 0.224,
      "step": 4131
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2924690761330049,
      "learning_rate": 8.262189595869283e-06,
      "loss": 0.2049,
      "step": 4132
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.003057329625412,
      "learning_rate": 8.261311620639003e-06,
      "loss": 0.59,
      "step": 4133
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4397745786883145,
      "learning_rate": 8.260433470354472e-06,
      "loss": 0.2087,
      "step": 4134
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1862558652117978,
      "learning_rate": 8.259555145062827e-06,
      "loss": 0.216,
      "step": 4135
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4587167794676854,
      "learning_rate": 8.25867664481121e-06,
      "loss": 0.2237,
      "step": 4136
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3210007316834875,
      "learning_rate": 8.257797969646776e-06,
      "loss": 0.2035,
      "step": 4137
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.66442182008484,
      "learning_rate": 8.25691911961669e-06,
      "loss": 0.2371,
      "step": 4138
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4066213884903278,
      "learning_rate": 8.256040094768124e-06,
      "loss": 0.2208,
      "step": 4139
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.355214208992719,
      "learning_rate": 8.255160895148263e-06,
      "loss": 0.228,
      "step": 4140
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.885406363426969,
      "learning_rate": 8.254281520804297e-06,
      "loss": 0.6964,
      "step": 4141
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4345638631716455,
      "learning_rate": 8.253401971783427e-06,
      "loss": 0.1567,
      "step": 4142
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.661640609282136,
      "learning_rate": 8.252522248132863e-06,
      "loss": 0.6906,
      "step": 4143
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4560462853892395,
      "learning_rate": 8.25164234989983e-06,
      "loss": 0.178,
      "step": 4144
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2758352610351231,
      "learning_rate": 8.25076227713155e-06,
      "loss": 0.1822,
      "step": 4145
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.118832600812816,
      "learning_rate": 8.249882029875267e-06,
      "loss": 0.4905,
      "step": 4146
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2449732874571167,
      "learning_rate": 8.249001608178226e-06,
      "loss": 0.1648,
      "step": 4147
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.981695559494868,
      "learning_rate": 8.24812101208769e-06,
      "loss": 0.5962,
      "step": 4148
    },
    {
      "epoch": 0.3,
      "grad_norm": 11.09725375725208,
      "learning_rate": 8.247240241650918e-06,
      "loss": 0.6791,
      "step": 4149
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.396648796654216,
      "learning_rate": 8.246359296915195e-06,
      "loss": 0.1778,
      "step": 4150
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2587879433554447,
      "learning_rate": 8.245478177927798e-06,
      "loss": 0.1634,
      "step": 4151
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.346562167063866,
      "learning_rate": 8.244596884736027e-06,
      "loss": 0.1886,
      "step": 4152
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4176954502337669,
      "learning_rate": 8.243715417387185e-06,
      "loss": 0.2251,
      "step": 4153
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.427726581021854,
      "learning_rate": 8.242833775928587e-06,
      "loss": 0.512,
      "step": 4154
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3579630019102151,
      "learning_rate": 8.241951960407556e-06,
      "loss": 0.1832,
      "step": 4155
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4056614544293775,
      "learning_rate": 8.241069970871422e-06,
      "loss": 0.1953,
      "step": 4156
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4421010053705154,
      "learning_rate": 8.240187807367528e-06,
      "loss": 0.1758,
      "step": 4157
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4562499241366997,
      "learning_rate": 8.239305469943227e-06,
      "loss": 0.167,
      "step": 4158
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3508505309317442,
      "learning_rate": 8.238422958645875e-06,
      "loss": 0.268,
      "step": 4159
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2752807956393788,
      "learning_rate": 8.237540273522844e-06,
      "loss": 0.2052,
      "step": 4160
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.181102513331286,
      "learning_rate": 8.236657414621516e-06,
      "loss": 0.5543,
      "step": 4161
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3618381320220463,
      "learning_rate": 8.235774381989273e-06,
      "loss": 0.226,
      "step": 4162
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.377872855858945,
      "learning_rate": 8.234891175673519e-06,
      "loss": 0.2175,
      "step": 4163
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1717665740809595,
      "learning_rate": 8.234007795721657e-06,
      "loss": 0.1697,
      "step": 4164
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.333651846452711,
      "learning_rate": 8.233124242181104e-06,
      "loss": 0.1883,
      "step": 4165
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2068486466799073,
      "learning_rate": 8.232240515099287e-06,
      "loss": 0.1865,
      "step": 4166
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2385403227931173,
      "learning_rate": 8.23135661452364e-06,
      "loss": 0.2125,
      "step": 4167
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4272564301174362,
      "learning_rate": 8.230472540501607e-06,
      "loss": 0.2323,
      "step": 4168
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3905431500471723,
      "learning_rate": 8.229588293080644e-06,
      "loss": 0.2249,
      "step": 4169
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3037801599855574,
      "learning_rate": 8.228703872308208e-06,
      "loss": 0.2058,
      "step": 4170
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3885224331416648,
      "learning_rate": 8.22781927823178e-06,
      "loss": 0.2046,
      "step": 4171
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6111665645795197,
      "learning_rate": 8.226934510898832e-06,
      "loss": 0.2054,
      "step": 4172
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.196500569041797,
      "learning_rate": 8.226049570356861e-06,
      "loss": 0.1782,
      "step": 4173
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3690664924664329,
      "learning_rate": 8.225164456653367e-06,
      "loss": 0.1905,
      "step": 4174
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.461886179277451,
      "learning_rate": 8.224279169835857e-06,
      "loss": 0.2135,
      "step": 4175
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1594626760721105,
      "learning_rate": 8.22339370995185e-06,
      "loss": 0.1837,
      "step": 4176
    },
    {
      "epoch": 0.3,
      "grad_norm": 19.656856543079726,
      "learning_rate": 8.222508077048875e-06,
      "loss": 0.4676,
      "step": 4177
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3163229086835568,
      "learning_rate": 8.221622271174467e-06,
      "loss": 0.2012,
      "step": 4178
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3506962459335055,
      "learning_rate": 8.220736292376177e-06,
      "loss": 0.1955,
      "step": 4179
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.888344368709606,
      "learning_rate": 8.219850140701557e-06,
      "loss": 0.6493,
      "step": 4180
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3622246846187454,
      "learning_rate": 8.218963816198174e-06,
      "loss": 0.2163,
      "step": 4181
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3470868804006588,
      "learning_rate": 8.2180773189136e-06,
      "loss": 0.1978,
      "step": 4182
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4109959877858365,
      "learning_rate": 8.217190648895421e-06,
      "loss": 0.2278,
      "step": 4183
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5902993487680914,
      "learning_rate": 8.21630380619123e-06,
      "loss": 0.2428,
      "step": 4184
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4217594417580592,
      "learning_rate": 8.215416790848626e-06,
      "loss": 0.2097,
      "step": 4185
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2774782747611606,
      "learning_rate": 8.214529602915226e-06,
      "loss": 0.1925,
      "step": 4186
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8001253571969031,
      "learning_rate": 8.213642242438646e-06,
      "loss": 0.2202,
      "step": 4187
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3165653675659297,
      "learning_rate": 8.212754709466519e-06,
      "loss": 0.229,
      "step": 4188
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4130863602956856,
      "learning_rate": 8.211867004046481e-06,
      "loss": 0.1852,
      "step": 4189
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2663213058231941,
      "learning_rate": 8.210979126226182e-06,
      "loss": 0.1919,
      "step": 4190
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.322538712326694,
      "learning_rate": 8.21009107605328e-06,
      "loss": 0.1965,
      "step": 4191
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5072965848132547,
      "learning_rate": 8.209202853575445e-06,
      "loss": 0.2,
      "step": 4192
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.731414517865259,
      "learning_rate": 8.208314458840347e-06,
      "loss": 0.5875,
      "step": 4193
    },
    {
      "epoch": 0.3,
      "grad_norm": 11.496024076905181,
      "learning_rate": 8.207425891895677e-06,
      "loss": 0.6528,
      "step": 4194
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3141460334551134,
      "learning_rate": 8.206537152789126e-06,
      "loss": 0.1871,
      "step": 4195
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3463192778368729,
      "learning_rate": 8.2056482415684e-06,
      "loss": 0.1703,
      "step": 4196
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4307792241044381,
      "learning_rate": 8.20475915828121e-06,
      "loss": 0.2128,
      "step": 4197
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5008272082930627,
      "learning_rate": 8.203869902975284e-06,
      "loss": 0.2647,
      "step": 4198
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3479888952877352,
      "learning_rate": 8.202980475698348e-06,
      "loss": 0.231,
      "step": 4199
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4640267857162308,
      "learning_rate": 8.202090876498144e-06,
      "loss": 0.2171,
      "step": 4200
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4754167364489201,
      "learning_rate": 8.201201105422423e-06,
      "loss": 0.2347,
      "step": 4201
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.191654897703027,
      "learning_rate": 8.200311162518946e-06,
      "loss": 0.183,
      "step": 4202
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.9070146062481172,
      "learning_rate": 8.199421047835478e-06,
      "loss": 0.5862,
      "step": 4203
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2632172450141754,
      "learning_rate": 8.1985307614198e-06,
      "loss": 0.2247,
      "step": 4204
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4556998963357939,
      "learning_rate": 8.197640303319698e-06,
      "loss": 0.2251,
      "step": 4205
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6009423598214585,
      "learning_rate": 8.19674967358297e-06,
      "loss": 0.2623,
      "step": 4206
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.376526142247953,
      "learning_rate": 8.195858872257416e-06,
      "loss": 0.2283,
      "step": 4207
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.409289209727124,
      "learning_rate": 8.194967899390856e-06,
      "loss": 0.198,
      "step": 4208
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2201765701022809,
      "learning_rate": 8.194076755031111e-06,
      "loss": 0.1498,
      "step": 4209
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.8405006723402275,
      "learning_rate": 8.193185439226015e-06,
      "loss": 0.7434,
      "step": 4210
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4568870718594595,
      "learning_rate": 8.192293952023412e-06,
      "loss": 0.1851,
      "step": 4211
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3328075707104006,
      "learning_rate": 8.191402293471151e-06,
      "loss": 0.1986,
      "step": 4212
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5489508345812042,
      "learning_rate": 8.190510463617093e-06,
      "loss": 0.261,
      "step": 4213
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4107171572549215,
      "learning_rate": 8.18961846250911e-06,
      "loss": 0.1796,
      "step": 4214
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1747404360071292,
      "learning_rate": 8.18872629019508e-06,
      "loss": 0.1952,
      "step": 4215
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3887880768437932,
      "learning_rate": 8.187833946722889e-06,
      "loss": 0.1736,
      "step": 4216
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.958813848424642,
      "learning_rate": 8.186941432140438e-06,
      "loss": 0.6375,
      "step": 4217
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3813345824735783,
      "learning_rate": 8.18604874649563e-06,
      "loss": 0.2122,
      "step": 4218
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4270433205816915,
      "learning_rate": 8.185155889836384e-06,
      "loss": 0.2226,
      "step": 4219
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.390120132128045,
      "learning_rate": 8.184262862210624e-06,
      "loss": 0.2549,
      "step": 4220
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2861797264493036,
      "learning_rate": 8.183369663666285e-06,
      "loss": 0.1872,
      "step": 4221
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1685513436165265,
      "learning_rate": 8.182476294251307e-06,
      "loss": 0.2081,
      "step": 4222
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.372084099564511,
      "learning_rate": 8.181582754013646e-06,
      "loss": 0.2063,
      "step": 4223
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4088541537318346,
      "learning_rate": 8.180689043001262e-06,
      "loss": 0.2232,
      "step": 4224
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.475689886526372,
      "learning_rate": 8.179795161262128e-06,
      "loss": 0.228,
      "step": 4225
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3221744112747513,
      "learning_rate": 8.17890110884422e-06,
      "loss": 0.1875,
      "step": 4226
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3833247216758258,
      "learning_rate": 8.178006885795531e-06,
      "loss": 0.246,
      "step": 4227
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4077368289885681,
      "learning_rate": 8.177112492164058e-06,
      "loss": 0.2531,
      "step": 4228
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3879596969444226,
      "learning_rate": 8.176217927997807e-06,
      "loss": 0.2028,
      "step": 4229
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.497118082063742,
      "learning_rate": 8.175323193344797e-06,
      "loss": 0.2079,
      "step": 4230
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1110221012685797,
      "learning_rate": 8.174428288253053e-06,
      "loss": 0.1755,
      "step": 4231
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4784721831246894,
      "learning_rate": 8.173533212770611e-06,
      "loss": 0.2288,
      "step": 4232
    },
    {
      "epoch": 0.3,
      "grad_norm": 7.765726509256648,
      "learning_rate": 8.172637966945513e-06,
      "loss": 0.6413,
      "step": 4233
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6041143094542212,
      "learning_rate": 8.171742550825814e-06,
      "loss": 0.2261,
      "step": 4234
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3054804726043723,
      "learning_rate": 8.170846964459573e-06,
      "loss": 0.2422,
      "step": 4235
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.456502236131684,
      "learning_rate": 8.169951207894866e-06,
      "loss": 0.2463,
      "step": 4236
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.377579537608348,
      "learning_rate": 8.169055281179771e-06,
      "loss": 0.222,
      "step": 4237
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3209362108055605,
      "learning_rate": 8.16815918436238e-06,
      "loss": 0.1945,
      "step": 4238
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.494896256215283,
      "learning_rate": 8.167262917490788e-06,
      "loss": 0.7202,
      "step": 4239
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3678659303048917,
      "learning_rate": 8.166366480613107e-06,
      "loss": 0.1867,
      "step": 4240
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3718732679223735,
      "learning_rate": 8.165469873777452e-06,
      "loss": 0.2378,
      "step": 4241
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3878805365798734,
      "learning_rate": 8.16457309703195e-06,
      "loss": 0.229,
      "step": 4242
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3234508201567718,
      "learning_rate": 8.163676150424736e-06,
      "loss": 0.2414,
      "step": 4243
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4080048019737115,
      "learning_rate": 8.162779034003955e-06,
      "loss": 0.2135,
      "step": 4244
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2611355383410465,
      "learning_rate": 8.161881747817761e-06,
      "loss": 0.1635,
      "step": 4245
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.08627799068848,
      "learning_rate": 8.160984291914316e-06,
      "loss": 0.6177,
      "step": 4246
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6488115455466497,
      "learning_rate": 8.160086666341794e-06,
      "loss": 0.2247,
      "step": 4247
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3016815243402677,
      "learning_rate": 8.159188871148372e-06,
      "loss": 0.1971,
      "step": 4248
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.472141529986145,
      "learning_rate": 8.15829090638224e-06,
      "loss": 0.2493,
      "step": 4249
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.386682010068181,
      "learning_rate": 8.157392772091603e-06,
      "loss": 0.1765,
      "step": 4250
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.388635899272824,
      "learning_rate": 8.156494468324665e-06,
      "loss": 0.202,
      "step": 4251
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.21510549038626,
      "learning_rate": 8.155595995129643e-06,
      "loss": 0.6997,
      "step": 4252
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.8853480843177355,
      "learning_rate": 8.154697352554766e-06,
      "loss": 0.525,
      "step": 4253
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6264553729219897,
      "learning_rate": 8.153798540648266e-06,
      "loss": 0.2556,
      "step": 4254
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3589438318867064,
      "learning_rate": 8.152899559458392e-06,
      "loss": 0.2011,
      "step": 4255
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.495531570310373,
      "learning_rate": 8.152000409033394e-06,
      "loss": 0.2154,
      "step": 4256
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3819337328695815,
      "learning_rate": 8.151101089421536e-06,
      "loss": 0.2226,
      "step": 4257
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3092968125303006,
      "learning_rate": 8.15020160067109e-06,
      "loss": 0.1944,
      "step": 4258
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5690378284594118,
      "learning_rate": 8.149301942830338e-06,
      "loss": 0.2143,
      "step": 4259
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.601018913250708,
      "learning_rate": 8.14840211594757e-06,
      "loss": 0.235,
      "step": 4260
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.972127047551957,
      "learning_rate": 8.147502120071084e-06,
      "loss": 0.6078,
      "step": 4261
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4222056146831985,
      "learning_rate": 8.146601955249187e-06,
      "loss": 0.2132,
      "step": 4262
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4553903088468665,
      "learning_rate": 8.1457016215302e-06,
      "loss": 0.2131,
      "step": 4263
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.628269829543279,
      "learning_rate": 8.144801118962447e-06,
      "loss": 0.2044,
      "step": 4264
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4375679569378084,
      "learning_rate": 8.143900447594261e-06,
      "loss": 0.2592,
      "step": 4265
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3060112939564996,
      "learning_rate": 8.142999607473992e-06,
      "loss": 0.2025,
      "step": 4266
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2252729190796254,
      "learning_rate": 8.142098598649987e-06,
      "loss": 0.1931,
      "step": 4267
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4526668673570355,
      "learning_rate": 8.141197421170617e-06,
      "loss": 0.2692,
      "step": 4268
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2920917421539715,
      "learning_rate": 8.140296075084246e-06,
      "loss": 0.196,
      "step": 4269
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2275237203838303,
      "learning_rate": 8.139394560439257e-06,
      "loss": 0.1951,
      "step": 4270
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.493707734851143,
      "learning_rate": 8.138492877284041e-06,
      "loss": 0.1878,
      "step": 4271
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3760960348368563,
      "learning_rate": 8.137591025666996e-06,
      "loss": 0.1898,
      "step": 4272
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.988827822730528,
      "learning_rate": 8.13668900563653e-06,
      "loss": 0.6292,
      "step": 4273
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.50941137823024,
      "learning_rate": 8.135786817241056e-06,
      "loss": 0.2468,
      "step": 4274
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5295258525481512,
      "learning_rate": 8.134884460529007e-06,
      "loss": 0.261,
      "step": 4275
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.591735840225546,
      "learning_rate": 8.133981935548814e-06,
      "loss": 0.6078,
      "step": 4276
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.334206505646596,
      "learning_rate": 8.13307924234892e-06,
      "loss": 0.1894,
      "step": 4277
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3394397827002784,
      "learning_rate": 8.13217638097778e-06,
      "loss": 0.1871,
      "step": 4278
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2901583074830638,
      "learning_rate": 8.131273351483857e-06,
      "loss": 0.2088,
      "step": 4279
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.382496241929535,
      "learning_rate": 8.13037015391562e-06,
      "loss": 0.2066,
      "step": 4280
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3771757940572398,
      "learning_rate": 8.129466788321547e-06,
      "loss": 0.2298,
      "step": 4281
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5392102482960015,
      "learning_rate": 8.128563254750132e-06,
      "loss": 0.2738,
      "step": 4282
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.447652591214568,
      "learning_rate": 8.127659553249869e-06,
      "loss": 0.206,
      "step": 4283
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3635992729973379,
      "learning_rate": 8.126755683869267e-06,
      "loss": 0.2077,
      "step": 4284
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2060758837323673,
      "learning_rate": 8.125851646656843e-06,
      "loss": 0.2036,
      "step": 4285
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4861654512601996,
      "learning_rate": 8.12494744166112e-06,
      "loss": 0.2661,
      "step": 4286
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.05700923836143,
      "learning_rate": 8.124043068930633e-06,
      "loss": 0.563,
      "step": 4287
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5468862518324757,
      "learning_rate": 8.123138528513926e-06,
      "loss": 0.2324,
      "step": 4288
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3193142074907194,
      "learning_rate": 8.12223382045955e-06,
      "loss": 0.2004,
      "step": 4289
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.38394573017972,
      "learning_rate": 8.121328944816067e-06,
      "loss": 0.2252,
      "step": 4290
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2461375197015399,
      "learning_rate": 8.120423901632046e-06,
      "loss": 0.1681,
      "step": 4291
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4568844544160655,
      "learning_rate": 8.119518690956066e-06,
      "loss": 0.2025,
      "step": 4292
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.530716963461995,
      "learning_rate": 8.118613312836717e-06,
      "loss": 0.2379,
      "step": 4293
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3471392189116826,
      "learning_rate": 8.117707767322594e-06,
      "loss": 0.2382,
      "step": 4294
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.253390141022074,
      "learning_rate": 8.116802054462305e-06,
      "loss": 0.2125,
      "step": 4295
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.247934787101151,
      "learning_rate": 8.115896174304465e-06,
      "loss": 0.195,
      "step": 4296
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2581168812556756,
      "learning_rate": 8.114990126897694e-06,
      "loss": 0.1834,
      "step": 4297
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2644714288631789,
      "learning_rate": 8.114083912290631e-06,
      "loss": 0.1927,
      "step": 4298
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2128161930807477,
      "learning_rate": 8.113177530531915e-06,
      "loss": 0.2015,
      "step": 4299
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3381978625508204,
      "learning_rate": 8.112270981670196e-06,
      "loss": 0.1946,
      "step": 4300
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7883791587152686,
      "learning_rate": 8.111364265754136e-06,
      "loss": 0.3014,
      "step": 4301
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2386293146835143,
      "learning_rate": 8.110457382832402e-06,
      "loss": 0.2171,
      "step": 4302
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.272227394407068,
      "learning_rate": 8.109550332953674e-06,
      "loss": 0.5293,
      "step": 4303
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.435415085467868,
      "learning_rate": 8.108643116166637e-06,
      "loss": 0.2429,
      "step": 4304
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3248174984390693,
      "learning_rate": 8.107735732519988e-06,
      "loss": 0.1872,
      "step": 4305
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.456474350496373,
      "learning_rate": 8.106828182062429e-06,
      "loss": 0.2241,
      "step": 4306
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5007446081463225,
      "learning_rate": 8.105920464842679e-06,
      "loss": 0.2472,
      "step": 4307
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6132215064587982,
      "learning_rate": 8.105012580909457e-06,
      "loss": 0.2207,
      "step": 4308
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4299033977950637,
      "learning_rate": 8.104104530311494e-06,
      "loss": 0.1958,
      "step": 4309
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4576235584286805,
      "learning_rate": 8.103196313097533e-06,
      "loss": 0.1843,
      "step": 4310
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3576010659856075,
      "learning_rate": 8.102287929316321e-06,
      "loss": 0.2277,
      "step": 4311
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4070119321783041,
      "learning_rate": 8.101379379016618e-06,
      "loss": 0.2218,
      "step": 4312
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3107927926177785,
      "learning_rate": 8.100470662247192e-06,
      "loss": 0.201,
      "step": 4313
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8606425157144113,
      "learning_rate": 8.099561779056816e-06,
      "loss": 0.1997,
      "step": 4314
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.443892050419447,
      "learning_rate": 8.098652729494281e-06,
      "loss": 0.2171,
      "step": 4315
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3595994519217165,
      "learning_rate": 8.097743513608377e-06,
      "loss": 0.215,
      "step": 4316
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.338811580758125,
      "learning_rate": 8.096834131447905e-06,
      "loss": 0.245,
      "step": 4317
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.748082379088702,
      "learning_rate": 8.095924583061682e-06,
      "loss": 0.2103,
      "step": 4318
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4114121941161277,
      "learning_rate": 8.095014868498529e-06,
      "loss": 0.1814,
      "step": 4319
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.4748395918483106,
      "learning_rate": 8.09410498780727e-06,
      "loss": 0.2016,
      "step": 4320
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5878975618580913,
      "learning_rate": 8.093194941036751e-06,
      "loss": 0.2284,
      "step": 4321
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.535523107075687,
      "learning_rate": 8.092284728235812e-06,
      "loss": 0.2302,
      "step": 4322
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4614526758990578,
      "learning_rate": 8.091374349453318e-06,
      "loss": 0.2008,
      "step": 4323
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4074784083936163,
      "learning_rate": 8.090463804738126e-06,
      "loss": 0.2143,
      "step": 4324
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.509964347716413,
      "learning_rate": 8.089553094139117e-06,
      "loss": 0.2223,
      "step": 4325
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4374751270016897,
      "learning_rate": 8.088642217705174e-06,
      "loss": 0.2135,
      "step": 4326
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.38759590397608,
      "learning_rate": 8.087731175485184e-06,
      "loss": 0.2104,
      "step": 4327
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4020177497597461,
      "learning_rate": 8.086819967528053e-06,
      "loss": 0.1779,
      "step": 4328
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4408923995659428,
      "learning_rate": 8.08590859388269e-06,
      "loss": 0.2091,
      "step": 4329
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5119556082885228,
      "learning_rate": 8.084997054598012e-06,
      "loss": 0.1954,
      "step": 4330
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2391208604655684,
      "learning_rate": 8.084085349722948e-06,
      "loss": 0.1726,
      "step": 4331
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3950546255052358,
      "learning_rate": 8.083173479306436e-06,
      "loss": 0.2474,
      "step": 4332
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.372392646739044,
      "learning_rate": 8.08226144339742e-06,
      "loss": 0.2225,
      "step": 4333
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.287458644525628,
      "learning_rate": 8.081349242044854e-06,
      "loss": 0.8016,
      "step": 4334
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5096815544220206,
      "learning_rate": 8.080436875297703e-06,
      "loss": 0.2277,
      "step": 4335
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5331512870551067,
      "learning_rate": 8.07952434320494e-06,
      "loss": 0.2211,
      "step": 4336
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.411875731148359,
      "learning_rate": 8.078611645815543e-06,
      "loss": 0.2585,
      "step": 4337
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.648258019095411,
      "learning_rate": 8.077698783178503e-06,
      "loss": 0.4517,
      "step": 4338
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3103017860123887,
      "learning_rate": 8.07678575534282e-06,
      "loss": 0.229,
      "step": 4339
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.029352018863754,
      "learning_rate": 8.075872562357502e-06,
      "loss": 0.71,
      "step": 4340
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.389703688882895,
      "learning_rate": 8.074959204271563e-06,
      "loss": 0.2061,
      "step": 4341
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3985193573074026,
      "learning_rate": 8.07404568113403e-06,
      "loss": 0.2061,
      "step": 4342
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4570942289517306,
      "learning_rate": 8.07313199299394e-06,
      "loss": 0.1781,
      "step": 4343
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.753879136287685,
      "learning_rate": 8.07221813990033e-06,
      "loss": 0.6069,
      "step": 4344
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3547257297122248,
      "learning_rate": 8.071304121902257e-06,
      "loss": 0.166,
      "step": 4345
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3039872159357055,
      "learning_rate": 8.070389939048782e-06,
      "loss": 0.2142,
      "step": 4346
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7021591994034695,
      "learning_rate": 8.06947559138897e-06,
      "loss": 0.2479,
      "step": 4347
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4644880783973198,
      "learning_rate": 8.068561078971905e-06,
      "loss": 0.245,
      "step": 4348
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.198600718998808,
      "learning_rate": 8.067646401846672e-06,
      "loss": 0.6469,
      "step": 4349
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4479598256047939,
      "learning_rate": 8.066731560062366e-06,
      "loss": 0.243,
      "step": 4350
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6751343075158596,
      "learning_rate": 8.065816553668094e-06,
      "loss": 0.2016,
      "step": 4351
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2865171264850355,
      "learning_rate": 8.064901382712968e-06,
      "loss": 0.2335,
      "step": 4352
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.379490490655004,
      "learning_rate": 8.063986047246113e-06,
      "loss": 0.6172,
      "step": 4353
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.367335215695112,
      "learning_rate": 8.06307054731666e-06,
      "loss": 0.2506,
      "step": 4354
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3726092380062855,
      "learning_rate": 8.06215488297375e-06,
      "loss": 0.1656,
      "step": 4355
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4612872418740448,
      "learning_rate": 8.06123905426653e-06,
      "loss": 0.2115,
      "step": 4356
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3516581278007367,
      "learning_rate": 8.060323061244158e-06,
      "loss": 0.2455,
      "step": 4357
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5130941615553144,
      "learning_rate": 8.059406903955803e-06,
      "loss": 0.253,
      "step": 4358
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2986153397418696,
      "learning_rate": 8.05849058245064e-06,
      "loss": 0.1944,
      "step": 4359
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.240657581779312,
      "learning_rate": 8.057574096777854e-06,
      "loss": 0.5496,
      "step": 4360
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4123866434694117,
      "learning_rate": 8.056657446986636e-06,
      "loss": 0.2001,
      "step": 4361
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4678821623436953,
      "learning_rate": 8.05574063312619e-06,
      "loss": 0.2681,
      "step": 4362
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1632388589280016,
      "learning_rate": 8.054823655245729e-06,
      "loss": 0.1684,
      "step": 4363
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4516975383265291,
      "learning_rate": 8.053906513394468e-06,
      "loss": 0.2378,
      "step": 4364
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3513223222369897,
      "learning_rate": 8.052989207621637e-06,
      "loss": 0.1845,
      "step": 4365
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.645830662841758,
      "learning_rate": 8.052071737976477e-06,
      "loss": 0.5025,
      "step": 4366
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2491152825534535,
      "learning_rate": 8.05115410450823e-06,
      "loss": 0.1452,
      "step": 4367
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.032912973155827,
      "learning_rate": 8.05023630726615e-06,
      "loss": 0.5316,
      "step": 4368
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5954177990220197,
      "learning_rate": 8.049318346299505e-06,
      "loss": 0.2024,
      "step": 4369
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.33378192565435,
      "learning_rate": 8.048400221657566e-06,
      "loss": 0.2095,
      "step": 4370
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6174045257043435,
      "learning_rate": 8.047481933389613e-06,
      "loss": 0.2282,
      "step": 4371
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.587260066239751,
      "learning_rate": 8.046563481544938e-06,
      "loss": 0.2182,
      "step": 4372
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.506557481235486,
      "learning_rate": 8.045644866172838e-06,
      "loss": 0.2541,
      "step": 4373
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2557971615331542,
      "learning_rate": 8.044726087322621e-06,
      "loss": 0.1778,
      "step": 4374
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2884354900647808,
      "learning_rate": 8.043807145043604e-06,
      "loss": 0.2031,
      "step": 4375
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4256522171814492,
      "learning_rate": 8.042888039385112e-06,
      "loss": 0.2385,
      "step": 4376
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4482712939344824,
      "learning_rate": 8.041968770396477e-06,
      "loss": 0.2087,
      "step": 4377
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4235077321191418,
      "learning_rate": 8.041049338127045e-06,
      "loss": 0.2584,
      "step": 4378
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1395763654669604,
      "learning_rate": 8.040129742626167e-06,
      "loss": 0.1867,
      "step": 4379
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4914504486091855,
      "learning_rate": 8.039209983943201e-06,
      "loss": 0.2077,
      "step": 4380
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2168371645015115,
      "learning_rate": 8.038290062127517e-06,
      "loss": 0.1904,
      "step": 4381
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.464070512158537,
      "learning_rate": 8.037369977228495e-06,
      "loss": 0.2286,
      "step": 4382
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3269698833704942,
      "learning_rate": 8.036449729295517e-06,
      "loss": 0.2199,
      "step": 4383
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.552754738543002,
      "learning_rate": 8.035529318377981e-06,
      "loss": 0.2402,
      "step": 4384
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.470719604985931,
      "learning_rate": 8.034608744525292e-06,
      "loss": 0.1935,
      "step": 4385
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.521670332806466,
      "learning_rate": 8.03368800778686e-06,
      "loss": 0.2012,
      "step": 4386
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.25690874284925,
      "learning_rate": 8.032767108212108e-06,
      "loss": 0.1617,
      "step": 4387
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2821137179733408,
      "learning_rate": 8.031846045850468e-06,
      "loss": 0.1724,
      "step": 4388
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.8731199406179995,
      "learning_rate": 8.030924820751375e-06,
      "loss": 0.61,
      "step": 4389
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.509132344249142,
      "learning_rate": 8.03000343296428e-06,
      "loss": 0.2107,
      "step": 4390
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3624206430473076,
      "learning_rate": 8.02908188253864e-06,
      "loss": 0.1738,
      "step": 4391
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.28185480192736,
      "learning_rate": 8.028160169523915e-06,
      "loss": 0.209,
      "step": 4392
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.326127703574001,
      "learning_rate": 8.027238293969583e-06,
      "loss": 0.1967,
      "step": 4393
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3773325555259772,
      "learning_rate": 8.026316255925127e-06,
      "loss": 0.192,
      "step": 4394
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2817350600827935,
      "learning_rate": 8.025394055440037e-06,
      "loss": 0.2089,
      "step": 4395
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.337795880724,
      "learning_rate": 8.024471692563814e-06,
      "loss": 0.2009,
      "step": 4396
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.447831148723988,
      "learning_rate": 8.023549167345966e-06,
      "loss": 0.2442,
      "step": 4397
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.460251885683025,
      "learning_rate": 8.02262647983601e-06,
      "loss": 0.228,
      "step": 4398
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4368620433986492,
      "learning_rate": 8.021703630083472e-06,
      "loss": 0.1974,
      "step": 4399
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4290927471456292,
      "learning_rate": 8.020780618137889e-06,
      "loss": 0.2137,
      "step": 4400
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.030068389467755,
      "learning_rate": 8.019857444048803e-06,
      "loss": 0.6593,
      "step": 4401
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2845583880557232,
      "learning_rate": 8.018934107865765e-06,
      "loss": 0.2239,
      "step": 4402
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.1816193331842175,
      "learning_rate": 8.018010609638341e-06,
      "loss": 0.6666,
      "step": 4403
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4196583890055376,
      "learning_rate": 8.017086949416095e-06,
      "loss": 0.2304,
      "step": 4404
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4869634830644582,
      "learning_rate": 8.01616312724861e-06,
      "loss": 0.2426,
      "step": 4405
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3900479989047305,
      "learning_rate": 8.015239143185469e-06,
      "loss": 0.1895,
      "step": 4406
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.366312103455014,
      "learning_rate": 8.01431499727627e-06,
      "loss": 0.2252,
      "step": 4407
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3988855544535694,
      "learning_rate": 8.013390689570616e-06,
      "loss": 0.1734,
      "step": 4408
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3595451562610792,
      "learning_rate": 8.012466220118125e-06,
      "loss": 0.2083,
      "step": 4409
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.593861433496469,
      "learning_rate": 8.011541588968412e-06,
      "loss": 0.572,
      "step": 4410
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3863799945928819,
      "learning_rate": 8.010616796171112e-06,
      "loss": 0.1874,
      "step": 4411
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2195654185749925,
      "learning_rate": 8.009691841775864e-06,
      "loss": 0.1855,
      "step": 4412
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3872269768531935,
      "learning_rate": 8.008766725832313e-06,
      "loss": 0.2477,
      "step": 4413
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4138598883588593,
      "learning_rate": 8.007841448390116e-06,
      "loss": 0.2256,
      "step": 4414
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.242908211430397,
      "learning_rate": 8.006916009498943e-06,
      "loss": 0.1791,
      "step": 4415
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4630881752965528,
      "learning_rate": 8.005990409208463e-06,
      "loss": 0.215,
      "step": 4416
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.235703887416425,
      "learning_rate": 8.005064647568359e-06,
      "loss": 0.1845,
      "step": 4417
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4957633828146948,
      "learning_rate": 8.004138724628324e-06,
      "loss": 0.2361,
      "step": 4418
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1913757851587805,
      "learning_rate": 8.003212640438057e-06,
      "loss": 0.157,
      "step": 4419
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2943225304644,
      "learning_rate": 8.002286395047267e-06,
      "loss": 0.1635,
      "step": 4420
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6563952861503435,
      "learning_rate": 8.00135998850567e-06,
      "loss": 0.1996,
      "step": 4421
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6545270968008416,
      "learning_rate": 8.000433420862995e-06,
      "loss": 0.1775,
      "step": 4422
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6186880325413322,
      "learning_rate": 7.99950669216897e-06,
      "loss": 0.2464,
      "step": 4423
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5681421963143796,
      "learning_rate": 7.998579802473345e-06,
      "loss": 0.224,
      "step": 4424
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2641915288411145,
      "learning_rate": 7.997652751825868e-06,
      "loss": 0.2296,
      "step": 4425
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5499752972234897,
      "learning_rate": 7.9967255402763e-06,
      "loss": 0.2158,
      "step": 4426
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3982138383879292,
      "learning_rate": 7.995798167874412e-06,
      "loss": 0.2203,
      "step": 4427
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4403325618617224,
      "learning_rate": 7.994870634669978e-06,
      "loss": 0.2534,
      "step": 4428
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4993248958797845,
      "learning_rate": 7.993942940712789e-06,
      "loss": 0.2509,
      "step": 4429
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5977269404897865,
      "learning_rate": 7.993015086052634e-06,
      "loss": 0.2619,
      "step": 4430
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3053874858221115,
      "learning_rate": 7.99208707073932e-06,
      "loss": 0.2075,
      "step": 4431
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.9933211059830316,
      "learning_rate": 7.991158894822662e-06,
      "loss": 0.5616,
      "step": 4432
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.025397648166224,
      "learning_rate": 7.990230558352476e-06,
      "loss": 0.5474,
      "step": 4433
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3890792692972895,
      "learning_rate": 7.989302061378593e-06,
      "loss": 0.2156,
      "step": 4434
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5062850658802256,
      "learning_rate": 7.98837340395085e-06,
      "loss": 0.2149,
      "step": 4435
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4609356968777376,
      "learning_rate": 7.987444586119098e-06,
      "loss": 0.1916,
      "step": 4436
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2760646227945385,
      "learning_rate": 7.986515607933186e-06,
      "loss": 0.1852,
      "step": 4437
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3883782457986629,
      "learning_rate": 7.985586469442983e-06,
      "loss": 0.2497,
      "step": 4438
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3294698931618316,
      "learning_rate": 7.984657170698358e-06,
      "loss": 0.1991,
      "step": 4439
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2418933256847227,
      "learning_rate": 7.983727711749194e-06,
      "loss": 0.1883,
      "step": 4440
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3218579735000666,
      "learning_rate": 7.982798092645381e-06,
      "loss": 0.2025,
      "step": 4441
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3669601194605174,
      "learning_rate": 7.981868313436815e-06,
      "loss": 0.2249,
      "step": 4442
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6158819205955401,
      "learning_rate": 7.980938374173405e-06,
      "loss": 0.2049,
      "step": 4443
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.178648280313349,
      "learning_rate": 7.980008274905067e-06,
      "loss": 0.1892,
      "step": 4444
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5124556239064182,
      "learning_rate": 7.979078015681723e-06,
      "loss": 0.1968,
      "step": 4445
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.060394445214173,
      "learning_rate": 7.978147596553305e-06,
      "loss": 0.6798,
      "step": 4446
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2979858984801371,
      "learning_rate": 7.977217017569756e-06,
      "loss": 0.1698,
      "step": 4447
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4453727436678074,
      "learning_rate": 7.976286278781027e-06,
      "loss": 0.2332,
      "step": 4448
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2491791254477551,
      "learning_rate": 7.975355380237073e-06,
      "loss": 0.2006,
      "step": 4449
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.552121570871094,
      "learning_rate": 7.974424321987864e-06,
      "loss": 0.4388,
      "step": 4450
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1382705219898712,
      "learning_rate": 7.973493104083373e-06,
      "loss": 0.1635,
      "step": 4451
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.344768779122071,
      "learning_rate": 7.972561726573584e-06,
      "loss": 0.2652,
      "step": 4452
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.906376285964811,
      "learning_rate": 7.971630189508494e-06,
      "loss": 0.6874,
      "step": 4453
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3952139362505775,
      "learning_rate": 7.970698492938099e-06,
      "loss": 0.2032,
      "step": 4454
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5678841917122002,
      "learning_rate": 7.969766636912411e-06,
      "loss": 0.2017,
      "step": 4455
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3956741388414295,
      "learning_rate": 7.96883462148145e-06,
      "loss": 0.2545,
      "step": 4456
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4088872109774793,
      "learning_rate": 7.96790244669524e-06,
      "loss": 0.2637,
      "step": 4457
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3093415022609916,
      "learning_rate": 7.966970112603816e-06,
      "loss": 0.2268,
      "step": 4458
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.295417617695109,
      "learning_rate": 7.966037619257225e-06,
      "loss": 0.1831,
      "step": 4459
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1115885543034163,
      "learning_rate": 7.965104966705518e-06,
      "loss": 0.1778,
      "step": 4460
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4145462216345506,
      "learning_rate": 7.964172154998755e-06,
      "loss": 0.2022,
      "step": 4461
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4518580632109828,
      "learning_rate": 7.963239184187008e-06,
      "loss": 0.2346,
      "step": 4462
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.364393664812921,
      "learning_rate": 7.962306054320353e-06,
      "loss": 0.2088,
      "step": 4463
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3584626301828335,
      "learning_rate": 7.96137276544888e-06,
      "loss": 0.235,
      "step": 4464
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4934977840503671,
      "learning_rate": 7.960439317622679e-06,
      "loss": 0.1944,
      "step": 4465
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3521829875729687,
      "learning_rate": 7.959505710891857e-06,
      "loss": 0.2037,
      "step": 4466
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2833183485163702,
      "learning_rate": 7.958571945306526e-06,
      "loss": 0.1858,
      "step": 4467
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4047627421962094,
      "learning_rate": 7.957638020916807e-06,
      "loss": 0.222,
      "step": 4468
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.391299484235729,
      "learning_rate": 7.95670393777283e-06,
      "loss": 0.1888,
      "step": 4469
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.419979642048531,
      "learning_rate": 7.95576969592473e-06,
      "loss": 0.6293,
      "step": 4470
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.623453409076198,
      "learning_rate": 7.954835295422659e-06,
      "loss": 0.276,
      "step": 4471
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4137906497530288,
      "learning_rate": 7.953900736316766e-06,
      "loss": 0.1908,
      "step": 4472
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3033307833561416,
      "learning_rate": 7.952966018657217e-06,
      "loss": 0.1881,
      "step": 4473
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2493001610096695,
      "learning_rate": 7.952031142494184e-06,
      "loss": 0.2532,
      "step": 4474
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2457485934423147,
      "learning_rate": 7.951096107877845e-06,
      "loss": 0.2027,
      "step": 4475
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3660009618879005,
      "learning_rate": 7.950160914858392e-06,
      "loss": 0.215,
      "step": 4476
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2726755001039265,
      "learning_rate": 7.949225563486022e-06,
      "loss": 0.2347,
      "step": 4477
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3641903138221871,
      "learning_rate": 7.94829005381094e-06,
      "loss": 0.2108,
      "step": 4478
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3290355546439336,
      "learning_rate": 7.947354385883363e-06,
      "loss": 0.2056,
      "step": 4479
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3146752574988225,
      "learning_rate": 7.946418559753509e-06,
      "loss": 0.1862,
      "step": 4480
    },
    {
      "epoch": 0.32,
      "grad_norm": 32.4022285563669,
      "learning_rate": 7.945482575471614e-06,
      "loss": 0.6941,
      "step": 4481
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.281183678234258,
      "learning_rate": 7.944546433087915e-06,
      "loss": 0.1883,
      "step": 4482
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3384358782648709,
      "learning_rate": 7.943610132652661e-06,
      "loss": 0.2398,
      "step": 4483
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2019106210608,
      "learning_rate": 7.94267367421611e-06,
      "loss": 0.1882,
      "step": 4484
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5677833275684363,
      "learning_rate": 7.941737057828528e-06,
      "loss": 0.2568,
      "step": 4485
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3380521824873923,
      "learning_rate": 7.940800283540187e-06,
      "loss": 0.2321,
      "step": 4486
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.293598817422054,
      "learning_rate": 7.93986335140137e-06,
      "loss": 0.2249,
      "step": 4487
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3043607263182484,
      "learning_rate": 7.938926261462366e-06,
      "loss": 0.2352,
      "step": 4488
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.452477829001722,
      "learning_rate": 7.937989013773478e-06,
      "loss": 0.2357,
      "step": 4489
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2167689531023806,
      "learning_rate": 7.93705160838501e-06,
      "loss": 0.2094,
      "step": 4490
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.145762137923544,
      "learning_rate": 7.936114045347283e-06,
      "loss": 0.5817,
      "step": 4491
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4686505885351229,
      "learning_rate": 7.935176324710617e-06,
      "loss": 0.2028,
      "step": 4492
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.342718985548421,
      "learning_rate": 7.934238446525346e-06,
      "loss": 0.1552,
      "step": 4493
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3295684792728466,
      "learning_rate": 7.933300410841812e-06,
      "loss": 0.2243,
      "step": 4494
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.53041440632651,
      "learning_rate": 7.932362217710366e-06,
      "loss": 0.754,
      "step": 4495
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.386632453836793,
      "learning_rate": 7.931423867181365e-06,
      "loss": 0.6938,
      "step": 4496
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2008989402072303,
      "learning_rate": 7.93048535930518e-06,
      "loss": 0.1883,
      "step": 4497
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.375885058889189,
      "learning_rate": 7.929546694132179e-06,
      "loss": 0.2413,
      "step": 4498
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.5219274529473505,
      "learning_rate": 7.928607871712753e-06,
      "loss": 0.6843,
      "step": 4499
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.319016846670825,
      "learning_rate": 7.927668892097288e-06,
      "loss": 0.1827,
      "step": 4500
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3139024377211619,
      "learning_rate": 7.92672975533619e-06,
      "loss": 0.2161,
      "step": 4501
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.628506686462941,
      "learning_rate": 7.925790461479866e-06,
      "loss": 0.5458,
      "step": 4502
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5399219850173202,
      "learning_rate": 7.924851010578734e-06,
      "loss": 0.2258,
      "step": 4503
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2731638508450254,
      "learning_rate": 7.92391140268322e-06,
      "loss": 0.187,
      "step": 4504
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.7069782511242755,
      "learning_rate": 7.922971637843755e-06,
      "loss": 0.5683,
      "step": 4505
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5213544284024967,
      "learning_rate": 7.922031716110788e-06,
      "loss": 0.1985,
      "step": 4506
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.547754680437872,
      "learning_rate": 7.921091637534765e-06,
      "loss": 0.6847,
      "step": 4507
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.657259229431092,
      "learning_rate": 7.920151402166148e-06,
      "loss": 0.6026,
      "step": 4508
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.944326504417041,
      "learning_rate": 7.919211010055406e-06,
      "loss": 0.7102,
      "step": 4509
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2567871898564404,
      "learning_rate": 7.918270461253014e-06,
      "loss": 0.1745,
      "step": 4510
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.290050534445731,
      "learning_rate": 7.917329755809458e-06,
      "loss": 0.1733,
      "step": 4511
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5044430412753775,
      "learning_rate": 7.916388893775229e-06,
      "loss": 0.2878,
      "step": 4512
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.599422334813816,
      "learning_rate": 7.91544787520083e-06,
      "loss": 0.2655,
      "step": 4513
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2565495235343223,
      "learning_rate": 7.914506700136775e-06,
      "loss": 0.2354,
      "step": 4514
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3105938408365039,
      "learning_rate": 7.913565368633576e-06,
      "loss": 0.1971,
      "step": 4515
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3045700943891678,
      "learning_rate": 7.912623880741762e-06,
      "loss": 0.1938,
      "step": 4516
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.331556175888798,
      "learning_rate": 7.911682236511873e-06,
      "loss": 0.2158,
      "step": 4517
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4155291192537862,
      "learning_rate": 7.910740435994446e-06,
      "loss": 0.2161,
      "step": 4518
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.309813488037048,
      "learning_rate": 7.909798479240038e-06,
      "loss": 0.2254,
      "step": 4519
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3126323049373394,
      "learning_rate": 7.908856366299206e-06,
      "loss": 0.1611,
      "step": 4520
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.34334380472764,
      "learning_rate": 7.907914097222522e-06,
      "loss": 0.4557,
      "step": 4521
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2643675430235892,
      "learning_rate": 7.906971672060562e-06,
      "loss": 0.1811,
      "step": 4522
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3109301655779633,
      "learning_rate": 7.906029090863912e-06,
      "loss": 0.1686,
      "step": 4523
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3316416981307055,
      "learning_rate": 7.905086353683162e-06,
      "loss": 0.1974,
      "step": 4524
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2334280157535553,
      "learning_rate": 7.90414346056892e-06,
      "loss": 0.1724,
      "step": 4525
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.386982235604832,
      "learning_rate": 7.903200411571795e-06,
      "loss": 0.1995,
      "step": 4526
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5810861345088132,
      "learning_rate": 7.902257206742405e-06,
      "loss": 0.2221,
      "step": 4527
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4632766554285028,
      "learning_rate": 7.90131384613138e-06,
      "loss": 0.2758,
      "step": 4528
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2969663486403764,
      "learning_rate": 7.900370329789352e-06,
      "loss": 0.1631,
      "step": 4529
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.361265045295215,
      "learning_rate": 7.899426657766966e-06,
      "loss": 0.2102,
      "step": 4530
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3465527966386448,
      "learning_rate": 7.898482830114878e-06,
      "loss": 0.2478,
      "step": 4531
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.586004046349359,
      "learning_rate": 7.897538846883748e-06,
      "loss": 0.7201,
      "step": 4532
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.435943251686969,
      "learning_rate": 7.896594708124242e-06,
      "loss": 0.2208,
      "step": 4533
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4133963762648394,
      "learning_rate": 7.89565041388704e-06,
      "loss": 0.1993,
      "step": 4534
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.533070930780637,
      "learning_rate": 7.894705964222826e-06,
      "loss": 0.2286,
      "step": 4535
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3409986235406122,
      "learning_rate": 7.893761359182297e-06,
      "loss": 0.1788,
      "step": 4536
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.048763170674597,
      "learning_rate": 7.892816598816155e-06,
      "loss": 0.5688,
      "step": 4537
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2466152956002763,
      "learning_rate": 7.891871683175112e-06,
      "loss": 0.1801,
      "step": 4538
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2224506125160157,
      "learning_rate": 7.890926612309884e-06,
      "loss": 0.1761,
      "step": 4539
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6167614418535203,
      "learning_rate": 7.889981386271202e-06,
      "loss": 0.2408,
      "step": 4540
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3278833914792667,
      "learning_rate": 7.889036005109798e-06,
      "loss": 0.2077,
      "step": 4541
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5177657304852616,
      "learning_rate": 7.888090468876422e-06,
      "loss": 0.2343,
      "step": 4542
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4306884377432565,
      "learning_rate": 7.88714477762182e-06,
      "loss": 0.2323,
      "step": 4543
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2736924401106264,
      "learning_rate": 7.886198931396758e-06,
      "loss": 0.1468,
      "step": 4544
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3046940231608952,
      "learning_rate": 7.885252930252002e-06,
      "loss": 0.2075,
      "step": 4545
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3195738101274577,
      "learning_rate": 7.884306774238336e-06,
      "loss": 0.1682,
      "step": 4546
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4643083128801333,
      "learning_rate": 7.883360463406535e-06,
      "loss": 0.2307,
      "step": 4547
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3938845011705756,
      "learning_rate": 7.882413997807404e-06,
      "loss": 0.2056,
      "step": 4548
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.979216308183173,
      "learning_rate": 7.881467377491738e-06,
      "loss": 0.5259,
      "step": 4549
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3866044063845284,
      "learning_rate": 7.880520602510353e-06,
      "loss": 0.1798,
      "step": 4550
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5197792969642114,
      "learning_rate": 7.879573672914062e-06,
      "loss": 0.2407,
      "step": 4551
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4897730321258797,
      "learning_rate": 7.8786265887537e-06,
      "loss": 0.2742,
      "step": 4552
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.420884554330261,
      "learning_rate": 7.877679350080096e-06,
      "loss": 0.1812,
      "step": 4553
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4602262454738961,
      "learning_rate": 7.876731956944098e-06,
      "loss": 0.233,
      "step": 4554
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4039803370357302,
      "learning_rate": 7.87578440939656e-06,
      "loss": 0.2128,
      "step": 4555
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5613483760983986,
      "learning_rate": 7.874836707488336e-06,
      "loss": 0.2309,
      "step": 4556
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.964179051533017,
      "learning_rate": 7.8738888512703e-06,
      "loss": 0.6407,
      "step": 4557
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2420390396839507,
      "learning_rate": 7.872940840793328e-06,
      "loss": 0.1863,
      "step": 4558
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3584788758460387,
      "learning_rate": 7.871992676108305e-06,
      "loss": 0.2115,
      "step": 4559
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2545314082817982,
      "learning_rate": 7.871044357266124e-06,
      "loss": 0.1917,
      "step": 4560
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.016830104946275,
      "learning_rate": 7.87009588431769e-06,
      "loss": 0.6992,
      "step": 4561
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3867711001481327,
      "learning_rate": 7.869147257313909e-06,
      "loss": 0.205,
      "step": 4562
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3366054887906549,
      "learning_rate": 7.868198476305705e-06,
      "loss": 0.212,
      "step": 4563
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.313308469373452,
      "learning_rate": 7.867249541343997e-06,
      "loss": 0.1852,
      "step": 4564
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4770904645777039,
      "learning_rate": 7.866300452479728e-06,
      "loss": 0.2484,
      "step": 4565
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2539209550778694,
      "learning_rate": 7.865351209763837e-06,
      "loss": 0.1497,
      "step": 4566
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5019930504895755,
      "learning_rate": 7.864401813247276e-06,
      "loss": 0.1874,
      "step": 4567
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.836866351635279,
      "learning_rate": 7.863452262981006e-06,
      "loss": 0.6518,
      "step": 4568
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3652888718031775,
      "learning_rate": 7.862502559015994e-06,
      "loss": 0.2377,
      "step": 4569
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.324201279005588,
      "learning_rate": 7.861552701403218e-06,
      "loss": 0.1833,
      "step": 4570
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2785338879288153,
      "learning_rate": 7.86060269019366e-06,
      "loss": 0.2494,
      "step": 4571
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3962329492726893,
      "learning_rate": 7.859652525438314e-06,
      "loss": 0.1974,
      "step": 4572
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4065724187551483,
      "learning_rate": 7.858702207188183e-06,
      "loss": 0.2068,
      "step": 4573
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.4585713462910705,
      "learning_rate": 7.857751735494272e-06,
      "loss": 0.6454,
      "step": 4574
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.630148422752979,
      "learning_rate": 7.856801110407602e-06,
      "loss": 0.2458,
      "step": 4575
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4243636421074484,
      "learning_rate": 7.855850331979199e-06,
      "loss": 0.2192,
      "step": 4576
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4326285195979387,
      "learning_rate": 7.854899400260094e-06,
      "loss": 0.1993,
      "step": 4577
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3265202734027455,
      "learning_rate": 7.853948315301334e-06,
      "loss": 0.2432,
      "step": 4578
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2278220732696887,
      "learning_rate": 7.852997077153964e-06,
      "loss": 0.1889,
      "step": 4579
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4881727868391827,
      "learning_rate": 7.852045685869046e-06,
      "loss": 0.2092,
      "step": 4580
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5693306002705179,
      "learning_rate": 7.851094141497645e-06,
      "loss": 0.189,
      "step": 4581
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.502471023510673,
      "learning_rate": 7.85014244409084e-06,
      "loss": 0.2488,
      "step": 4582
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.280772540551916,
      "learning_rate": 7.84919059369971e-06,
      "loss": 0.2087,
      "step": 4583
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.220497804121692,
      "learning_rate": 7.848238590375348e-06,
      "loss": 0.6535,
      "step": 4584
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4399212367337306,
      "learning_rate": 7.847286434168853e-06,
      "loss": 0.2293,
      "step": 4585
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.479232273595728,
      "learning_rate": 7.846334125131336e-06,
      "loss": 0.2363,
      "step": 4586
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2934676487345758,
      "learning_rate": 7.84538166331391e-06,
      "loss": 0.1556,
      "step": 4587
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.275788358162262,
      "learning_rate": 7.844429048767702e-06,
      "loss": 0.1907,
      "step": 4588
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.22902838957051,
      "learning_rate": 7.843476281543841e-06,
      "loss": 0.1996,
      "step": 4589
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.105408356676575,
      "learning_rate": 7.84252336169347e-06,
      "loss": 0.1606,
      "step": 4590
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.337862630638302,
      "learning_rate": 7.841570289267738e-06,
      "loss": 0.4815,
      "step": 4591
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.276848463714875,
      "learning_rate": 7.8406170643178e-06,
      "loss": 0.2029,
      "step": 4592
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.153834564616883,
      "learning_rate": 7.839663686894825e-06,
      "loss": 0.7063,
      "step": 4593
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.432277600197333,
      "learning_rate": 7.838710157049985e-06,
      "loss": 0.1743,
      "step": 4594
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3261365291246432,
      "learning_rate": 7.83775647483446e-06,
      "loss": 0.1693,
      "step": 4595
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3728823721554628,
      "learning_rate": 7.836802640299442e-06,
      "loss": 0.2287,
      "step": 4596
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3941663022899486,
      "learning_rate": 7.835848653496129e-06,
      "loss": 0.2308,
      "step": 4597
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2712959913348558,
      "learning_rate": 7.834894514475725e-06,
      "loss": 0.1847,
      "step": 4598
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1684241777978779,
      "learning_rate": 7.833940223289448e-06,
      "loss": 0.1804,
      "step": 4599
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3654626862260337,
      "learning_rate": 7.832985779988518e-06,
      "loss": 0.1906,
      "step": 4600
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4105204397528188,
      "learning_rate": 7.832031184624165e-06,
      "loss": 0.2074,
      "step": 4601
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1032821708609535,
      "learning_rate": 7.83107643724763e-06,
      "loss": 0.1545,
      "step": 4602
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3644410751507485,
      "learning_rate": 7.830121537910158e-06,
      "loss": 0.2136,
      "step": 4603
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4604238950197626,
      "learning_rate": 7.829166486663005e-06,
      "loss": 0.2198,
      "step": 4604
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2273127316468662,
      "learning_rate": 7.828211283557436e-06,
      "loss": 0.1811,
      "step": 4605
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3262955536888117,
      "learning_rate": 7.827255928644721e-06,
      "loss": 0.2259,
      "step": 4606
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1904641910498672,
      "learning_rate": 7.82630042197614e-06,
      "loss": 0.1425,
      "step": 4607
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4375485519569122,
      "learning_rate": 7.825344763602982e-06,
      "loss": 0.2344,
      "step": 4608
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.419910912165066,
      "learning_rate": 7.824388953576539e-06,
      "loss": 0.2269,
      "step": 4609
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4266323744991785,
      "learning_rate": 7.823432991948119e-06,
      "loss": 0.2204,
      "step": 4610
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.4004901939796905,
      "learning_rate": 7.822476878769032e-06,
      "loss": 0.641,
      "step": 4611
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.096382652051679,
      "learning_rate": 7.8215206140906e-06,
      "loss": 0.7195,
      "step": 4612
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4505301139004232,
      "learning_rate": 7.82056419796415e-06,
      "loss": 0.2221,
      "step": 4613
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4106869332709613,
      "learning_rate": 7.81960763044102e-06,
      "loss": 0.2102,
      "step": 4614
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3495953342102984,
      "learning_rate": 7.818650911572553e-06,
      "loss": 0.1709,
      "step": 4615
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4223874091694961,
      "learning_rate": 7.817694041410106e-06,
      "loss": 0.2692,
      "step": 4616
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.120940129272454,
      "learning_rate": 7.816737020005035e-06,
      "loss": 0.1582,
      "step": 4617
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4745310663883042,
      "learning_rate": 7.815779847408711e-06,
      "loss": 0.218,
      "step": 4618
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3128387347753592,
      "learning_rate": 7.81482252367251e-06,
      "loss": 0.2088,
      "step": 4619
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2424369624303544,
      "learning_rate": 7.81386504884782e-06,
      "loss": 0.1734,
      "step": 4620
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.411049224954434,
      "learning_rate": 7.812907422986033e-06,
      "loss": 0.1643,
      "step": 4621
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2902778313369658,
      "learning_rate": 7.81194964613855e-06,
      "loss": 0.2288,
      "step": 4622
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3890905780208214,
      "learning_rate": 7.810991718356781e-06,
      "loss": 0.2134,
      "step": 4623
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3872279940987733,
      "learning_rate": 7.810033639692147e-06,
      "loss": 0.2421,
      "step": 4624
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4755240719102296,
      "learning_rate": 7.809075410196067e-06,
      "loss": 0.2016,
      "step": 4625
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2844593142517982,
      "learning_rate": 7.80811702991998e-06,
      "loss": 0.2184,
      "step": 4626
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.188589907562974,
      "learning_rate": 7.807158498915329e-06,
      "loss": 0.1861,
      "step": 4627
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3521513491386523,
      "learning_rate": 7.80619981723356e-06,
      "loss": 0.2325,
      "step": 4628
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2807056943208444,
      "learning_rate": 7.805240984926134e-06,
      "loss": 0.2034,
      "step": 4629
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6474080972030807,
      "learning_rate": 7.804282002044517e-06,
      "loss": 0.2607,
      "step": 4630
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5412601689299397,
      "learning_rate": 7.803322868640182e-06,
      "loss": 0.2523,
      "step": 4631
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.822059582538499,
      "learning_rate": 7.802363584764613e-06,
      "loss": 0.2321,
      "step": 4632
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3025015177536587,
      "learning_rate": 7.8014041504693e-06,
      "loss": 0.1839,
      "step": 4633
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3294454676960608,
      "learning_rate": 7.80044456580574e-06,
      "loss": 0.1626,
      "step": 4634
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.347201365237797,
      "learning_rate": 7.799484830825443e-06,
      "loss": 0.2122,
      "step": 4635
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.91421616385361,
      "learning_rate": 7.798524945579922e-06,
      "loss": 0.4876,
      "step": 4636
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.346536420197388,
      "learning_rate": 7.7975649101207e-06,
      "loss": 0.2262,
      "step": 4637
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3037138629132499,
      "learning_rate": 7.796604724499306e-06,
      "loss": 0.1921,
      "step": 4638
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3549954527577046,
      "learning_rate": 7.795644388767283e-06,
      "loss": 0.2078,
      "step": 4639
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5127267773769109,
      "learning_rate": 7.794683902976175e-06,
      "loss": 0.2413,
      "step": 4640
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4248220359700412,
      "learning_rate": 7.793723267177538e-06,
      "loss": 0.2114,
      "step": 4641
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.277380730332711,
      "learning_rate": 7.792762481422933e-06,
      "loss": 0.5926,
      "step": 4642
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.44824350460334,
      "learning_rate": 7.791801545763934e-06,
      "loss": 0.7087,
      "step": 4643
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.495388866003692,
      "learning_rate": 7.790840460252121e-06,
      "loss": 0.2244,
      "step": 4644
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4371861560000003,
      "learning_rate": 7.789879224939078e-06,
      "loss": 0.246,
      "step": 4645
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3935489834158172,
      "learning_rate": 7.788917839876402e-06,
      "loss": 0.2361,
      "step": 4646
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.416779449322759,
      "learning_rate": 7.787956305115696e-06,
      "loss": 0.1843,
      "step": 4647
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4125387282742232,
      "learning_rate": 7.786994620708572e-06,
      "loss": 0.2139,
      "step": 4648
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1033852742707633,
      "learning_rate": 7.786032786706648e-06,
      "loss": 0.1402,
      "step": 4649
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3597511933142212,
      "learning_rate": 7.785070803161552e-06,
      "loss": 0.1829,
      "step": 4650
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1907186733056092,
      "learning_rate": 7.784108670124921e-06,
      "loss": 0.1561,
      "step": 4651
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.30763833288827,
      "learning_rate": 7.783146387648397e-06,
      "loss": 0.2175,
      "step": 4652
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2270376205196245,
      "learning_rate": 7.782183955783633e-06,
      "loss": 0.1577,
      "step": 4653
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5646695070806511,
      "learning_rate": 7.781221374582284e-06,
      "loss": 0.2265,
      "step": 4654
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2820686465104565,
      "learning_rate": 7.780258644096024e-06,
      "loss": 0.2394,
      "step": 4655
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4805879866501594,
      "learning_rate": 7.779295764376524e-06,
      "loss": 0.2045,
      "step": 4656
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.439745454232342,
      "learning_rate": 7.77833273547547e-06,
      "loss": 0.1985,
      "step": 4657
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4349080124112894,
      "learning_rate": 7.77736955744455e-06,
      "loss": 0.237,
      "step": 4658
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2474915637670567,
      "learning_rate": 7.776406230335469e-06,
      "loss": 0.183,
      "step": 4659
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5339912123169865,
      "learning_rate": 7.775442754199929e-06,
      "loss": 0.2519,
      "step": 4660
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3489278063475347,
      "learning_rate": 7.77447912908965e-06,
      "loss": 0.2419,
      "step": 4661
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3182120957097267,
      "learning_rate": 7.773515355056354e-06,
      "loss": 0.2106,
      "step": 4662
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3401895304346232,
      "learning_rate": 7.772551432151771e-06,
      "loss": 0.1999,
      "step": 4663
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5535310494308856,
      "learning_rate": 7.771587360427642e-06,
      "loss": 0.1939,
      "step": 4664
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3822008916216864,
      "learning_rate": 7.770623139935716e-06,
      "loss": 0.2231,
      "step": 4665
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3795615224891262,
      "learning_rate": 7.769658770727745e-06,
      "loss": 0.238,
      "step": 4666
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4620561520826667,
      "learning_rate": 7.768694252855495e-06,
      "loss": 0.2123,
      "step": 4667
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3518772674923205,
      "learning_rate": 7.767729586370738e-06,
      "loss": 0.187,
      "step": 4668
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1649761230185158,
      "learning_rate": 7.76676477132525e-06,
      "loss": 0.1477,
      "step": 4669
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.465256565603295,
      "learning_rate": 7.765799807770824e-06,
      "loss": 0.2208,
      "step": 4670
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.386945854291886,
      "learning_rate": 7.764834695759251e-06,
      "loss": 0.2204,
      "step": 4671
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2121196514853203,
      "learning_rate": 7.763869435342335e-06,
      "loss": 0.1797,
      "step": 4672
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5221129127495716,
      "learning_rate": 7.762904026571889e-06,
      "loss": 0.2141,
      "step": 4673
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5241753922235284,
      "learning_rate": 7.76193846949973e-06,
      "loss": 0.2699,
      "step": 4674
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.778930221261666,
      "learning_rate": 7.760972764177688e-06,
      "loss": 0.6438,
      "step": 4675
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5107383821748674,
      "learning_rate": 7.760006910657597e-06,
      "loss": 0.24,
      "step": 4676
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5143363602372204,
      "learning_rate": 7.759040908991297e-06,
      "loss": 0.2391,
      "step": 4677
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.4099940483474835,
      "learning_rate": 7.758074759230645e-06,
      "loss": 0.7578,
      "step": 4678
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3891058934766316,
      "learning_rate": 7.757108461427496e-06,
      "loss": 0.2182,
      "step": 4679
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.116495119673267,
      "learning_rate": 7.75614201563372e-06,
      "loss": 0.6878,
      "step": 4680
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3542336247691311,
      "learning_rate": 7.755175421901187e-06,
      "loss": 0.1645,
      "step": 4681
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.277862209020888,
      "learning_rate": 7.754208680281784e-06,
      "loss": 0.2081,
      "step": 4682
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.467077621457075,
      "learning_rate": 7.753241790827401e-06,
      "loss": 0.2276,
      "step": 4683
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4178612377713071,
      "learning_rate": 7.752274753589939e-06,
      "loss": 0.2288,
      "step": 4684
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2707158965681835,
      "learning_rate": 7.7513075686213e-06,
      "loss": 0.1798,
      "step": 4685
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4861475442673708,
      "learning_rate": 7.750340235973401e-06,
      "loss": 0.2341,
      "step": 4686
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4580276865070532,
      "learning_rate": 7.749372755698166e-06,
      "loss": 0.2155,
      "step": 4687
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.313186658018327,
      "learning_rate": 7.748405127847525e-06,
      "loss": 0.1932,
      "step": 4688
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.708257396340362,
      "learning_rate": 7.747437352473414e-06,
      "loss": 0.2348,
      "step": 4689
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2572847058322358,
      "learning_rate": 7.74646942962778e-06,
      "loss": 0.2156,
      "step": 4690
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3831811530875395,
      "learning_rate": 7.74550135936258e-06,
      "loss": 0.2012,
      "step": 4691
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4782713571556692,
      "learning_rate": 7.744533141729773e-06,
      "loss": 0.1928,
      "step": 4692
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6087742949685924,
      "learning_rate": 7.743564776781332e-06,
      "loss": 0.2297,
      "step": 4693
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.319235132797053,
      "learning_rate": 7.742596264569232e-06,
      "loss": 0.1819,
      "step": 4694
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5307696053948605,
      "learning_rate": 7.74162760514546e-06,
      "loss": 0.206,
      "step": 4695
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.290977741433551,
      "learning_rate": 7.74065879856201e-06,
      "loss": 0.6766,
      "step": 4696
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5126675736856585,
      "learning_rate": 7.739689844870885e-06,
      "loss": 0.204,
      "step": 4697
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4653002924408978,
      "learning_rate": 7.738720744124093e-06,
      "loss": 0.2388,
      "step": 4698
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.90522809641348,
      "learning_rate": 7.737751496373652e-06,
      "loss": 0.7529,
      "step": 4699
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.747649106380913,
      "learning_rate": 7.736782101671587e-06,
      "loss": 0.5075,
      "step": 4700
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.422438147751286,
      "learning_rate": 7.73581256006993e-06,
      "loss": 0.1962,
      "step": 4701
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.332620916718465,
      "learning_rate": 7.734842871620725e-06,
      "loss": 0.1919,
      "step": 4702
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5814793775956724,
      "learning_rate": 7.733873036376021e-06,
      "loss": 0.234,
      "step": 4703
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2915242772272948,
      "learning_rate": 7.732903054387872e-06,
      "loss": 0.2102,
      "step": 4704
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3007399495509682,
      "learning_rate": 7.731932925708343e-06,
      "loss": 0.1809,
      "step": 4705
    },
    {
      "epoch": 0.34,
      "grad_norm": 11.620738510847799,
      "learning_rate": 7.73096265038951e-06,
      "loss": 0.4606,
      "step": 4706
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5156201529372535,
      "learning_rate": 7.72999222848345e-06,
      "loss": 0.2464,
      "step": 4707
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5030260185893738,
      "learning_rate": 7.729021660042254e-06,
      "loss": 0.2317,
      "step": 4708
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.423551542191018,
      "learning_rate": 7.728050945118017e-06,
      "loss": 0.2266,
      "step": 4709
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3631395992829514,
      "learning_rate": 7.727080083762844e-06,
      "loss": 0.2004,
      "step": 4710
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.7113638193465475,
      "learning_rate": 7.726109076028846e-06,
      "loss": 0.5942,
      "step": 4711
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3866158352213205,
      "learning_rate": 7.72513792196814e-06,
      "loss": 0.1941,
      "step": 4712
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.321018366837503,
      "learning_rate": 7.724166621632859e-06,
      "loss": 0.2031,
      "step": 4713
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.344041606929519,
      "learning_rate": 7.723195175075136e-06,
      "loss": 0.2091,
      "step": 4714
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.313191087104459,
      "learning_rate": 7.722223582347115e-06,
      "loss": 0.2147,
      "step": 4715
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.181209704056137,
      "learning_rate": 7.721251843500948e-06,
      "loss": 0.5853,
      "step": 4716
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3420655405288728,
      "learning_rate": 7.720279958588791e-06,
      "loss": 0.219,
      "step": 4717
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4515739332436,
      "learning_rate": 7.719307927662813e-06,
      "loss": 0.2124,
      "step": 4718
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2212428798120076,
      "learning_rate": 7.71833575077519e-06,
      "loss": 0.1587,
      "step": 4719
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4844113039597002,
      "learning_rate": 7.717363427978103e-06,
      "loss": 0.2062,
      "step": 4720
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.410865759148572,
      "learning_rate": 7.716390959323743e-06,
      "loss": 0.2498,
      "step": 4721
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2130805585552855,
      "learning_rate": 7.715418344864306e-06,
      "loss": 0.1841,
      "step": 4722
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.438362118314582,
      "learning_rate": 7.714445584652001e-06,
      "loss": 0.2043,
      "step": 4723
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.99252549522735,
      "learning_rate": 7.713472678739042e-06,
      "loss": 0.4975,
      "step": 4724
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3662144993163154,
      "learning_rate": 7.712499627177648e-06,
      "loss": 0.199,
      "step": 4725
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.471292440398072,
      "learning_rate": 7.711526430020052e-06,
      "loss": 0.2291,
      "step": 4726
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.868256489088615,
      "learning_rate": 7.710553087318489e-06,
      "loss": 0.5817,
      "step": 4727
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3954135172494164,
      "learning_rate": 7.709579599125205e-06,
      "loss": 0.2368,
      "step": 4728
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.463593919186997,
      "learning_rate": 7.708605965492451e-06,
      "loss": 0.2477,
      "step": 4729
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.428317301164114,
      "learning_rate": 7.70763218647249e-06,
      "loss": 0.2102,
      "step": 4730
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3068161816616652,
      "learning_rate": 7.706658262117592e-06,
      "loss": 0.2075,
      "step": 4731
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4412562126913349,
      "learning_rate": 7.705684192480029e-06,
      "loss": 0.201,
      "step": 4732
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4737874705637126,
      "learning_rate": 7.70470997761209e-06,
      "loss": 0.1997,
      "step": 4733
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.523453737076902,
      "learning_rate": 7.703735617566063e-06,
      "loss": 0.2544,
      "step": 4734
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3015823117521228,
      "learning_rate": 7.70276111239425e-06,
      "loss": 0.1964,
      "step": 4735
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2820393677384432,
      "learning_rate": 7.701786462148958e-06,
      "loss": 0.1702,
      "step": 4736
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4204634617817746,
      "learning_rate": 7.700811666882501e-06,
      "loss": 0.192,
      "step": 4737
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4655529705751784,
      "learning_rate": 7.699836726647206e-06,
      "loss": 0.2383,
      "step": 4738
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3828304068236836,
      "learning_rate": 7.6988616414954e-06,
      "loss": 0.2036,
      "step": 4739
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3099561276681655,
      "learning_rate": 7.697886411479422e-06,
      "loss": 0.1854,
      "step": 4740
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5053759040935586,
      "learning_rate": 7.696911036651621e-06,
      "loss": 0.2496,
      "step": 4741
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5182686435592552,
      "learning_rate": 7.695935517064352e-06,
      "loss": 0.1943,
      "step": 4742
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4235657024967352,
      "learning_rate": 7.694959852769972e-06,
      "loss": 0.2272,
      "step": 4743
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2952225674386235,
      "learning_rate": 7.693984043820852e-06,
      "loss": 0.2159,
      "step": 4744
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2622448496007699,
      "learning_rate": 7.693008090269374e-06,
      "loss": 0.173,
      "step": 4745
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2482007353236018,
      "learning_rate": 7.692031992167921e-06,
      "loss": 0.1932,
      "step": 4746
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.328284774859831,
      "learning_rate": 7.691055749568885e-06,
      "loss": 0.211,
      "step": 4747
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2402447991480978,
      "learning_rate": 7.690079362524666e-06,
      "loss": 0.1489,
      "step": 4748
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4901467009292249,
      "learning_rate": 7.689102831087673e-06,
      "loss": 0.2525,
      "step": 4749
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.878067706926046,
      "learning_rate": 7.688126155310326e-06,
      "loss": 0.6885,
      "step": 4750
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3817010171751853,
      "learning_rate": 7.687149335245046e-06,
      "loss": 0.2032,
      "step": 4751
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3808639596573962,
      "learning_rate": 7.686172370944262e-06,
      "loss": 0.2068,
      "step": 4752
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3069442104023203,
      "learning_rate": 7.68519526246042e-06,
      "loss": 0.2101,
      "step": 4753
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3325687667717678,
      "learning_rate": 7.684218009845963e-06,
      "loss": 0.1952,
      "step": 4754
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5643501164981082,
      "learning_rate": 7.683240613153346e-06,
      "loss": 0.2434,
      "step": 4755
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.5634315246853765,
      "learning_rate": 7.682263072435035e-06,
      "loss": 0.5861,
      "step": 4756
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.338700688562351,
      "learning_rate": 7.681285387743495e-06,
      "loss": 0.1885,
      "step": 4757
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2621647667930365,
      "learning_rate": 7.680307559131211e-06,
      "loss": 0.1917,
      "step": 4758
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.284168673309812,
      "learning_rate": 7.679329586650666e-06,
      "loss": 0.5075,
      "step": 4759
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5555761774869115,
      "learning_rate": 7.67835147035435e-06,
      "loss": 0.2592,
      "step": 4760
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4112778162698283,
      "learning_rate": 7.677373210294769e-06,
      "loss": 0.2775,
      "step": 4761
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.227435460165308,
      "learning_rate": 7.676394806524432e-06,
      "loss": 0.1941,
      "step": 4762
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3303070283760783,
      "learning_rate": 7.675416259095854e-06,
      "loss": 0.2043,
      "step": 4763
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.359128076090484,
      "learning_rate": 7.674437568061559e-06,
      "loss": 0.2013,
      "step": 4764
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2868546486487915,
      "learning_rate": 7.673458733474081e-06,
      "loss": 0.1952,
      "step": 4765
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4590554233493203,
      "learning_rate": 7.672479755385958e-06,
      "loss": 0.2081,
      "step": 4766
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.418259566522473,
      "learning_rate": 7.67150063384974e-06,
      "loss": 0.1737,
      "step": 4767
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5189724354844958,
      "learning_rate": 7.670521368917983e-06,
      "loss": 0.1758,
      "step": 4768
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.958703495896905,
      "learning_rate": 7.669541960643246e-06,
      "loss": 0.5713,
      "step": 4769
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3350124994230945,
      "learning_rate": 7.668562409078102e-06,
      "loss": 0.2308,
      "step": 4770
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3516542855167446,
      "learning_rate": 7.66758271427513e-06,
      "loss": 0.1947,
      "step": 4771
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2413588946742415,
      "learning_rate": 7.666602876286917e-06,
      "loss": 0.1946,
      "step": 4772
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.213847339434656,
      "learning_rate": 7.665622895166054e-06,
      "loss": 0.1356,
      "step": 4773
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3719635367607725,
      "learning_rate": 7.664642770965144e-06,
      "loss": 0.2083,
      "step": 4774
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4746489253933357,
      "learning_rate": 7.663662503736796e-06,
      "loss": 0.2234,
      "step": 4775
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4277292171734157,
      "learning_rate": 7.662682093533628e-06,
      "loss": 0.1991,
      "step": 4776
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2535939652339498,
      "learning_rate": 7.661701540408263e-06,
      "loss": 0.1693,
      "step": 4777
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.291477683063257,
      "learning_rate": 7.660720844413333e-06,
      "loss": 0.5333,
      "step": 4778
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4080758817313526,
      "learning_rate": 7.65974000560148e-06,
      "loss": 0.2139,
      "step": 4779
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4034870364239027,
      "learning_rate": 7.658759024025349e-06,
      "loss": 0.2346,
      "step": 4780
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2476731993490076,
      "learning_rate": 7.657777899737596e-06,
      "loss": 0.1847,
      "step": 4781
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.353810377694333,
      "learning_rate": 7.656796632790885e-06,
      "loss": 0.6565,
      "step": 4782
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3285412345757042,
      "learning_rate": 7.655815223237884e-06,
      "loss": 0.1629,
      "step": 4783
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.380001667541261,
      "learning_rate": 7.654833671131274e-06,
      "loss": 0.2283,
      "step": 4784
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3471407908316917,
      "learning_rate": 7.65385197652374e-06,
      "loss": 0.2059,
      "step": 4785
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4080061119496534,
      "learning_rate": 7.652870139467975e-06,
      "loss": 0.2357,
      "step": 4786
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3394925069781158,
      "learning_rate": 7.65188816001668e-06,
      "loss": 0.2184,
      "step": 4787
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4593354599939896,
      "learning_rate": 7.650906038222563e-06,
      "loss": 0.207,
      "step": 4788
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.812776945242237,
      "learning_rate": 7.649923774138344e-06,
      "loss": 0.6408,
      "step": 4789
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3236409211252762,
      "learning_rate": 7.648941367816742e-06,
      "loss": 0.1794,
      "step": 4790
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.545015814424405,
      "learning_rate": 7.647958819310491e-06,
      "loss": 0.2179,
      "step": 4791
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3692024652124706,
      "learning_rate": 7.646976128672332e-06,
      "loss": 0.1866,
      "step": 4792
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5617845371835737,
      "learning_rate": 7.64599329595501e-06,
      "loss": 0.253,
      "step": 4793
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2905704416991175,
      "learning_rate": 7.64501032121128e-06,
      "loss": 0.1779,
      "step": 4794
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3308460365962715,
      "learning_rate": 7.644027204493902e-06,
      "loss": 0.2042,
      "step": 4795
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3124957898029632,
      "learning_rate": 7.643043945855648e-06,
      "loss": 0.2094,
      "step": 4796
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5019655768267048,
      "learning_rate": 7.642060545349296e-06,
      "loss": 0.2519,
      "step": 4797
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.962778050704384,
      "learning_rate": 7.641077003027632e-06,
      "loss": 0.5218,
      "step": 4798
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3518032245550393,
      "learning_rate": 7.640093318943445e-06,
      "loss": 0.2273,
      "step": 4799
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3206915323461959,
      "learning_rate": 7.639109493149537e-06,
      "loss": 0.191,
      "step": 4800
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4623275872791461,
      "learning_rate": 7.638125525698717e-06,
      "loss": 0.2079,
      "step": 4801
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2915364454995029,
      "learning_rate": 7.637141416643801e-06,
      "loss": 0.1795,
      "step": 4802
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.445577362855475,
      "learning_rate": 7.636157166037608e-06,
      "loss": 0.4998,
      "step": 4803
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4853517079752392,
      "learning_rate": 7.635172773932972e-06,
      "loss": 0.2161,
      "step": 4804
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.349138158557549,
      "learning_rate": 7.63418824038273e-06,
      "loss": 0.6782,
      "step": 4805
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4365065464431361,
      "learning_rate": 7.633203565439729e-06,
      "loss": 0.2246,
      "step": 4806
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2390027699622674,
      "learning_rate": 7.632218749156822e-06,
      "loss": 0.1683,
      "step": 4807
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.290893284934911,
      "learning_rate": 7.63123379158687e-06,
      "loss": 0.1888,
      "step": 4808
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4570974272516766,
      "learning_rate": 7.63024869278274e-06,
      "loss": 0.2488,
      "step": 4809
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.375698858780777,
      "learning_rate": 7.629263452797311e-06,
      "loss": 0.2316,
      "step": 4810
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5072108475588435,
      "learning_rate": 7.628278071683465e-06,
      "loss": 0.1793,
      "step": 4811
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.526017044000886,
      "learning_rate": 7.627292549494092e-06,
      "loss": 0.713,
      "step": 4812
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3093291958536024,
      "learning_rate": 7.626306886282096e-06,
      "loss": 0.19,
      "step": 4813
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.662294652638992,
      "learning_rate": 7.6253210821003765e-06,
      "loss": 0.5742,
      "step": 4814
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.563646496477359,
      "learning_rate": 7.624335137001854e-06,
      "loss": 0.1871,
      "step": 4815
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2678218628667255,
      "learning_rate": 7.623349051039446e-06,
      "loss": 0.1952,
      "step": 4816
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4776598187788588,
      "learning_rate": 7.622362824266083e-06,
      "loss": 0.1878,
      "step": 4817
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3459846639995374,
      "learning_rate": 7.621376456734702e-06,
      "loss": 0.1907,
      "step": 4818
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5095567454306926,
      "learning_rate": 7.620389948498245e-06,
      "loss": 0.2214,
      "step": 4819
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3728387013238597,
      "learning_rate": 7.6194032996096685e-06,
      "loss": 0.1911,
      "step": 4820
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.558272752468376,
      "learning_rate": 7.618416510121928e-06,
      "loss": 0.6037,
      "step": 4821
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4598006541076647,
      "learning_rate": 7.617429580087992e-06,
      "loss": 0.2251,
      "step": 4822
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2905467003056512,
      "learning_rate": 7.616442509560835e-06,
      "loss": 0.2051,
      "step": 4823
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2464350907294492,
      "learning_rate": 7.6154552985934385e-06,
      "loss": 0.2223,
      "step": 4824
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2923541791153246,
      "learning_rate": 7.614467947238791e-06,
      "loss": 0.187,
      "step": 4825
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2937071697459692,
      "learning_rate": 7.613480455549892e-06,
      "loss": 0.181,
      "step": 4826
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3702656641539492,
      "learning_rate": 7.612492823579744e-06,
      "loss": 0.1843,
      "step": 4827
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3470974453348925,
      "learning_rate": 7.611505051381363e-06,
      "loss": 0.2037,
      "step": 4828
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4379575616220603,
      "learning_rate": 7.610517139007763e-06,
      "loss": 0.2116,
      "step": 4829
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.681876417899939,
      "learning_rate": 7.609529086511974e-06,
      "loss": 0.487,
      "step": 4830
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.303871482302147,
      "learning_rate": 7.608540893947032e-06,
      "loss": 0.195,
      "step": 4831
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4187441867158963,
      "learning_rate": 7.6075525613659775e-06,
      "loss": 0.2202,
      "step": 4832
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4298521316600983,
      "learning_rate": 7.6065640888218595e-06,
      "loss": 0.1985,
      "step": 4833
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.6041872257246235,
      "learning_rate": 7.605575476367739e-06,
      "loss": 0.5279,
      "step": 4834
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2670522491520646,
      "learning_rate": 7.604586724056677e-06,
      "loss": 0.1873,
      "step": 4835
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5206554895647546,
      "learning_rate": 7.603597831941747e-06,
      "loss": 0.2176,
      "step": 4836
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4900278091852655,
      "learning_rate": 7.6026088000760276e-06,
      "loss": 0.2839,
      "step": 4837
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3689246018042849,
      "learning_rate": 7.601619628512611e-06,
      "loss": 0.2406,
      "step": 4838
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.130549102392845,
      "learning_rate": 7.600630317304586e-06,
      "loss": 0.6326,
      "step": 4839
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4184896893429064,
      "learning_rate": 7.599640866505058e-06,
      "loss": 0.2299,
      "step": 4840
    },
    {
      "epoch": 0.35,
      "grad_norm": 12.325807647443925,
      "learning_rate": 7.598651276167135e-06,
      "loss": 0.6177,
      "step": 4841
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.344380938165053,
      "learning_rate": 7.597661546343936e-06,
      "loss": 0.1916,
      "step": 4842
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.47371052658536,
      "learning_rate": 7.596671677088585e-06,
      "loss": 0.1702,
      "step": 4843
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.760704218726979,
      "learning_rate": 7.5956816684542154e-06,
      "loss": 0.2053,
      "step": 4844
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5055142206570484,
      "learning_rate": 7.594691520493965e-06,
      "loss": 0.2568,
      "step": 4845
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.459961605910439,
      "learning_rate": 7.593701233260983e-06,
      "loss": 0.1919,
      "step": 4846
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.468617622570324,
      "learning_rate": 7.592710806808422e-06,
      "loss": 0.2301,
      "step": 4847
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2709344896206303,
      "learning_rate": 7.591720241189446e-06,
      "loss": 0.1909,
      "step": 4848
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2501506566761205,
      "learning_rate": 7.590729536457222e-06,
      "loss": 0.1912,
      "step": 4849
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.271986900844157,
      "learning_rate": 7.58973869266493e-06,
      "loss": 0.1704,
      "step": 4850
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.528409270726896,
      "learning_rate": 7.588747709865753e-06,
      "loss": 0.2483,
      "step": 4851
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.316591437389964,
      "learning_rate": 7.587756588112884e-06,
      "loss": 0.2133,
      "step": 4852
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3276773688122625,
      "learning_rate": 7.586765327459523e-06,
      "loss": 0.222,
      "step": 4853
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.175065077542618,
      "learning_rate": 7.585773927958875e-06,
      "loss": 0.2033,
      "step": 4854
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.171127420129231,
      "learning_rate": 7.5847823896641545e-06,
      "loss": 0.1816,
      "step": 4855
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3604538157774908,
      "learning_rate": 7.583790712628585e-06,
      "loss": 0.2668,
      "step": 4856
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2119980473902272,
      "learning_rate": 7.582798896905396e-06,
      "loss": 0.1755,
      "step": 4857
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.454400368314169,
      "learning_rate": 7.58180694254782e-06,
      "loss": 0.1672,
      "step": 4858
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.368637701492504,
      "learning_rate": 7.580814849609107e-06,
      "loss": 0.2037,
      "step": 4859
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.79332036474255,
      "learning_rate": 7.579822618142505e-06,
      "loss": 0.7615,
      "step": 4860
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3204940761501438,
      "learning_rate": 7.578830248201274e-06,
      "loss": 0.2147,
      "step": 4861
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2068251132945396,
      "learning_rate": 7.5778377398386815e-06,
      "loss": 0.1884,
      "step": 4862
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.962422720932961,
      "learning_rate": 7.576845093107998e-06,
      "loss": 0.5353,
      "step": 4863
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3713051258115976,
      "learning_rate": 7.575852308062509e-06,
      "loss": 0.24,
      "step": 4864
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3252313777225955,
      "learning_rate": 7.574859384755502e-06,
      "loss": 0.1787,
      "step": 4865
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5692679421811029,
      "learning_rate": 7.5738663232402706e-06,
      "loss": 0.2674,
      "step": 4866
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3828773822083427,
      "learning_rate": 7.572873123570122e-06,
      "loss": 0.202,
      "step": 4867
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7611567492044915,
      "learning_rate": 7.571879785798364e-06,
      "loss": 0.2849,
      "step": 4868
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2841391409561893,
      "learning_rate": 7.570886309978318e-06,
      "loss": 0.21,
      "step": 4869
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2021566397634278,
      "learning_rate": 7.569892696163308e-06,
      "loss": 0.1678,
      "step": 4870
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3162514003607486,
      "learning_rate": 7.568898944406667e-06,
      "loss": 0.195,
      "step": 4871
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.507754483745572,
      "learning_rate": 7.5679050547617375e-06,
      "loss": 0.1845,
      "step": 4872
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2782065369795192,
      "learning_rate": 7.566911027281867e-06,
      "loss": 0.1817,
      "step": 4873
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3476634111381744,
      "learning_rate": 7.565916862020408e-06,
      "loss": 0.1991,
      "step": 4874
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.068654419698149,
      "learning_rate": 7.564922559030727e-06,
      "loss": 0.6658,
      "step": 4875
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3277446711224548,
      "learning_rate": 7.563928118366193e-06,
      "loss": 0.1878,
      "step": 4876
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.226406401741414,
      "learning_rate": 7.562933540080184e-06,
      "loss": 0.6546,
      "step": 4877
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2223421587364767,
      "learning_rate": 7.561938824226085e-06,
      "loss": 0.1844,
      "step": 4878
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5586431550918087,
      "learning_rate": 7.560943970857286e-06,
      "loss": 0.2188,
      "step": 4879
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.991888915391443,
      "learning_rate": 7.559948980027189e-06,
      "loss": 0.4627,
      "step": 4880
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2109543725805108,
      "learning_rate": 7.558953851789202e-06,
      "loss": 0.1307,
      "step": 4881
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5019326719899215,
      "learning_rate": 7.5579585861967384e-06,
      "loss": 0.2221,
      "step": 4882
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4867330949771582,
      "learning_rate": 7.55696318330322e-06,
      "loss": 0.2136,
      "step": 4883
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3573556113783107,
      "learning_rate": 7.5559676431620745e-06,
      "loss": 0.1958,
      "step": 4884
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.351894349261488,
      "learning_rate": 7.554971965826743e-06,
      "loss": 0.2218,
      "step": 4885
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1705583896548382,
      "learning_rate": 7.553976151350666e-06,
      "loss": 0.1662,
      "step": 4886
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4584942278928406,
      "learning_rate": 7.5529801997872945e-06,
      "loss": 0.2114,
      "step": 4887
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3951963575091393,
      "learning_rate": 7.551984111190088e-06,
      "loss": 0.2144,
      "step": 4888
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4317257168999753,
      "learning_rate": 7.550987885612514e-06,
      "loss": 0.2462,
      "step": 4889
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3874513555107229,
      "learning_rate": 7.5499915231080445e-06,
      "loss": 0.1742,
      "step": 4890
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4657648796551586,
      "learning_rate": 7.548995023730161e-06,
      "loss": 0.2518,
      "step": 4891
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1510718228769996,
      "learning_rate": 7.5479983875323515e-06,
      "loss": 0.193,
      "step": 4892
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4221449394956425,
      "learning_rate": 7.54700161456811e-06,
      "loss": 0.2088,
      "step": 4893
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4121771519367425,
      "learning_rate": 7.546004704890941e-06,
      "loss": 0.1969,
      "step": 4894
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3725266277942887,
      "learning_rate": 7.545007658554355e-06,
      "loss": 0.1933,
      "step": 4895
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3565179177323987,
      "learning_rate": 7.544010475611868e-06,
      "loss": 0.1809,
      "step": 4896
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.43429813419383,
      "learning_rate": 7.543013156117005e-06,
      "loss": 0.234,
      "step": 4897
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4277102171831735,
      "learning_rate": 7.5420157001233e-06,
      "loss": 0.2495,
      "step": 4898
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2805273655553038,
      "learning_rate": 7.54101810768429e-06,
      "loss": 0.1747,
      "step": 4899
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2491244302454414,
      "learning_rate": 7.540020378853523e-06,
      "loss": 0.2092,
      "step": 4900
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4530146738132,
      "learning_rate": 7.539022513684554e-06,
      "loss": 0.2364,
      "step": 4901
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1580658916765,
      "learning_rate": 7.538024512230942e-06,
      "loss": 0.1536,
      "step": 4902
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2738472211245093,
      "learning_rate": 7.537026374546259e-06,
      "loss": 0.2019,
      "step": 4903
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.27540036808571,
      "learning_rate": 7.536028100684078e-06,
      "loss": 0.2428,
      "step": 4904
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3213004179854244,
      "learning_rate": 7.535029690697984e-06,
      "loss": 0.2214,
      "step": 4905
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4292236653344172,
      "learning_rate": 7.534031144641567e-06,
      "loss": 0.2201,
      "step": 4906
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3656069957100705,
      "learning_rate": 7.533032462568426e-06,
      "loss": 0.2183,
      "step": 4907
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3169538418677602,
      "learning_rate": 7.532033644532166e-06,
      "loss": 0.2135,
      "step": 4908
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.842073703047555,
      "learning_rate": 7.531034690586397e-06,
      "loss": 0.6519,
      "step": 4909
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6822239306366573,
      "learning_rate": 7.530035600784743e-06,
      "loss": 0.2638,
      "step": 4910
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1971238627600487,
      "learning_rate": 7.529036375180829e-06,
      "loss": 0.159,
      "step": 4911
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4499414260005141,
      "learning_rate": 7.528037013828289e-06,
      "loss": 0.2585,
      "step": 4912
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3732065734740053,
      "learning_rate": 7.527037516780766e-06,
      "loss": 0.2526,
      "step": 4913
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1866339609962981,
      "learning_rate": 7.5260378840919075e-06,
      "loss": 0.1833,
      "step": 4914
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.2923380895358165,
      "learning_rate": 7.525038115815372e-06,
      "loss": 0.1872,
      "step": 4915
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4692185236673518,
      "learning_rate": 7.524038212004822e-06,
      "loss": 0.1853,
      "step": 4916
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.366637234095254,
      "learning_rate": 7.523038172713928e-06,
      "loss": 0.2053,
      "step": 4917
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2856167046513394,
      "learning_rate": 7.522037997996367e-06,
      "loss": 0.2188,
      "step": 4918
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.594769744316143,
      "learning_rate": 7.521037687905828e-06,
      "loss": 0.5826,
      "step": 4919
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3259501384597088,
      "learning_rate": 7.520037242496e-06,
      "loss": 0.1832,
      "step": 4920
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4562357405237034,
      "learning_rate": 7.519036661820585e-06,
      "loss": 0.2462,
      "step": 4921
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3075853379762876,
      "learning_rate": 7.518035945933289e-06,
      "loss": 0.1945,
      "step": 4922
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.353088020037506,
      "learning_rate": 7.5170350948878276e-06,
      "loss": 0.2283,
      "step": 4923
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.274677746850781,
      "learning_rate": 7.516034108737922e-06,
      "loss": 0.1893,
      "step": 4924
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3695201697438923,
      "learning_rate": 7.5150329875373005e-06,
      "loss": 0.2687,
      "step": 4925
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1606398021160842,
      "learning_rate": 7.5140317313397004e-06,
      "loss": 0.157,
      "step": 4926
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4315485020922114,
      "learning_rate": 7.513030340198865e-06,
      "loss": 0.2356,
      "step": 4927
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.163736619001254,
      "learning_rate": 7.512028814168543e-06,
      "loss": 0.1776,
      "step": 4928
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4943158448350171,
      "learning_rate": 7.511027153302497e-06,
      "loss": 0.234,
      "step": 4929
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.291155598321692,
      "learning_rate": 7.510025357654487e-06,
      "loss": 0.2008,
      "step": 4930
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4704581664077705,
      "learning_rate": 7.509023427278288e-06,
      "loss": 0.1824,
      "step": 4931
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3906552140004318,
      "learning_rate": 7.50802136222768e-06,
      "loss": 0.2197,
      "step": 4932
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.435666578778386,
      "learning_rate": 7.507019162556448e-06,
      "loss": 0.1937,
      "step": 4933
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.519805774388088,
      "learning_rate": 7.506016828318387e-06,
      "loss": 0.2248,
      "step": 4934
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3163819304563045,
      "learning_rate": 7.5050143595673e-06,
      "loss": 0.1942,
      "step": 4935
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2351041680437558,
      "learning_rate": 7.504011756356993e-06,
      "loss": 0.1595,
      "step": 4936
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.346938297916784,
      "learning_rate": 7.503009018741284e-06,
      "loss": 0.2252,
      "step": 4937
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3211131714154514,
      "learning_rate": 7.502006146773994e-06,
      "loss": 0.1969,
      "step": 4938
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.563772423638981,
      "learning_rate": 7.501003140508953e-06,
      "loss": 0.2431,
      "step": 4939
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3086536119191392,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.231,
      "step": 4940
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2805202486805651,
      "learning_rate": 7.498996725300981e-06,
      "loss": 0.1908,
      "step": 4941
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4735548834670145,
      "learning_rate": 7.4979933164657435e-06,
      "loss": 0.2256,
      "step": 4942
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3386817786513228,
      "learning_rate": 7.496989773548149e-06,
      "loss": 0.1723,
      "step": 4943
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.32719791301756,
      "learning_rate": 7.495986096602064e-06,
      "loss": 0.2302,
      "step": 4944
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3953854996827593,
      "learning_rate": 7.494982285681361e-06,
      "loss": 0.179,
      "step": 4945
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4277518925399062,
      "learning_rate": 7.493978340839922e-06,
      "loss": 0.2028,
      "step": 4946
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3257678355133768,
      "learning_rate": 7.4929742621316316e-06,
      "loss": 0.2195,
      "step": 4947
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.391522824989136,
      "learning_rate": 7.491970049610388e-06,
      "loss": 0.2035,
      "step": 4948
    },
    {
      "epoch": 0.35,
      "grad_norm": 9.302298565299465,
      "learning_rate": 7.490965703330092e-06,
      "loss": 0.6846,
      "step": 4949
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2947533257862585,
      "learning_rate": 7.489961223344654e-06,
      "loss": 0.2165,
      "step": 4950
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.812313190245047,
      "learning_rate": 7.488956609707988e-06,
      "loss": 0.585,
      "step": 4951
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.338333064193312,
      "learning_rate": 7.487951862474021e-06,
      "loss": 0.2193,
      "step": 4952
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.801694758161189,
      "learning_rate": 7.486946981696681e-06,
      "loss": 0.6365,
      "step": 4953
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2531530747128843,
      "learning_rate": 7.485941967429908e-06,
      "loss": 0.1742,
      "step": 4954
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.30309347691331,
      "learning_rate": 7.484936819727646e-06,
      "loss": 0.1511,
      "step": 4955
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4588277513293393,
      "learning_rate": 7.483931538643847e-06,
      "loss": 0.1658,
      "step": 4956
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.482204001242645,
      "learning_rate": 7.482926124232473e-06,
      "loss": 0.2778,
      "step": 4957
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4228124615462518,
      "learning_rate": 7.481920576547488e-06,
      "loss": 0.2163,
      "step": 4958
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5064408305912127,
      "learning_rate": 7.480914895642866e-06,
      "loss": 0.1719,
      "step": 4959
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4639589919492935,
      "learning_rate": 7.479909081572587e-06,
      "loss": 0.1989,
      "step": 4960
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3416977824803633,
      "learning_rate": 7.4789031343906425e-06,
      "loss": 0.1786,
      "step": 4961
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5187715510673072,
      "learning_rate": 7.4778970541510265e-06,
      "loss": 0.2705,
      "step": 4962
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3792651359763892,
      "learning_rate": 7.47689084090774e-06,
      "loss": 0.232,
      "step": 4963
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.329243693926206,
      "learning_rate": 7.475884494714794e-06,
      "loss": 0.7428,
      "step": 4964
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4654588554639674,
      "learning_rate": 7.474878015626204e-06,
      "loss": 0.2442,
      "step": 4965
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4817947980953543,
      "learning_rate": 7.473871403695994e-06,
      "loss": 0.2063,
      "step": 4966
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4288554842016112,
      "learning_rate": 7.472864658978195e-06,
      "loss": 0.2141,
      "step": 4967
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4722007821868395,
      "learning_rate": 7.471857781526846e-06,
      "loss": 0.19,
      "step": 4968
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2411222914807758,
      "learning_rate": 7.47085077139599e-06,
      "loss": 0.2233,
      "step": 4969
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3518056905884852,
      "learning_rate": 7.469843628639682e-06,
      "loss": 0.2266,
      "step": 4970
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2749149974591274,
      "learning_rate": 7.468836353311981e-06,
      "loss": 0.1832,
      "step": 4971
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3128472700591396,
      "learning_rate": 7.467828945466951e-06,
      "loss": 0.2163,
      "step": 4972
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.783606297757659,
      "learning_rate": 7.466821405158669e-06,
      "loss": 0.5181,
      "step": 4973
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3840725379667707,
      "learning_rate": 7.465813732441213e-06,
      "loss": 0.2244,
      "step": 4974
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.204090294273979,
      "learning_rate": 7.464805927368672e-06,
      "loss": 0.1766,
      "step": 4975
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3995168247664902,
      "learning_rate": 7.463797989995141e-06,
      "loss": 0.2371,
      "step": 4976
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2351626883583433,
      "learning_rate": 7.462789920374722e-06,
      "loss": 0.2327,
      "step": 4977
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3104804963697543,
      "learning_rate": 7.461781718561524e-06,
      "loss": 0.2225,
      "step": 4978
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5308322627951092,
      "learning_rate": 7.4607733846096645e-06,
      "loss": 0.1825,
      "step": 4979
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.090044475468448,
      "learning_rate": 7.459764918573264e-06,
      "loss": 0.5416,
      "step": 4980
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3605899422455203,
      "learning_rate": 7.4587563205064575e-06,
      "loss": 0.2072,
      "step": 4981
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5297551866753034,
      "learning_rate": 7.457747590463378e-06,
      "loss": 0.2346,
      "step": 4982
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.516775900207488,
      "learning_rate": 7.456738728498171e-06,
      "loss": 0.1988,
      "step": 4983
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4080765249319216,
      "learning_rate": 7.455729734664993e-06,
      "loss": 0.2006,
      "step": 4984
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.359848000505531,
      "learning_rate": 7.454720609017996e-06,
      "loss": 0.2065,
      "step": 4985
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4725352821496327,
      "learning_rate": 7.453711351611349e-06,
      "loss": 0.2247,
      "step": 4986
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.326877550427805,
      "learning_rate": 7.452701962499225e-06,
      "loss": 0.211,
      "step": 4987
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.302379900783371,
      "learning_rate": 7.451692441735804e-06,
      "loss": 0.2093,
      "step": 4988
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4113435272542305,
      "learning_rate": 7.450682789375272e-06,
      "loss": 0.1995,
      "step": 4989
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2861232622733887,
      "learning_rate": 7.449673005471825e-06,
      "loss": 0.1867,
      "step": 4990
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2516408707609386,
      "learning_rate": 7.448663090079662e-06,
      "loss": 0.206,
      "step": 4991
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3681929554335204,
      "learning_rate": 7.447653043252993e-06,
      "loss": 0.2437,
      "step": 4992
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4781755994456356,
      "learning_rate": 7.446642865046032e-06,
      "loss": 0.1872,
      "step": 4993
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2963522326464425,
      "learning_rate": 7.445632555513003e-06,
      "loss": 0.1986,
      "step": 4994
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3966555867825088,
      "learning_rate": 7.444622114708134e-06,
      "loss": 0.171,
      "step": 4995
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2734060457982501,
      "learning_rate": 7.443611542685663e-06,
      "loss": 0.1551,
      "step": 4996
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4455191339960127,
      "learning_rate": 7.44260083949983e-06,
      "loss": 0.2435,
      "step": 4997
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.254516001195114,
      "learning_rate": 7.441590005204888e-06,
      "loss": 0.1769,
      "step": 4998
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2699722072751873,
      "learning_rate": 7.440579039855097e-06,
      "loss": 0.201,
      "step": 4999
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2785232134454927,
      "learning_rate": 7.4395679435047175e-06,
      "loss": 0.1789,
      "step": 5000
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2685419615302125,
      "learning_rate": 7.4385567162080215e-06,
      "loss": 0.2495,
      "step": 5001
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4486776913062192,
      "learning_rate": 7.437545358019291e-06,
      "loss": 0.2326,
      "step": 5002
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3121029817118894,
      "learning_rate": 7.436533868992808e-06,
      "loss": 0.2147,
      "step": 5003
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0753230726305032,
      "learning_rate": 7.435522249182868e-06,
      "loss": 0.1657,
      "step": 5004
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.116535998391369,
      "learning_rate": 7.434510498643769e-06,
      "loss": 0.5774,
      "step": 5005
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4567701753845885,
      "learning_rate": 7.433498617429817e-06,
      "loss": 0.2316,
      "step": 5006
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.254519145806552,
      "learning_rate": 7.432486605595328e-06,
      "loss": 0.1963,
      "step": 5007
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2047559223408137,
      "learning_rate": 7.431474463194624e-06,
      "loss": 0.1915,
      "step": 5008
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5299436975800416,
      "learning_rate": 7.430462190282027e-06,
      "loss": 0.2719,
      "step": 5009
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.367391509733238,
      "learning_rate": 7.429449786911876e-06,
      "loss": 0.204,
      "step": 5010
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4626050551548924,
      "learning_rate": 7.428437253138512e-06,
      "loss": 0.2344,
      "step": 5011
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3667750570565864,
      "learning_rate": 7.4274245890162854e-06,
      "loss": 0.2626,
      "step": 5012
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3496837911785873,
      "learning_rate": 7.426411794599551e-06,
      "loss": 0.2079,
      "step": 5013
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.604965542930624,
      "learning_rate": 7.425398869942669e-06,
      "loss": 0.1951,
      "step": 5014
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4626663462069416,
      "learning_rate": 7.424385815100011e-06,
      "loss": 0.2113,
      "step": 5015
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3737570541114905,
      "learning_rate": 7.4233726301259555e-06,
      "loss": 0.1906,
      "step": 5016
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2279924735078325,
      "learning_rate": 7.4223593150748855e-06,
      "loss": 0.2047,
      "step": 5017
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3911663855664134,
      "learning_rate": 7.421345870001191e-06,
      "loss": 0.2609,
      "step": 5018
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.316822990250142,
      "learning_rate": 7.420332294959267e-06,
      "loss": 0.2206,
      "step": 5019
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.62683803517904,
      "learning_rate": 7.419318590003524e-06,
      "loss": 0.2201,
      "step": 5020
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2661009154476908,
      "learning_rate": 7.418304755188368e-06,
      "loss": 0.1816,
      "step": 5021
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4826038741199117,
      "learning_rate": 7.417290790568221e-06,
      "loss": 0.2125,
      "step": 5022
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3029836311122123,
      "learning_rate": 7.416276696197508e-06,
      "loss": 0.1898,
      "step": 5023
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4352710852586592,
      "learning_rate": 7.415262472130662e-06,
      "loss": 0.2133,
      "step": 5024
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4075647467950272,
      "learning_rate": 7.41424811842212e-06,
      "loss": 0.208,
      "step": 5025
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3499260887375084,
      "learning_rate": 7.413233635126332e-06,
      "loss": 0.1969,
      "step": 5026
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1583077733819755,
      "learning_rate": 7.412219022297749e-06,
      "loss": 0.1915,
      "step": 5027
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3369953391197438,
      "learning_rate": 7.411204279990832e-06,
      "loss": 0.2444,
      "step": 5028
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.159860390306262,
      "learning_rate": 7.410189408260049e-06,
      "loss": 0.594,
      "step": 5029
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3597841296002897,
      "learning_rate": 7.409174407159873e-06,
      "loss": 0.2075,
      "step": 5030
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.197193250817317,
      "learning_rate": 7.4081592767447865e-06,
      "loss": 0.173,
      "step": 5031
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2284514099036052,
      "learning_rate": 7.407144017069278e-06,
      "loss": 0.1967,
      "step": 5032
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3039656277051226,
      "learning_rate": 7.406128628187841e-06,
      "loss": 0.2295,
      "step": 5033
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3152298045258015,
      "learning_rate": 7.405113110154978e-06,
      "loss": 0.2023,
      "step": 5034
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.29135420448655,
      "learning_rate": 7.404097463025197e-06,
      "loss": 0.1686,
      "step": 5035
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.064973389530389,
      "learning_rate": 7.403081686853017e-06,
      "loss": 0.6202,
      "step": 5036
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4050148428714593,
      "learning_rate": 7.4020657816929594e-06,
      "loss": 0.2047,
      "step": 5037
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.682322779335673,
      "learning_rate": 7.401049747599552e-06,
      "loss": 0.2331,
      "step": 5038
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4284679347685394,
      "learning_rate": 7.400033584627333e-06,
      "loss": 0.2463,
      "step": 5039
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.514762780384517,
      "learning_rate": 7.399017292830848e-06,
      "loss": 0.5583,
      "step": 5040
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3018475750798593,
      "learning_rate": 7.3980008722646435e-06,
      "loss": 0.2189,
      "step": 5041
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1931038158556406,
      "learning_rate": 7.396984322983279e-06,
      "loss": 0.1665,
      "step": 5042
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2562281002231366,
      "learning_rate": 7.39596764504132e-06,
      "loss": 0.1892,
      "step": 5043
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4741267738891257,
      "learning_rate": 7.394950838493334e-06,
      "loss": 0.2071,
      "step": 5044
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.572257998058601,
      "learning_rate": 7.393933903393904e-06,
      "loss": 0.362,
      "step": 5045
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1954920667259894,
      "learning_rate": 7.3929168397976126e-06,
      "loss": 0.1574,
      "step": 5046
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5499893582186544,
      "learning_rate": 7.391899647759052e-06,
      "loss": 0.2428,
      "step": 5047
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.8688060894010246,
      "learning_rate": 7.390882327332819e-06,
      "loss": 0.5788,
      "step": 5048
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.408068472863165,
      "learning_rate": 7.389864878573523e-06,
      "loss": 0.2189,
      "step": 5049
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.145050207407742,
      "learning_rate": 7.3888473015357744e-06,
      "loss": 0.1393,
      "step": 5050
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3516528938735064,
      "learning_rate": 7.387829596274192e-06,
      "loss": 0.2101,
      "step": 5051
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5577477638591355,
      "learning_rate": 7.386811762843404e-06,
      "loss": 0.2137,
      "step": 5052
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.353687514234991,
      "learning_rate": 7.3857938012980425e-06,
      "loss": 0.2129,
      "step": 5053
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3785065904693468,
      "learning_rate": 7.384775711692749e-06,
      "loss": 0.2153,
      "step": 5054
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3118505601478194,
      "learning_rate": 7.383757494082169e-06,
      "loss": 0.2231,
      "step": 5055
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.441600778334505,
      "learning_rate": 7.3827391485209555e-06,
      "loss": 0.2646,
      "step": 5056
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.197618172506486,
      "learning_rate": 7.381720675063772e-06,
      "loss": 0.5099,
      "step": 5057
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4236147305355868,
      "learning_rate": 7.380702073765285e-06,
      "loss": 0.1943,
      "step": 5058
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4851612016128881,
      "learning_rate": 7.379683344680169e-06,
      "loss": 0.2379,
      "step": 5059
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3235883095300325,
      "learning_rate": 7.3786644878631035e-06,
      "loss": 0.2027,
      "step": 5060
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.244484782361441,
      "learning_rate": 7.377645503368778e-06,
      "loss": 0.151,
      "step": 5061
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1174083810085311,
      "learning_rate": 7.3766263912518885e-06,
      "loss": 0.1237,
      "step": 5062
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.018050125766409,
      "learning_rate": 7.375607151567137e-06,
      "loss": 0.4748,
      "step": 5063
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.218435133616388,
      "learning_rate": 7.374587784369231e-06,
      "loss": 0.5892,
      "step": 5064
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.64271679144367,
      "learning_rate": 7.3735682897128854e-06,
      "loss": 0.5502,
      "step": 5065
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3555756495555293,
      "learning_rate": 7.372548667652826e-06,
      "loss": 0.229,
      "step": 5066
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.48899415805759,
      "learning_rate": 7.3715289182437775e-06,
      "loss": 0.2458,
      "step": 5067
    },
    {
      "epoch": 0.36,
      "grad_norm": 10.8270945351268,
      "learning_rate": 7.37050904154048e-06,
      "loss": 0.6074,
      "step": 5068
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.529923760001668,
      "learning_rate": 7.369489037597673e-06,
      "loss": 0.8744,
      "step": 5069
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3148280549662714,
      "learning_rate": 7.368468906470109e-06,
      "loss": 0.1528,
      "step": 5070
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3451109039714237,
      "learning_rate": 7.367448648212543e-06,
      "loss": 0.1729,
      "step": 5071
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.477737095536321,
      "learning_rate": 7.366428262879739e-06,
      "loss": 0.5628,
      "step": 5072
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4162625413971304,
      "learning_rate": 7.365407750526468e-06,
      "loss": 0.1858,
      "step": 5073
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4037819036938215,
      "learning_rate": 7.364387111207504e-06,
      "loss": 0.2041,
      "step": 5074
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4081080941664654,
      "learning_rate": 7.363366344977635e-06,
      "loss": 0.2325,
      "step": 5075
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3169872743863023,
      "learning_rate": 7.362345451891649e-06,
      "loss": 0.2137,
      "step": 5076
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.967329823486751,
      "learning_rate": 7.361324432004345e-06,
      "loss": 0.4909,
      "step": 5077
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2649711289107375,
      "learning_rate": 7.360303285370526e-06,
      "loss": 0.1927,
      "step": 5078
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4192690443747504,
      "learning_rate": 7.359282012045004e-06,
      "loss": 0.204,
      "step": 5079
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.298050556452496,
      "learning_rate": 7.358260612082596e-06,
      "loss": 0.2232,
      "step": 5080
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.869120480193776,
      "learning_rate": 7.357239085538128e-06,
      "loss": 0.5494,
      "step": 5081
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4819430223275167,
      "learning_rate": 7.356217432466429e-06,
      "loss": 0.2633,
      "step": 5082
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3298582142137947,
      "learning_rate": 7.355195652922341e-06,
      "loss": 0.2124,
      "step": 5083
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2909246133832941,
      "learning_rate": 7.3541737469607066e-06,
      "loss": 0.191,
      "step": 5084
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7024423232529686,
      "learning_rate": 7.353151714636378e-06,
      "loss": 0.2213,
      "step": 5085
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3654462733634802,
      "learning_rate": 7.352129556004213e-06,
      "loss": 0.2088,
      "step": 5086
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4398679995804622,
      "learning_rate": 7.351107271119079e-06,
      "loss": 0.2176,
      "step": 5087
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.907328569565744,
      "learning_rate": 7.350084860035846e-06,
      "loss": 0.575,
      "step": 5088
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.339368158161762,
      "learning_rate": 7.349062322809396e-06,
      "loss": 0.2181,
      "step": 5089
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.481302468033255,
      "learning_rate": 7.34803965949461e-06,
      "loss": 0.2312,
      "step": 5090
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2469803170214144,
      "learning_rate": 7.3470168701463865e-06,
      "loss": 0.1908,
      "step": 5091
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4101426289432533,
      "learning_rate": 7.34599395481962e-06,
      "loss": 0.2301,
      "step": 5092
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4102732734700403,
      "learning_rate": 7.344970913569218e-06,
      "loss": 0.2305,
      "step": 5093
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3134751850961166,
      "learning_rate": 7.343947746450093e-06,
      "loss": 0.1995,
      "step": 5094
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4417431845009951,
      "learning_rate": 7.342924453517166e-06,
      "loss": 0.2053,
      "step": 5095
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.468886601041917,
      "learning_rate": 7.3419010348253626e-06,
      "loss": 0.1979,
      "step": 5096
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3699813739498823,
      "learning_rate": 7.340877490429616e-06,
      "loss": 0.2274,
      "step": 5097
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3513958122951433,
      "learning_rate": 7.339853820384863e-06,
      "loss": 0.2256,
      "step": 5098
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6218953351697176,
      "learning_rate": 7.338830024746054e-06,
      "loss": 0.2863,
      "step": 5099
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.366117596342071,
      "learning_rate": 7.3378061035681415e-06,
      "loss": 0.1755,
      "step": 5100
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4907351789004128,
      "learning_rate": 7.336782056906085e-06,
      "loss": 0.2415,
      "step": 5101
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.83784183925294,
      "learning_rate": 7.33575788481485e-06,
      "loss": 0.6952,
      "step": 5102
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3915068076300086,
      "learning_rate": 7.3347335873494115e-06,
      "loss": 0.2049,
      "step": 5103
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.853785239163227,
      "learning_rate": 7.33370916456475e-06,
      "loss": 0.62,
      "step": 5104
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.653635022474864,
      "learning_rate": 7.332684616515853e-06,
      "loss": 0.6255,
      "step": 5105
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2317747594837243,
      "learning_rate": 7.331659943257712e-06,
      "loss": 0.1988,
      "step": 5106
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3242076092525958,
      "learning_rate": 7.330635144845326e-06,
      "loss": 0.2096,
      "step": 5107
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3359612460151886,
      "learning_rate": 7.329610221333707e-06,
      "loss": 0.2327,
      "step": 5108
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.362601045135596,
      "learning_rate": 7.328585172777866e-06,
      "loss": 0.206,
      "step": 5109
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2320770293602212,
      "learning_rate": 7.327559999232826e-06,
      "loss": 0.1625,
      "step": 5110
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.416593582190875,
      "learning_rate": 7.3265347007536105e-06,
      "loss": 0.2249,
      "step": 5111
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3491679041254274,
      "learning_rate": 7.3255092773952545e-06,
      "loss": 0.1808,
      "step": 5112
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3289184341102411,
      "learning_rate": 7.324483729212799e-06,
      "loss": 0.1716,
      "step": 5113
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3547225452016027,
      "learning_rate": 7.3234580562612925e-06,
      "loss": 0.205,
      "step": 5114
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2992165382839085,
      "learning_rate": 7.322432258595789e-06,
      "loss": 0.1617,
      "step": 5115
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3429405705094573,
      "learning_rate": 7.321406336271346e-06,
      "loss": 0.2288,
      "step": 5116
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5744046977208355,
      "learning_rate": 7.320380289343034e-06,
      "loss": 0.2784,
      "step": 5117
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2475813397793685,
      "learning_rate": 7.319354117865928e-06,
      "loss": 0.2135,
      "step": 5118
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5539987133417845,
      "learning_rate": 7.318327821895106e-06,
      "loss": 0.2094,
      "step": 5119
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.395083805081527,
      "learning_rate": 7.317301401485657e-06,
      "loss": 0.2186,
      "step": 5120
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6799260528271898,
      "learning_rate": 7.316274856692676e-06,
      "loss": 0.2524,
      "step": 5121
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3154543855523566,
      "learning_rate": 7.315248187571261e-06,
      "loss": 0.2151,
      "step": 5122
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3493201538042643,
      "learning_rate": 7.314221394176521e-06,
      "loss": 0.1744,
      "step": 5123
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3188766631770004,
      "learning_rate": 7.313194476563572e-06,
      "loss": 0.1962,
      "step": 5124
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1916186652604503,
      "learning_rate": 7.312167434787533e-06,
      "loss": 0.1423,
      "step": 5125
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3734549448807392,
      "learning_rate": 7.311140268903532e-06,
      "loss": 0.2174,
      "step": 5126
    },
    {
      "epoch": 0.37,
      "grad_norm": 25.814919012522356,
      "learning_rate": 7.310112978966703e-06,
      "loss": 0.7954,
      "step": 5127
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.354703031760036,
      "learning_rate": 7.309085565032186e-06,
      "loss": 0.1984,
      "step": 5128
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.209679246354488,
      "learning_rate": 7.30805802715513e-06,
      "loss": 0.2166,
      "step": 5129
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4148499414399502,
      "learning_rate": 7.3070303653906885e-06,
      "loss": 0.2083,
      "step": 5130
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.680965311972458,
      "learning_rate": 7.306002579794022e-06,
      "loss": 0.5361,
      "step": 5131
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.293801887529542,
      "learning_rate": 7.304974670420299e-06,
      "loss": 0.5016,
      "step": 5132
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3298934060847296,
      "learning_rate": 7.3039466373246924e-06,
      "loss": 0.1938,
      "step": 5133
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4031811427091343,
      "learning_rate": 7.302918480562384e-06,
      "loss": 0.2408,
      "step": 5134
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3155849047084702,
      "learning_rate": 7.301890200188559e-06,
      "loss": 0.2167,
      "step": 5135
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3504393618522972,
      "learning_rate": 7.300861796258414e-06,
      "loss": 0.1803,
      "step": 5136
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3594597575281275,
      "learning_rate": 7.299833268827149e-06,
      "loss": 0.2355,
      "step": 5137
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2982296238885702,
      "learning_rate": 7.298804617949971e-06,
      "loss": 0.1852,
      "step": 5138
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.63364757781956,
      "learning_rate": 7.297775843682092e-06,
      "loss": 0.5781,
      "step": 5139
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.434353908119797,
      "learning_rate": 7.296746946078737e-06,
      "loss": 0.2157,
      "step": 5140
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4063621249688407,
      "learning_rate": 7.295717925195127e-06,
      "loss": 0.2054,
      "step": 5141
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3270107107523625,
      "learning_rate": 7.294688781086502e-06,
      "loss": 0.2167,
      "step": 5142
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.355016926810944,
      "learning_rate": 7.293659513808099e-06,
      "loss": 0.2151,
      "step": 5143
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.3304327626455645,
      "learning_rate": 7.292630123415165e-06,
      "loss": 0.5147,
      "step": 5144
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4360423980230157,
      "learning_rate": 7.291600609962954e-06,
      "loss": 0.2479,
      "step": 5145
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4132294470869529,
      "learning_rate": 7.290570973506728e-06,
      "loss": 0.2067,
      "step": 5146
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3341659203066594,
      "learning_rate": 7.289541214101751e-06,
      "loss": 0.1688,
      "step": 5147
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.258479962165081,
      "learning_rate": 7.288511331803296e-06,
      "loss": 0.2207,
      "step": 5148
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3573728195272545,
      "learning_rate": 7.287481326666646e-06,
      "loss": 0.2329,
      "step": 5149
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4824645746943406,
      "learning_rate": 7.286451198747086e-06,
      "loss": 0.1872,
      "step": 5150
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4317296056318793,
      "learning_rate": 7.285420948099909e-06,
      "loss": 0.2185,
      "step": 5151
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4843550812587933,
      "learning_rate": 7.2843905747804155e-06,
      "loss": 0.2202,
      "step": 5152
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3528765347796194,
      "learning_rate": 7.283360078843911e-06,
      "loss": 0.2115,
      "step": 5153
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.108385858553031,
      "learning_rate": 7.282329460345708e-06,
      "loss": 0.1663,
      "step": 5154
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3827773904075156,
      "learning_rate": 7.2812987193411275e-06,
      "loss": 0.172,
      "step": 5155
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2629200472943685,
      "learning_rate": 7.280267855885495e-06,
      "loss": 0.18,
      "step": 5156
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3035043518947358,
      "learning_rate": 7.279236870034143e-06,
      "loss": 0.1644,
      "step": 5157
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2148600687824793,
      "learning_rate": 7.278205761842411e-06,
      "loss": 0.1629,
      "step": 5158
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4951490784323807,
      "learning_rate": 7.277174531365644e-06,
      "loss": 0.2304,
      "step": 5159
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3028294533111069,
      "learning_rate": 7.276143178659195e-06,
      "loss": 0.216,
      "step": 5160
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3263851490994678,
      "learning_rate": 7.275111703778424e-06,
      "loss": 0.1586,
      "step": 5161
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2576312475354818,
      "learning_rate": 7.274080106778693e-06,
      "loss": 0.1633,
      "step": 5162
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2577592983473234,
      "learning_rate": 7.273048387715378e-06,
      "loss": 0.182,
      "step": 5163
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3001077297600752,
      "learning_rate": 7.272016546643856e-06,
      "loss": 0.2038,
      "step": 5164
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2678935690023225,
      "learning_rate": 7.270984583619511e-06,
      "loss": 0.1884,
      "step": 5165
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3723187033295654,
      "learning_rate": 7.269952498697734e-06,
      "loss": 0.2079,
      "step": 5166
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3042531944184668,
      "learning_rate": 7.2689202919339275e-06,
      "loss": 0.2059,
      "step": 5167
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.606630411940769,
      "learning_rate": 7.267887963383493e-06,
      "loss": 0.2616,
      "step": 5168
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.463249577946215,
      "learning_rate": 7.2668555131018436e-06,
      "loss": 0.2197,
      "step": 5169
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3729679382122428,
      "learning_rate": 7.265822941144394e-06,
      "loss": 0.1936,
      "step": 5170
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2525458466183756,
      "learning_rate": 7.264790247566571e-06,
      "loss": 0.1573,
      "step": 5171
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5284074909251033,
      "learning_rate": 7.263757432423806e-06,
      "loss": 0.2175,
      "step": 5172
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.334218645271567,
      "learning_rate": 7.262724495771535e-06,
      "loss": 0.212,
      "step": 5173
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.614828006929252,
      "learning_rate": 7.2616914376652025e-06,
      "loss": 0.7485,
      "step": 5174
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.9766280174403414,
      "learning_rate": 7.260658258160258e-06,
      "loss": 0.6285,
      "step": 5175
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1761404133082543,
      "learning_rate": 7.259624957312161e-06,
      "loss": 0.1793,
      "step": 5176
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5528252059853986,
      "learning_rate": 7.258591535176372e-06,
      "loss": 0.1793,
      "step": 5177
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.43773186882904,
      "learning_rate": 7.257557991808364e-06,
      "loss": 0.2338,
      "step": 5178
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.293354125614021,
      "learning_rate": 7.256524327263609e-06,
      "loss": 0.1839,
      "step": 5179
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3773617489408085,
      "learning_rate": 7.255490541597594e-06,
      "loss": 0.2221,
      "step": 5180
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.693038900622433,
      "learning_rate": 7.254456634865809e-06,
      "loss": 0.5299,
      "step": 5181
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4212298092844184,
      "learning_rate": 7.253422607123747e-06,
      "loss": 0.1883,
      "step": 5182
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.196121250077038,
      "learning_rate": 7.252388458426911e-06,
      "loss": 0.1595,
      "step": 5183
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4078144708601932,
      "learning_rate": 7.251354188830811e-06,
      "loss": 0.2341,
      "step": 5184
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2453539056806038,
      "learning_rate": 7.250319798390963e-06,
      "loss": 0.1738,
      "step": 5185
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3614467982557197,
      "learning_rate": 7.249285287162889e-06,
      "loss": 0.2094,
      "step": 5186
    },
    {
      "epoch": 0.37,
      "grad_norm": 11.46518890857137,
      "learning_rate": 7.2482506552021156e-06,
      "loss": 0.4895,
      "step": 5187
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.804094126042983,
      "learning_rate": 7.24721590256418e-06,
      "loss": 0.6617,
      "step": 5188
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4887594601316252,
      "learning_rate": 7.246181029304621e-06,
      "loss": 0.2113,
      "step": 5189
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.4328976786609635,
      "learning_rate": 7.245146035478989e-06,
      "loss": 0.5084,
      "step": 5190
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3998945603461392,
      "learning_rate": 7.244110921142836e-06,
      "loss": 0.2193,
      "step": 5191
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.390308470260549,
      "learning_rate": 7.243075686351725e-06,
      "loss": 0.1643,
      "step": 5192
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.361451300515782,
      "learning_rate": 7.242040331161223e-06,
      "loss": 0.7866,
      "step": 5193
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3752222453678717,
      "learning_rate": 7.241004855626903e-06,
      "loss": 0.209,
      "step": 5194
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4297853317980782,
      "learning_rate": 7.239969259804344e-06,
      "loss": 0.1841,
      "step": 5195
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.415871817163078,
      "learning_rate": 7.238933543749135e-06,
      "loss": 0.2279,
      "step": 5196
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3726852462029897,
      "learning_rate": 7.237897707516869e-06,
      "loss": 0.1966,
      "step": 5197
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3050725056091215,
      "learning_rate": 7.236861751163144e-06,
      "loss": 0.1535,
      "step": 5198
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3298373560564996,
      "learning_rate": 7.235825674743566e-06,
      "loss": 0.1961,
      "step": 5199
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.350616649716634,
      "learning_rate": 7.2347894783137485e-06,
      "loss": 0.1691,
      "step": 5200
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4097265758634332,
      "learning_rate": 7.233753161929311e-06,
      "loss": 0.1901,
      "step": 5201
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3850589952964674,
      "learning_rate": 7.232716725645878e-06,
      "loss": 0.2317,
      "step": 5202
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.189507890619141,
      "learning_rate": 7.2316801695190805e-06,
      "loss": 0.1611,
      "step": 5203
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3547494435062506,
      "learning_rate": 7.230643493604557e-06,
      "loss": 0.1927,
      "step": 5204
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3850060634800934,
      "learning_rate": 7.229606697957954e-06,
      "loss": 0.2457,
      "step": 5205
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4276054498014128,
      "learning_rate": 7.228569782634921e-06,
      "loss": 0.2012,
      "step": 5206
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.909443532007172,
      "learning_rate": 7.227532747691116e-06,
      "loss": 0.7191,
      "step": 5207
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.556164160010817,
      "learning_rate": 7.2264955931822025e-06,
      "loss": 0.2227,
      "step": 5208
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.324432853537748,
      "learning_rate": 7.22545831916385e-06,
      "loss": 0.184,
      "step": 5209
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2981119903414269,
      "learning_rate": 7.2244209256917365e-06,
      "loss": 0.2007,
      "step": 5210
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3081605669408995,
      "learning_rate": 7.2233834128215454e-06,
      "loss": 0.2103,
      "step": 5211
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4321957307194026,
      "learning_rate": 7.222345780608966e-06,
      "loss": 0.2192,
      "step": 5212
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3353790485087782,
      "learning_rate": 7.2213080291096925e-06,
      "loss": 0.212,
      "step": 5213
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.118043416651116,
      "learning_rate": 7.22027015837943e-06,
      "loss": 0.146,
      "step": 5214
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4286772145067639,
      "learning_rate": 7.219232168473886e-06,
      "loss": 0.1961,
      "step": 5215
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3224601038674204,
      "learning_rate": 7.218194059448775e-06,
      "loss": 0.2374,
      "step": 5216
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2536616816707513,
      "learning_rate": 7.21715583135982e-06,
      "loss": 0.1675,
      "step": 5217
    },
    {
      "epoch": 0.37,
      "grad_norm": 13.416246544080483,
      "learning_rate": 7.216117484262748e-06,
      "loss": 0.5001,
      "step": 5218
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4573004238493596,
      "learning_rate": 7.215079018213294e-06,
      "loss": 0.2159,
      "step": 5219
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4739864880204603,
      "learning_rate": 7.2140404332671986e-06,
      "loss": 0.2388,
      "step": 5220
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5674819679957297,
      "learning_rate": 7.213001729480207e-06,
      "loss": 0.2596,
      "step": 5221
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.554984172802207,
      "learning_rate": 7.2119629069080766e-06,
      "loss": 0.2518,
      "step": 5222
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.242590750946218,
      "learning_rate": 7.210923965606564e-06,
      "loss": 0.1768,
      "step": 5223
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5459015741480204,
      "learning_rate": 7.209884905631437e-06,
      "loss": 0.257,
      "step": 5224
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.73546361333772,
      "learning_rate": 7.208845727038466e-06,
      "loss": 0.5381,
      "step": 5225
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1787149298978972,
      "learning_rate": 7.207806429883433e-06,
      "loss": 0.1362,
      "step": 5226
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2675783787813169,
      "learning_rate": 7.206767014222121e-06,
      "loss": 0.1912,
      "step": 5227
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3112684966441426,
      "learning_rate": 7.2057274801103235e-06,
      "loss": 0.1946,
      "step": 5228
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.227659156923823,
      "learning_rate": 7.204687827603838e-06,
      "loss": 0.159,
      "step": 5229
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1808104269558863,
      "learning_rate": 7.203648056758468e-06,
      "loss": 0.1622,
      "step": 5230
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3114681554794212,
      "learning_rate": 7.202608167630026e-06,
      "loss": 0.1721,
      "step": 5231
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.285441981582618,
      "learning_rate": 7.201568160274327e-06,
      "loss": 0.1705,
      "step": 5232
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4060944873286094,
      "learning_rate": 7.2005280347471955e-06,
      "loss": 0.2034,
      "step": 5233
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.323400990736908,
      "learning_rate": 7.19948779110446e-06,
      "loss": 0.2173,
      "step": 5234
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4023750873886776,
      "learning_rate": 7.19844742940196e-06,
      "loss": 0.1865,
      "step": 5235
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.426244152049011,
      "learning_rate": 7.1974069496955365e-06,
      "loss": 0.2052,
      "step": 5236
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.412262846832108,
      "learning_rate": 7.196366352041037e-06,
      "loss": 0.2137,
      "step": 5237
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4667088460909592,
      "learning_rate": 7.195325636494317e-06,
      "loss": 0.2022,
      "step": 5238
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3802805101133935,
      "learning_rate": 7.194284803111239e-06,
      "loss": 0.2336,
      "step": 5239
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3926828310978259,
      "learning_rate": 7.19324385194767e-06,
      "loss": 0.2349,
      "step": 5240
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2199814686947814,
      "learning_rate": 7.1922027830594846e-06,
      "loss": 0.135,
      "step": 5241
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4158870466344178,
      "learning_rate": 7.1911615965025615e-06,
      "loss": 0.1884,
      "step": 5242
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2549814342067525,
      "learning_rate": 7.19012029233279e-06,
      "loss": 0.2112,
      "step": 5243
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.333600611200091,
      "learning_rate": 7.189078870606063e-06,
      "loss": 0.206,
      "step": 5244
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.796546380376461,
      "learning_rate": 7.188037331378278e-06,
      "loss": 0.6358,
      "step": 5245
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2673757158726637,
      "learning_rate": 7.1869956747053414e-06,
      "loss": 0.1917,
      "step": 5246
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.0815786492865875,
      "learning_rate": 7.185953900643165e-06,
      "loss": 0.5622,
      "step": 5247
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.727191632389053,
      "learning_rate": 7.1849120092476685e-06,
      "loss": 0.2434,
      "step": 5248
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2395147299528873,
      "learning_rate": 7.183870000574777e-06,
      "loss": 0.1758,
      "step": 5249
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4297513270738393,
      "learning_rate": 7.182827874680417e-06,
      "loss": 0.2116,
      "step": 5250
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2794268923172978,
      "learning_rate": 7.18178563162053e-06,
      "loss": 0.2004,
      "step": 5251
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3069182235896368,
      "learning_rate": 7.180743271451058e-06,
      "loss": 0.2113,
      "step": 5252
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2648456227397045,
      "learning_rate": 7.179700794227952e-06,
      "loss": 0.1927,
      "step": 5253
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.294624020087715,
      "learning_rate": 7.178658200007165e-06,
      "loss": 0.2005,
      "step": 5254
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4407690305450147,
      "learning_rate": 7.177615488844661e-06,
      "loss": 0.2539,
      "step": 5255
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.29352991089323,
      "learning_rate": 7.17657266079641e-06,
      "loss": 0.7696,
      "step": 5256
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4421601640421118,
      "learning_rate": 7.175529715918386e-06,
      "loss": 0.1713,
      "step": 5257
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.251778709842344,
      "learning_rate": 7.1744866542665695e-06,
      "loss": 0.5524,
      "step": 5258
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3505063954947236,
      "learning_rate": 7.173443475896946e-06,
      "loss": 0.1916,
      "step": 5259
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.985356346821925,
      "learning_rate": 7.172400180865514e-06,
      "loss": 0.6054,
      "step": 5260
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4297995671407833,
      "learning_rate": 7.17135676922827e-06,
      "loss": 0.2121,
      "step": 5261
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.297481235706431,
      "learning_rate": 7.1703132410412215e-06,
      "loss": 0.1685,
      "step": 5262
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.274773710041479,
      "learning_rate": 7.169269596360379e-06,
      "loss": 0.1991,
      "step": 5263
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2977346617550918,
      "learning_rate": 7.168225835241764e-06,
      "loss": 0.1615,
      "step": 5264
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.258783013672195,
      "learning_rate": 7.167181957741398e-06,
      "loss": 0.1726,
      "step": 5265
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.325941785774385,
      "learning_rate": 7.166137963915316e-06,
      "loss": 0.1958,
      "step": 5266
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3681856554132266,
      "learning_rate": 7.165093853819553e-06,
      "loss": 0.1977,
      "step": 5267
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.911135145452195,
      "learning_rate": 7.164049627510154e-06,
      "loss": 0.7471,
      "step": 5268
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4731608476626972,
      "learning_rate": 7.163005285043168e-06,
      "loss": 0.2396,
      "step": 5269
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.415189789601481,
      "learning_rate": 7.161960826474651e-06,
      "loss": 0.2088,
      "step": 5270
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.355802769028095,
      "learning_rate": 7.160916251860667e-06,
      "loss": 0.2392,
      "step": 5271
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4139478354102717,
      "learning_rate": 7.159871561257282e-06,
      "loss": 0.2008,
      "step": 5272
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3553352440241615,
      "learning_rate": 7.158826754720574e-06,
      "loss": 0.224,
      "step": 5273
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.299653602677173,
      "learning_rate": 7.1577818323066205e-06,
      "loss": 0.2476,
      "step": 5274
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4214623239103048,
      "learning_rate": 7.156736794071512e-06,
      "loss": 0.2407,
      "step": 5275
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3338898738432636,
      "learning_rate": 7.155691640071341e-06,
      "loss": 0.175,
      "step": 5276
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3545367785801545,
      "learning_rate": 7.154646370362206e-06,
      "loss": 0.2039,
      "step": 5277
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3833067920256379,
      "learning_rate": 7.153600985000213e-06,
      "loss": 0.2379,
      "step": 5278
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2123594881388255,
      "learning_rate": 7.1525554840414765e-06,
      "loss": 0.1682,
      "step": 5279
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2718267930716303,
      "learning_rate": 7.1515098675421125e-06,
      "loss": 0.1545,
      "step": 5280
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.470272208182743,
      "learning_rate": 7.150464135558246e-06,
      "loss": 0.1835,
      "step": 5281
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3271528128938859,
      "learning_rate": 7.14941828814601e-06,
      "loss": 0.1789,
      "step": 5282
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4291837746951697,
      "learning_rate": 7.148372325361538e-06,
      "loss": 0.2197,
      "step": 5283
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.280551710575636,
      "learning_rate": 7.147326247260973e-06,
      "loss": 0.1947,
      "step": 5284
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5145956947959534,
      "learning_rate": 7.146280053900468e-06,
      "loss": 0.2038,
      "step": 5285
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.996376436278586,
      "learning_rate": 7.145233745336176e-06,
      "loss": 0.5867,
      "step": 5286
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1775284987475882,
      "learning_rate": 7.144187321624259e-06,
      "loss": 0.1727,
      "step": 5287
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.476337945416581,
      "learning_rate": 7.143140782820885e-06,
      "loss": 0.5064,
      "step": 5288
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3488345043328493,
      "learning_rate": 7.142094128982228e-06,
      "loss": 0.2399,
      "step": 5289
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.401625436230305,
      "learning_rate": 7.141047360164469e-06,
      "loss": 0.1985,
      "step": 5290
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3370891656015966,
      "learning_rate": 7.140000476423793e-06,
      "loss": 0.1926,
      "step": 5291
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.43850975081692,
      "learning_rate": 7.138953477816393e-06,
      "loss": 0.2138,
      "step": 5292
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2688690255232633,
      "learning_rate": 7.137906364398468e-06,
      "loss": 0.2017,
      "step": 5293
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4473672598169522,
      "learning_rate": 7.136859136226223e-06,
      "loss": 0.2312,
      "step": 5294
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5216866236271294,
      "learning_rate": 7.135811793355869e-06,
      "loss": 0.2202,
      "step": 5295
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4016946341096583,
      "learning_rate": 7.1347643358436226e-06,
      "loss": 0.2069,
      "step": 5296
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3453228880814028,
      "learning_rate": 7.133716763745707e-06,
      "loss": 0.1776,
      "step": 5297
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4409939984676357,
      "learning_rate": 7.132669077118352e-06,
      "loss": 0.2084,
      "step": 5298
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.334354563803864,
      "learning_rate": 7.131621276017796e-06,
      "loss": 0.2318,
      "step": 5299
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.820991565415937,
      "learning_rate": 7.130573360500277e-06,
      "loss": 0.6317,
      "step": 5300
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.378037039414726,
      "learning_rate": 7.1295253306220435e-06,
      "loss": 0.1945,
      "step": 5301
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.165590661477633,
      "learning_rate": 7.128477186439352e-06,
      "loss": 0.1617,
      "step": 5302
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3848916198514492,
      "learning_rate": 7.127428928008461e-06,
      "loss": 0.2173,
      "step": 5303
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3587961795281016,
      "learning_rate": 7.126380555385637e-06,
      "loss": 0.2179,
      "step": 5304
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4334153696673626,
      "learning_rate": 7.125332068627151e-06,
      "loss": 0.2367,
      "step": 5305
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2331758708682101,
      "learning_rate": 7.124283467789284e-06,
      "loss": 0.1777,
      "step": 5306
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3629330195691678,
      "learning_rate": 7.12323475292832e-06,
      "loss": 0.2468,
      "step": 5307
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2608640531997062,
      "learning_rate": 7.12218592410055e-06,
      "loss": 0.2012,
      "step": 5308
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2878358055964931,
      "learning_rate": 7.12113698136227e-06,
      "loss": 0.1885,
      "step": 5309
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.7216822643518075,
      "learning_rate": 7.120087924769783e-06,
      "loss": 0.6202,
      "step": 5310
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4701954639598007,
      "learning_rate": 7.119038754379399e-06,
      "loss": 0.1826,
      "step": 5311
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3926629610644652,
      "learning_rate": 7.117989470247434e-06,
      "loss": 0.2239,
      "step": 5312
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.861720419306191,
      "learning_rate": 7.116940072430208e-06,
      "loss": 0.5447,
      "step": 5313
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4597522882713152,
      "learning_rate": 7.1158905609840486e-06,
      "loss": 0.244,
      "step": 5314
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.406652505729443,
      "learning_rate": 7.1148409359652914e-06,
      "loss": 0.2269,
      "step": 5315
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6689972780200897,
      "learning_rate": 7.113791197430275e-06,
      "loss": 0.2185,
      "step": 5316
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3922548995863004,
      "learning_rate": 7.112741345435342e-06,
      "loss": 0.2114,
      "step": 5317
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1602190523344074,
      "learning_rate": 7.111691380036848e-06,
      "loss": 0.1707,
      "step": 5318
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3844979994689708,
      "learning_rate": 7.1106413012911514e-06,
      "loss": 0.2481,
      "step": 5319
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2174944224488164,
      "learning_rate": 7.109591109254614e-06,
      "loss": 0.148,
      "step": 5320
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3159546984056836,
      "learning_rate": 7.108540803983608e-06,
      "loss": 0.1736,
      "step": 5321
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2458224018515767,
      "learning_rate": 7.107490385534506e-06,
      "loss": 0.2056,
      "step": 5322
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.328504744734417,
      "learning_rate": 7.106439853963694e-06,
      "loss": 0.2347,
      "step": 5323
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2488937313417505,
      "learning_rate": 7.10538920932756e-06,
      "loss": 0.2115,
      "step": 5324
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3151866389456683,
      "learning_rate": 7.1043384516824965e-06,
      "loss": 0.2091,
      "step": 5325
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3148738407977136,
      "learning_rate": 7.103287581084905e-06,
      "loss": 0.227,
      "step": 5326
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1227393727508552,
      "learning_rate": 7.102236597591193e-06,
      "loss": 0.1129,
      "step": 5327
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.341605284158933,
      "learning_rate": 7.1011855012577705e-06,
      "loss": 0.1787,
      "step": 5328
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4239463000165027,
      "learning_rate": 7.100134292141058e-06,
      "loss": 0.2186,
      "step": 5329
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2798347196932893,
      "learning_rate": 7.099082970297481e-06,
      "loss": 0.1981,
      "step": 5330
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3222489756026916,
      "learning_rate": 7.098031535783468e-06,
      "loss": 0.2075,
      "step": 5331
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.543461247996678,
      "learning_rate": 7.096979988655459e-06,
      "loss": 0.2504,
      "step": 5332
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4116409393313618,
      "learning_rate": 7.095928328969895e-06,
      "loss": 0.199,
      "step": 5333
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2768246204064162,
      "learning_rate": 7.094876556783225e-06,
      "loss": 0.1485,
      "step": 5334
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3235101589700204,
      "learning_rate": 7.093824672151901e-06,
      "loss": 0.2023,
      "step": 5335
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2038491894455357,
      "learning_rate": 7.09277267513239e-06,
      "loss": 0.1972,
      "step": 5336
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.199967832284442,
      "learning_rate": 7.091720565781156e-06,
      "loss": 0.1607,
      "step": 5337
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2800986957899048,
      "learning_rate": 7.090668344154671e-06,
      "loss": 0.182,
      "step": 5338
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2598876949846325,
      "learning_rate": 7.0896160103094145e-06,
      "loss": 0.2104,
      "step": 5339
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3473038713994927,
      "learning_rate": 7.088563564301874e-06,
      "loss": 0.2539,
      "step": 5340
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3985245043743806,
      "learning_rate": 7.0875110061885375e-06,
      "loss": 0.1926,
      "step": 5341
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3564563019131808,
      "learning_rate": 7.086458336025904e-06,
      "loss": 0.2008,
      "step": 5342
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3250471927629286,
      "learning_rate": 7.085405553870475e-06,
      "loss": 0.2039,
      "step": 5343
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.481494999284909,
      "learning_rate": 7.084352659778762e-06,
      "loss": 0.2043,
      "step": 5344
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4642368292048393,
      "learning_rate": 7.083299653807278e-06,
      "loss": 0.1954,
      "step": 5345
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5944475816441144,
      "learning_rate": 7.082246536012546e-06,
      "loss": 0.2429,
      "step": 5346
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1162649820923438,
      "learning_rate": 7.081193306451091e-06,
      "loss": 0.1749,
      "step": 5347
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6472149481192184,
      "learning_rate": 7.080139965179449e-06,
      "loss": 0.2093,
      "step": 5348
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4379596979127667,
      "learning_rate": 7.079086512254156e-06,
      "loss": 0.2052,
      "step": 5349
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2858700292192091,
      "learning_rate": 7.078032947731761e-06,
      "loss": 0.1583,
      "step": 5350
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4068751127747166,
      "learning_rate": 7.076979271668811e-06,
      "loss": 0.2762,
      "step": 5351
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4107479290003986,
      "learning_rate": 7.075925484121866e-06,
      "loss": 0.2243,
      "step": 5352
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4382586138466011,
      "learning_rate": 7.074871585147488e-06,
      "loss": 0.2122,
      "step": 5353
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2980422364784499,
      "learning_rate": 7.0738175748022464e-06,
      "loss": 0.1896,
      "step": 5354
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5171936244275739,
      "learning_rate": 7.0727634531427155e-06,
      "loss": 0.2524,
      "step": 5355
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.346182385629131,
      "learning_rate": 7.071709220225478e-06,
      "loss": 0.1979,
      "step": 5356
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4956888475605887,
      "learning_rate": 7.070654876107119e-06,
      "loss": 0.2067,
      "step": 5357
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3268758107537053,
      "learning_rate": 7.069600420844234e-06,
      "loss": 0.1832,
      "step": 5358
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3839862518756176,
      "learning_rate": 7.06854585449342e-06,
      "loss": 0.2191,
      "step": 5359
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.494683041128527,
      "learning_rate": 7.067491177111282e-06,
      "loss": 0.7203,
      "step": 5360
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2594638487488858,
      "learning_rate": 7.066436388754432e-06,
      "loss": 0.2007,
      "step": 5361
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4111417080392896,
      "learning_rate": 7.0653814894794836e-06,
      "loss": 0.1662,
      "step": 5362
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.264770813746771,
      "learning_rate": 7.064326479343065e-06,
      "loss": 0.2008,
      "step": 5363
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.447296836396725,
      "learning_rate": 7.063271358401802e-06,
      "loss": 0.1952,
      "step": 5364
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3848085938776455,
      "learning_rate": 7.062216126712328e-06,
      "loss": 0.2218,
      "step": 5365
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2776837965233663,
      "learning_rate": 7.061160784331286e-06,
      "loss": 0.1752,
      "step": 5366
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.342066014003571,
      "learning_rate": 7.060105331315321e-06,
      "loss": 0.2284,
      "step": 5367
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3254528944533246,
      "learning_rate": 7.059049767721087e-06,
      "loss": 0.1757,
      "step": 5368
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2195873163623143,
      "learning_rate": 7.05799409360524e-06,
      "loss": 0.1852,
      "step": 5369
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.560337500701452,
      "learning_rate": 7.056938309024448e-06,
      "loss": 0.6138,
      "step": 5370
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3259779319632197,
      "learning_rate": 7.05588241403538e-06,
      "loss": 0.1852,
      "step": 5371
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3521330819516233,
      "learning_rate": 7.054826408694711e-06,
      "loss": 0.1674,
      "step": 5372
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4023064735465687,
      "learning_rate": 7.053770293059121e-06,
      "loss": 0.1865,
      "step": 5373
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2892231521775988,
      "learning_rate": 7.052714067185303e-06,
      "loss": 0.2264,
      "step": 5374
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3557590815462028,
      "learning_rate": 7.051657731129949e-06,
      "loss": 0.2476,
      "step": 5375
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3966054176055451,
      "learning_rate": 7.050601284949759e-06,
      "loss": 0.2258,
      "step": 5376
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.321655033038373,
      "learning_rate": 7.049544728701439e-06,
      "loss": 0.1782,
      "step": 5377
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2085789231078083,
      "learning_rate": 7.0484880624416985e-06,
      "loss": 0.1836,
      "step": 5378
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.017034507951786,
      "learning_rate": 7.047431286227259e-06,
      "loss": 0.5426,
      "step": 5379
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4172221890454402,
      "learning_rate": 7.046374400114842e-06,
      "loss": 0.1927,
      "step": 5380
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.20038033563381,
      "learning_rate": 7.045317404161175e-06,
      "loss": 0.1884,
      "step": 5381
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5367987925486606,
      "learning_rate": 7.044260298422997e-06,
      "loss": 0.2141,
      "step": 5382
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3449502243111315,
      "learning_rate": 7.043203082957046e-06,
      "loss": 0.1729,
      "step": 5383
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.414854392595381,
      "learning_rate": 7.042145757820073e-06,
      "loss": 0.2144,
      "step": 5384
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3245857028642467,
      "learning_rate": 7.041088323068827e-06,
      "loss": 0.2018,
      "step": 5385
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1095325233540685,
      "learning_rate": 7.04003077876007e-06,
      "loss": 0.1691,
      "step": 5386
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4051072252962102,
      "learning_rate": 7.038973124950564e-06,
      "loss": 0.2538,
      "step": 5387
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2041851912153743,
      "learning_rate": 7.037915361697082e-06,
      "loss": 0.2134,
      "step": 5388
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5227259467220857,
      "learning_rate": 7.036857489056399e-06,
      "loss": 0.1876,
      "step": 5389
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4615992206052881,
      "learning_rate": 7.035799507085299e-06,
      "loss": 0.2376,
      "step": 5390
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4843549189083187,
      "learning_rate": 7.034741415840568e-06,
      "loss": 0.2813,
      "step": 5391
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3731244884684177,
      "learning_rate": 7.033683215379002e-06,
      "loss": 0.1952,
      "step": 5392
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3711030997553977,
      "learning_rate": 7.0326249057574e-06,
      "loss": 0.2118,
      "step": 5393
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3241887600159117,
      "learning_rate": 7.031566487032568e-06,
      "loss": 0.2089,
      "step": 5394
    },
    {
      "epoch": 0.39,
      "grad_norm": 9.17915184631606,
      "learning_rate": 7.030507959261319e-06,
      "loss": 0.7093,
      "step": 5395
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.34254517195933,
      "learning_rate": 7.029449322500469e-06,
      "loss": 0.1908,
      "step": 5396
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.280853949312569,
      "learning_rate": 7.028390576806843e-06,
      "loss": 0.191,
      "step": 5397
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3771367768269365,
      "learning_rate": 7.027331722237268e-06,
      "loss": 0.1945,
      "step": 5398
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3636017755883405,
      "learning_rate": 7.02627275884858e-06,
      "loss": 0.2113,
      "step": 5399
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3583402332203054,
      "learning_rate": 7.0252136866976205e-06,
      "loss": 0.1936,
      "step": 5400
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.229436370670822,
      "learning_rate": 7.024154505841237e-06,
      "loss": 0.204,
      "step": 5401
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.979406682598009,
      "learning_rate": 7.02309521633628e-06,
      "loss": 0.7524,
      "step": 5402
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.475900757305683,
      "learning_rate": 7.0220358182396095e-06,
      "loss": 0.5348,
      "step": 5403
    },
    {
      "epoch": 0.39,
      "grad_norm": 9.415485383395827,
      "learning_rate": 7.0209763116080895e-06,
      "loss": 0.5551,
      "step": 5404
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4152115843077044,
      "learning_rate": 7.019916696498591e-06,
      "loss": 0.2432,
      "step": 5405
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5153297551955902,
      "learning_rate": 7.018856972967988e-06,
      "loss": 0.2284,
      "step": 5406
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3783458263792205,
      "learning_rate": 7.017797141073163e-06,
      "loss": 0.2103,
      "step": 5407
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3352340767878301,
      "learning_rate": 7.0167372008710035e-06,
      "loss": 0.173,
      "step": 5408
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.453324817894412,
      "learning_rate": 7.015677152418404e-06,
      "loss": 0.7066,
      "step": 5409
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.381443235572736,
      "learning_rate": 7.014616995772263e-06,
      "loss": 0.2343,
      "step": 5410
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2977153085166835,
      "learning_rate": 7.013556730989484e-06,
      "loss": 0.2387,
      "step": 5411
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.406639170604141,
      "learning_rate": 7.012496358126979e-06,
      "loss": 0.2136,
      "step": 5412
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7038903414323416,
      "learning_rate": 7.011435877241666e-06,
      "loss": 0.2292,
      "step": 5413
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.292494944421032,
      "learning_rate": 7.010375288390466e-06,
      "loss": 0.2011,
      "step": 5414
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5602545772142864,
      "learning_rate": 7.009314591630306e-06,
      "loss": 0.2444,
      "step": 5415
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4289953442673606,
      "learning_rate": 7.008253787018124e-06,
      "loss": 0.1684,
      "step": 5416
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3671991359227755,
      "learning_rate": 7.0071928746108545e-06,
      "loss": 0.2221,
      "step": 5417
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2581705815559647,
      "learning_rate": 7.006131854465447e-06,
      "loss": 0.2291,
      "step": 5418
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2499704238689193,
      "learning_rate": 7.005070726638849e-06,
      "loss": 0.1659,
      "step": 5419
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1549711068570299,
      "learning_rate": 7.004009491188023e-06,
      "loss": 0.1751,
      "step": 5420
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4573035289690772,
      "learning_rate": 7.002948148169929e-06,
      "loss": 0.2338,
      "step": 5421
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4643026458633595,
      "learning_rate": 7.0018866976415345e-06,
      "loss": 0.1832,
      "step": 5422
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3546310321858084,
      "learning_rate": 7.000825139659815e-06,
      "loss": 0.3,
      "step": 5423
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3087462235220766,
      "learning_rate": 6.999763474281751e-06,
      "loss": 0.1996,
      "step": 5424
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3650633768781233,
      "learning_rate": 6.99870170156433e-06,
      "loss": 0.2212,
      "step": 5425
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3461634538583682,
      "learning_rate": 6.997639821564539e-06,
      "loss": 0.2282,
      "step": 5426
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5369389855175586,
      "learning_rate": 6.996577834339381e-06,
      "loss": 0.25,
      "step": 5427
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3633347594031406,
      "learning_rate": 6.995515739945855e-06,
      "loss": 0.1982,
      "step": 5428
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3766699480481996,
      "learning_rate": 6.994453538440974e-06,
      "loss": 0.1664,
      "step": 5429
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.126979028576072,
      "learning_rate": 6.9933912298817495e-06,
      "loss": 0.1305,
      "step": 5430
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2982718915641807,
      "learning_rate": 6.9923288143252036e-06,
      "loss": 0.1994,
      "step": 5431
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4822797868836177,
      "learning_rate": 6.99126629182836e-06,
      "loss": 0.2401,
      "step": 5432
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5061680226185863,
      "learning_rate": 6.990203662448256e-06,
      "loss": 0.2316,
      "step": 5433
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0945477657779545,
      "learning_rate": 6.989140926241924e-06,
      "loss": 0.1613,
      "step": 5434
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.049624372936082,
      "learning_rate": 6.988078083266411e-06,
      "loss": 0.5779,
      "step": 5435
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3504998789719542,
      "learning_rate": 6.987015133578763e-06,
      "loss": 0.2016,
      "step": 5436
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.509246475289816,
      "learning_rate": 6.985952077236037e-06,
      "loss": 0.5983,
      "step": 5437
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3638594343592807,
      "learning_rate": 6.9848889142952935e-06,
      "loss": 0.2302,
      "step": 5438
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3449807096523152,
      "learning_rate": 6.983825644813599e-06,
      "loss": 0.2057,
      "step": 5439
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3459576211057953,
      "learning_rate": 6.982762268848024e-06,
      "loss": 0.2333,
      "step": 5440
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3806908566081453,
      "learning_rate": 6.981698786455649e-06,
      "loss": 0.2349,
      "step": 5441
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3828263332108028,
      "learning_rate": 6.980635197693556e-06,
      "loss": 0.2067,
      "step": 5442
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3596307606629263,
      "learning_rate": 6.979571502618834e-06,
      "loss": 0.193,
      "step": 5443
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2764381400960554,
      "learning_rate": 6.978507701288577e-06,
      "loss": 0.2017,
      "step": 5444
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3404962934387084,
      "learning_rate": 6.977443793759887e-06,
      "loss": 0.1938,
      "step": 5445
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2246279123538737,
      "learning_rate": 6.976379780089872e-06,
      "loss": 0.2148,
      "step": 5446
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4996250744722766,
      "learning_rate": 6.975315660335641e-06,
      "loss": 0.1708,
      "step": 5447
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4649974659846794,
      "learning_rate": 6.974251434554311e-06,
      "loss": 0.2041,
      "step": 5448
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.267520895710077,
      "learning_rate": 6.97318710280301e-06,
      "loss": 0.202,
      "step": 5449
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.228808055712223,
      "learning_rate": 6.972122665138863e-06,
      "loss": 0.6716,
      "step": 5450
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5637211150013883,
      "learning_rate": 6.9710581216190075e-06,
      "loss": 0.2184,
      "step": 5451
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.329835555251607,
      "learning_rate": 6.96999347230058e-06,
      "loss": 0.1842,
      "step": 5452
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3091830818548698,
      "learning_rate": 6.968928717240731e-06,
      "loss": 0.2013,
      "step": 5453
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6450888668766965,
      "learning_rate": 6.967863856496612e-06,
      "loss": 0.2997,
      "step": 5454
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4801640725200875,
      "learning_rate": 6.966798890125377e-06,
      "loss": 0.2033,
      "step": 5455
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4044223724676201,
      "learning_rate": 6.965733818184193e-06,
      "loss": 0.2347,
      "step": 5456
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3380437086676884,
      "learning_rate": 6.964668640730225e-06,
      "loss": 0.2336,
      "step": 5457
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2845347581958746,
      "learning_rate": 6.9636033578206495e-06,
      "loss": 0.1472,
      "step": 5458
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3182752854169142,
      "learning_rate": 6.962537969512648e-06,
      "loss": 0.1622,
      "step": 5459
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3649039153082538,
      "learning_rate": 6.961472475863406e-06,
      "loss": 0.2531,
      "step": 5460
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3216202290703494,
      "learning_rate": 6.960406876930113e-06,
      "loss": 0.2433,
      "step": 5461
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2192405156866233,
      "learning_rate": 6.959341172769966e-06,
      "loss": 0.2079,
      "step": 5462
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4646986136743858,
      "learning_rate": 6.9582753634401704e-06,
      "loss": 0.2305,
      "step": 5463
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.215859986299084,
      "learning_rate": 6.957209448997932e-06,
      "loss": 0.1838,
      "step": 5464
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3856115455902704,
      "learning_rate": 6.956143429500467e-06,
      "loss": 0.2376,
      "step": 5465
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3726881407635687,
      "learning_rate": 6.955077305004992e-06,
      "loss": 0.2003,
      "step": 5466
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3922409624360077,
      "learning_rate": 6.954011075568737e-06,
      "loss": 0.2354,
      "step": 5467
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3932986112609804,
      "learning_rate": 6.952944741248929e-06,
      "loss": 0.2302,
      "step": 5468
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4091143853419041,
      "learning_rate": 6.951878302102807e-06,
      "loss": 0.174,
      "step": 5469
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3591130073679345,
      "learning_rate": 6.95081175818761e-06,
      "loss": 0.1643,
      "step": 5470
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3231129772971173,
      "learning_rate": 6.949745109560591e-06,
      "loss": 0.2075,
      "step": 5471
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3997652853893072,
      "learning_rate": 6.948678356278998e-06,
      "loss": 0.2142,
      "step": 5472
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2336107661111204,
      "learning_rate": 6.9476114984000945e-06,
      "loss": 0.1821,
      "step": 5473
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4804027467445875,
      "learning_rate": 6.946544535981141e-06,
      "loss": 0.1901,
      "step": 5474
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4334816150192307,
      "learning_rate": 6.945477469079411e-06,
      "loss": 0.1985,
      "step": 5475
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3239666120461713,
      "learning_rate": 6.94441029775218e-06,
      "loss": 0.1998,
      "step": 5476
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3778692413027909,
      "learning_rate": 6.943343022056729e-06,
      "loss": 0.218,
      "step": 5477
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3891162039987508,
      "learning_rate": 6.942275642050343e-06,
      "loss": 0.1768,
      "step": 5478
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.126610743877066,
      "learning_rate": 6.9412081577903186e-06,
      "loss": 0.1515,
      "step": 5479
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.872071658548364,
      "learning_rate": 6.940140569333953e-06,
      "loss": 0.5451,
      "step": 5480
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2669710216213697,
      "learning_rate": 6.939072876738549e-06,
      "loss": 0.2255,
      "step": 5481
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4534831120477456,
      "learning_rate": 6.938005080061414e-06,
      "loss": 0.1918,
      "step": 5482
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3807755411599583,
      "learning_rate": 6.936937179359869e-06,
      "loss": 0.2168,
      "step": 5483
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4334375026592392,
      "learning_rate": 6.935869174691229e-06,
      "loss": 0.2379,
      "step": 5484
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4130295307485325,
      "learning_rate": 6.934801066112824e-06,
      "loss": 0.2176,
      "step": 5485
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6787814601493884,
      "learning_rate": 6.9337328536819825e-06,
      "loss": 0.191,
      "step": 5486
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7966309013808608,
      "learning_rate": 6.932664537456045e-06,
      "loss": 0.1946,
      "step": 5487
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.746680268168713,
      "learning_rate": 6.931596117492353e-06,
      "loss": 0.5163,
      "step": 5488
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4681701638679818,
      "learning_rate": 6.930527593848256e-06,
      "loss": 0.2214,
      "step": 5489
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3769314701402668,
      "learning_rate": 6.929458966581106e-06,
      "loss": 0.2113,
      "step": 5490
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2690878350082115,
      "learning_rate": 6.928390235748264e-06,
      "loss": 0.2087,
      "step": 5491
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2910906935960704,
      "learning_rate": 6.927321401407096e-06,
      "loss": 0.1765,
      "step": 5492
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.327513611497493,
      "learning_rate": 6.926252463614972e-06,
      "loss": 0.1932,
      "step": 5493
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4762999936203332,
      "learning_rate": 6.925183422429268e-06,
      "loss": 0.232,
      "step": 5494
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3630700420103667,
      "learning_rate": 6.924114277907366e-06,
      "loss": 0.1736,
      "step": 5495
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3630609119162944,
      "learning_rate": 6.923045030106655e-06,
      "loss": 0.2032,
      "step": 5496
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4056445452757549,
      "learning_rate": 6.921975679084525e-06,
      "loss": 0.1619,
      "step": 5497
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3221458634519583,
      "learning_rate": 6.920906224898377e-06,
      "loss": 0.1909,
      "step": 5498
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3958109637965124,
      "learning_rate": 6.919836667605614e-06,
      "loss": 0.193,
      "step": 5499
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.302085135385465,
      "learning_rate": 6.918767007263646e-06,
      "loss": 0.1571,
      "step": 5500
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3132205881081422,
      "learning_rate": 6.917697243929887e-06,
      "loss": 0.2357,
      "step": 5501
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3380872205262417,
      "learning_rate": 6.91662737766176e-06,
      "loss": 0.183,
      "step": 5502
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.35832643457043,
      "learning_rate": 6.915557408516688e-06,
      "loss": 0.2139,
      "step": 5503
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4724196446921758,
      "learning_rate": 6.914487336552106e-06,
      "loss": 0.2205,
      "step": 5504
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2336497878247246,
      "learning_rate": 6.913417161825449e-06,
      "loss": 0.1612,
      "step": 5505
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3607706184220547,
      "learning_rate": 6.912346884394161e-06,
      "loss": 0.1695,
      "step": 5506
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4079452052457617,
      "learning_rate": 6.91127650431569e-06,
      "loss": 0.2263,
      "step": 5507
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4843204641975927,
      "learning_rate": 6.910206021647487e-06,
      "loss": 0.2341,
      "step": 5508
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5306619981529381,
      "learning_rate": 6.9091354364470156e-06,
      "loss": 0.1943,
      "step": 5509
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4043490780015915,
      "learning_rate": 6.908064748771738e-06,
      "loss": 0.2259,
      "step": 5510
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2845860301656613,
      "learning_rate": 6.906993958679126e-06,
      "loss": 0.1925,
      "step": 5511
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.587546955656274,
      "learning_rate": 6.9059230662266515e-06,
      "loss": 0.5324,
      "step": 5512
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.416903463322816,
      "learning_rate": 6.904852071471802e-06,
      "loss": 0.2003,
      "step": 5513
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3531753268997397,
      "learning_rate": 6.903780974472059e-06,
      "loss": 0.1796,
      "step": 5514
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.996130215595993,
      "learning_rate": 6.902709775284917e-06,
      "loss": 0.7079,
      "step": 5515
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3704174787830783,
      "learning_rate": 6.9016384739678735e-06,
      "loss": 0.1737,
      "step": 5516
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.416586084777156,
      "learning_rate": 6.900567070578431e-06,
      "loss": 0.2119,
      "step": 5517
    },
    {
      "epoch": 0.39,
      "grad_norm": 9.963018669013438,
      "learning_rate": 6.899495565174099e-06,
      "loss": 0.4257,
      "step": 5518
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5199529408285737,
      "learning_rate": 6.898423957812392e-06,
      "loss": 0.1975,
      "step": 5519
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4478822221640517,
      "learning_rate": 6.897352248550828e-06,
      "loss": 0.2744,
      "step": 5520
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.427538214887589,
      "learning_rate": 6.8962804374469314e-06,
      "loss": 0.211,
      "step": 5521
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3044939390893289,
      "learning_rate": 6.8952085245582376e-06,
      "loss": 0.1981,
      "step": 5522
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4431798240896716,
      "learning_rate": 6.894136509942278e-06,
      "loss": 0.1824,
      "step": 5523
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.421751206974571,
      "learning_rate": 6.893064393656596e-06,
      "loss": 0.2322,
      "step": 5524
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4173752264386053,
      "learning_rate": 6.891992175758739e-06,
      "loss": 0.2112,
      "step": 5525
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1231312836131055,
      "learning_rate": 6.890919856306257e-06,
      "loss": 0.1576,
      "step": 5526
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.7020199159754856,
      "learning_rate": 6.88984743535671e-06,
      "loss": 0.5266,
      "step": 5527
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.785817339048343,
      "learning_rate": 6.88877491296766e-06,
      "loss": 0.5494,
      "step": 5528
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.489519780110349,
      "learning_rate": 6.887702289196677e-06,
      "loss": 0.2483,
      "step": 5529
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5012710029410992,
      "learning_rate": 6.886629564101335e-06,
      "loss": 0.2485,
      "step": 5530
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.315510356681793,
      "learning_rate": 6.8855567377392114e-06,
      "loss": 0.1983,
      "step": 5531
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.34327927053197,
      "learning_rate": 6.884483810167896e-06,
      "loss": 0.2165,
      "step": 5532
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.374256567675251,
      "learning_rate": 6.883410781444974e-06,
      "loss": 0.2331,
      "step": 5533
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5501373669937877,
      "learning_rate": 6.882337651628044e-06,
      "loss": 0.2694,
      "step": 5534
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4138076226603336,
      "learning_rate": 6.881264420774707e-06,
      "loss": 0.2489,
      "step": 5535
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3696030268079102,
      "learning_rate": 6.88019108894257e-06,
      "loss": 0.1817,
      "step": 5536
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4388805252758328,
      "learning_rate": 6.8791176561892445e-06,
      "loss": 0.2189,
      "step": 5537
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.255780910094656,
      "learning_rate": 6.878044122572349e-06,
      "loss": 0.1761,
      "step": 5538
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.464304972671263,
      "learning_rate": 6.876970488149507e-06,
      "loss": 0.1901,
      "step": 5539
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4026020197478846,
      "learning_rate": 6.875896752978345e-06,
      "loss": 0.1867,
      "step": 5540
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3417660176466784,
      "learning_rate": 6.874822917116497e-06,
      "loss": 0.1707,
      "step": 5541
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.87636241256386,
      "learning_rate": 6.873748980621604e-06,
      "loss": 0.6684,
      "step": 5542
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.635198967281351,
      "learning_rate": 6.872674943551311e-06,
      "loss": 0.511,
      "step": 5543
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4265213957806706,
      "learning_rate": 6.871600805963265e-06,
      "loss": 0.1896,
      "step": 5544
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3990489038233138,
      "learning_rate": 6.870526567915123e-06,
      "loss": 0.2034,
      "step": 5545
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.6926395423565115,
      "learning_rate": 6.869452229464547e-06,
      "loss": 0.5113,
      "step": 5546
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.224148616305966,
      "learning_rate": 6.868377790669203e-06,
      "loss": 0.1821,
      "step": 5547
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3059819063753604,
      "learning_rate": 6.867303251586761e-06,
      "loss": 0.187,
      "step": 5548
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.407287523803495,
      "learning_rate": 6.8662286122748965e-06,
      "loss": 0.2338,
      "step": 5549
    },
    {
      "epoch": 0.4,
      "grad_norm": 9.649997347402051,
      "learning_rate": 6.865153872791297e-06,
      "loss": 0.6762,
      "step": 5550
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.267629056663605,
      "learning_rate": 6.864079033193648e-06,
      "loss": 0.1811,
      "step": 5551
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3098261226588972,
      "learning_rate": 6.863004093539641e-06,
      "loss": 0.2184,
      "step": 5552
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.508380338709397,
      "learning_rate": 6.861929053886975e-06,
      "loss": 0.2215,
      "step": 5553
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5479807821706266,
      "learning_rate": 6.860853914293354e-06,
      "loss": 0.2122,
      "step": 5554
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1717678582352498,
      "learning_rate": 6.859778674816488e-06,
      "loss": 0.184,
      "step": 5555
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3997633658897108,
      "learning_rate": 6.858703335514093e-06,
      "loss": 0.2505,
      "step": 5556
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1880016976382144,
      "learning_rate": 6.8576278964438865e-06,
      "loss": 0.1548,
      "step": 5557
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5156639715768516,
      "learning_rate": 6.856552357663595e-06,
      "loss": 0.1811,
      "step": 5558
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4042744538306366,
      "learning_rate": 6.855476719230947e-06,
      "loss": 0.2125,
      "step": 5559
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.7597815200588975,
      "learning_rate": 6.85440098120368e-06,
      "loss": 0.7087,
      "step": 5560
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.859582509334346,
      "learning_rate": 6.853325143639537e-06,
      "loss": 0.4197,
      "step": 5561
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.591397920738184,
      "learning_rate": 6.852249206596263e-06,
      "loss": 0.5732,
      "step": 5562
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1116464862503945,
      "learning_rate": 6.851173170131609e-06,
      "loss": 0.1248,
      "step": 5563
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3032199753861151,
      "learning_rate": 6.850097034303334e-06,
      "loss": 0.1781,
      "step": 5564
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2526028623669665,
      "learning_rate": 6.849020799169201e-06,
      "loss": 0.2041,
      "step": 5565
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.599648992525517,
      "learning_rate": 6.847944464786977e-06,
      "loss": 0.2251,
      "step": 5566
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3575918628520225,
      "learning_rate": 6.846868031214435e-06,
      "loss": 0.1866,
      "step": 5567
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1710525756292172,
      "learning_rate": 6.845791498509354e-06,
      "loss": 0.1807,
      "step": 5568
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3301206328347441,
      "learning_rate": 6.84471486672952e-06,
      "loss": 0.2128,
      "step": 5569
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2307937675753777,
      "learning_rate": 6.84363813593272e-06,
      "loss": 0.1941,
      "step": 5570
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4485222151481953,
      "learning_rate": 6.842561306176748e-06,
      "loss": 0.2173,
      "step": 5571
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4337920419380417,
      "learning_rate": 6.8414843775194066e-06,
      "loss": 0.1882,
      "step": 5572
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4296636398082319,
      "learning_rate": 6.840407350018499e-06,
      "loss": 0.2026,
      "step": 5573
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4097176286519408,
      "learning_rate": 6.839330223731837e-06,
      "loss": 0.2469,
      "step": 5574
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4105148026192633,
      "learning_rate": 6.838252998717234e-06,
      "loss": 0.1928,
      "step": 5575
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4587265494013628,
      "learning_rate": 6.8371756750325145e-06,
      "loss": 0.2201,
      "step": 5576
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3402577320249627,
      "learning_rate": 6.836098252735504e-06,
      "loss": 0.224,
      "step": 5577
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3594176110604488,
      "learning_rate": 6.835020731884034e-06,
      "loss": 0.2413,
      "step": 5578
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3143514713280868,
      "learning_rate": 6.833943112535939e-06,
      "loss": 0.2131,
      "step": 5579
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.459414761332634,
      "learning_rate": 6.832865394749065e-06,
      "loss": 0.1876,
      "step": 5580
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.617409977829471,
      "learning_rate": 6.831787578581259e-06,
      "loss": 0.4895,
      "step": 5581
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4607782064684072,
      "learning_rate": 6.830709664090372e-06,
      "loss": 0.1918,
      "step": 5582
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8059841837737545,
      "learning_rate": 6.829631651334263e-06,
      "loss": 0.2749,
      "step": 5583
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.023705694200093,
      "learning_rate": 6.828553540370796e-06,
      "loss": 0.7598,
      "step": 5584
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1538320438027967,
      "learning_rate": 6.827475331257839e-06,
      "loss": 0.1662,
      "step": 5585
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4485756229003985,
      "learning_rate": 6.826397024053267e-06,
      "loss": 0.2315,
      "step": 5586
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.226058775726674,
      "learning_rate": 6.825318618814959e-06,
      "loss": 0.1556,
      "step": 5587
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.363053413472059,
      "learning_rate": 6.824240115600798e-06,
      "loss": 0.2029,
      "step": 5588
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3321334761367272,
      "learning_rate": 6.823161514468676e-06,
      "loss": 0.1963,
      "step": 5589
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3763765335929299,
      "learning_rate": 6.822082815476486e-06,
      "loss": 0.1846,
      "step": 5590
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.263566843650097,
      "learning_rate": 6.82100401868213e-06,
      "loss": 0.1639,
      "step": 5591
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4228771044736253,
      "learning_rate": 6.819925124143513e-06,
      "loss": 0.1874,
      "step": 5592
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4247225358616686,
      "learning_rate": 6.818846131918545e-06,
      "loss": 0.2197,
      "step": 5593
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.390647622950931,
      "learning_rate": 6.817767042065142e-06,
      "loss": 0.1853,
      "step": 5594
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3395484757380494,
      "learning_rate": 6.816687854641227e-06,
      "loss": 0.1952,
      "step": 5595
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.850204089984936,
      "learning_rate": 6.8156085697047234e-06,
      "loss": 0.622,
      "step": 5596
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.620831558801329,
      "learning_rate": 6.814529187313568e-06,
      "loss": 0.225,
      "step": 5597
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3770183157654075,
      "learning_rate": 6.813449707525692e-06,
      "loss": 0.1756,
      "step": 5598
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5029098051648315,
      "learning_rate": 6.812370130399041e-06,
      "loss": 0.2018,
      "step": 5599
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4312658274499137,
      "learning_rate": 6.811290455991561e-06,
      "loss": 0.1681,
      "step": 5600
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3536552620855886,
      "learning_rate": 6.810210684361206e-06,
      "loss": 0.226,
      "step": 5601
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.307051465859498,
      "learning_rate": 6.809130815565932e-06,
      "loss": 0.1979,
      "step": 5602
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2819170194886846,
      "learning_rate": 6.8080508496637045e-06,
      "loss": 0.1659,
      "step": 5603
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2715534584094037,
      "learning_rate": 6.806970786712489e-06,
      "loss": 0.2147,
      "step": 5604
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2244668718626839,
      "learning_rate": 6.8058906267702604e-06,
      "loss": 0.1865,
      "step": 5605
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.854578091455639,
      "learning_rate": 6.804810369894998e-06,
      "loss": 0.6661,
      "step": 5606
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5679196216002247,
      "learning_rate": 6.803730016144686e-06,
      "loss": 0.2461,
      "step": 5607
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2946201530421633,
      "learning_rate": 6.802649565577311e-06,
      "loss": 0.1775,
      "step": 5608
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.457927186726712,
      "learning_rate": 6.8015690182508694e-06,
      "loss": 0.1969,
      "step": 5609
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4387634054821987,
      "learning_rate": 6.800488374223361e-06,
      "loss": 0.1888,
      "step": 5610
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2634647040079436,
      "learning_rate": 6.7994076335527895e-06,
      "loss": 0.1804,
      "step": 5611
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2913501710486008,
      "learning_rate": 6.7983267962971646e-06,
      "loss": 0.221,
      "step": 5612
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4463424421840554,
      "learning_rate": 6.7972458625145e-06,
      "loss": 0.2387,
      "step": 5613
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1865963361719756,
      "learning_rate": 6.79616483226282e-06,
      "loss": 0.2194,
      "step": 5614
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.333945270110938,
      "learning_rate": 6.795083705600147e-06,
      "loss": 0.2099,
      "step": 5615
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5012291318917845,
      "learning_rate": 6.794002482584511e-06,
      "loss": 0.1558,
      "step": 5616
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.303196211375126,
      "learning_rate": 6.792921163273949e-06,
      "loss": 0.5494,
      "step": 5617
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2807506122899042,
      "learning_rate": 6.7918397477265e-06,
      "loss": 0.1928,
      "step": 5618
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3710794262638142,
      "learning_rate": 6.790758236000217e-06,
      "loss": 0.2087,
      "step": 5619
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2816986408208557,
      "learning_rate": 6.7896766281531435e-06,
      "loss": 0.1958,
      "step": 5620
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.30825114772741,
      "learning_rate": 6.788594924243339e-06,
      "loss": 0.1722,
      "step": 5621
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3143265759851857,
      "learning_rate": 6.787513124328863e-06,
      "loss": 0.219,
      "step": 5622
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.244290675604723,
      "learning_rate": 6.786431228467786e-06,
      "loss": 0.1723,
      "step": 5623
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3061247617568577,
      "learning_rate": 6.785349236718178e-06,
      "loss": 0.1745,
      "step": 5624
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4454913451299987,
      "learning_rate": 6.784267149138114e-06,
      "loss": 0.2168,
      "step": 5625
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.346377383351855,
      "learning_rate": 6.783184965785678e-06,
      "loss": 0.1954,
      "step": 5626
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4333145754571348,
      "learning_rate": 6.782102686718959e-06,
      "loss": 0.2161,
      "step": 5627
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3016011363404663,
      "learning_rate": 6.781020311996046e-06,
      "loss": 0.2111,
      "step": 5628
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.668174190513239,
      "learning_rate": 6.779937841675039e-06,
      "loss": 0.6214,
      "step": 5629
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7338934574235783,
      "learning_rate": 6.77885527581404e-06,
      "loss": 0.1882,
      "step": 5630
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4228248548587812,
      "learning_rate": 6.777772614471157e-06,
      "loss": 0.1925,
      "step": 5631
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4672302890507156,
      "learning_rate": 6.7766898577045025e-06,
      "loss": 0.2483,
      "step": 5632
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2541623101007504,
      "learning_rate": 6.775607005572196e-06,
      "loss": 0.2088,
      "step": 5633
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.275631519261516,
      "learning_rate": 6.7745240581323564e-06,
      "loss": 0.5352,
      "step": 5634
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3223278163794383,
      "learning_rate": 6.773441015443118e-06,
      "loss": 0.2313,
      "step": 5635
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7273539639985924,
      "learning_rate": 6.772357877562611e-06,
      "loss": 0.1882,
      "step": 5636
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1681766348586549,
      "learning_rate": 6.771274644548974e-06,
      "loss": 0.1408,
      "step": 5637
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.340093170375682,
      "learning_rate": 6.770191316460351e-06,
      "loss": 0.2237,
      "step": 5638
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2664269241767787,
      "learning_rate": 6.7691078933548906e-06,
      "loss": 0.2034,
      "step": 5639
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5414294694474973,
      "learning_rate": 6.768024375290747e-06,
      "loss": 0.2414,
      "step": 5640
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4095764473750485,
      "learning_rate": 6.7669407623260795e-06,
      "loss": 0.205,
      "step": 5641
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6811992346321818,
      "learning_rate": 6.765857054519049e-06,
      "loss": 0.2113,
      "step": 5642
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3459242541542007,
      "learning_rate": 6.764773251927829e-06,
      "loss": 0.1846,
      "step": 5643
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.252840276206209,
      "learning_rate": 6.763689354610593e-06,
      "loss": 0.1494,
      "step": 5644
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.40161470572911,
      "learning_rate": 6.762605362625519e-06,
      "loss": 0.2016,
      "step": 5645
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.284078616107917,
      "learning_rate": 6.761521276030789e-06,
      "loss": 0.1993,
      "step": 5646
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.362782444433252,
      "learning_rate": 6.760437094884598e-06,
      "loss": 0.2081,
      "step": 5647
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4501700912105644,
      "learning_rate": 6.7593528192451365e-06,
      "loss": 0.2136,
      "step": 5648
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.472602136339102,
      "learning_rate": 6.758268449170604e-06,
      "loss": 0.7433,
      "step": 5649
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2875915689205593,
      "learning_rate": 6.757183984719208e-06,
      "loss": 0.1697,
      "step": 5650
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2809689059353946,
      "learning_rate": 6.756099425949155e-06,
      "loss": 0.1538,
      "step": 5651
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3446215599964646,
      "learning_rate": 6.7550147729186635e-06,
      "loss": 0.2251,
      "step": 5652
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2453541675547508,
      "learning_rate": 6.753930025685951e-06,
      "loss": 0.211,
      "step": 5653
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2107568020733361,
      "learning_rate": 6.752845184309241e-06,
      "loss": 0.1778,
      "step": 5654
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.242823192084643,
      "learning_rate": 6.751760248846766e-06,
      "loss": 0.1985,
      "step": 5655
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.285560873477995,
      "learning_rate": 6.7506752193567605e-06,
      "loss": 0.1797,
      "step": 5656
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.294694911178069,
      "learning_rate": 6.749590095897464e-06,
      "loss": 0.1829,
      "step": 5657
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4382540341219394,
      "learning_rate": 6.748504878527124e-06,
      "loss": 0.2331,
      "step": 5658
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.490172429199845,
      "learning_rate": 6.747419567303985e-06,
      "loss": 0.6948,
      "step": 5659
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5703790270565599,
      "learning_rate": 6.7463341622863074e-06,
      "loss": 0.2278,
      "step": 5660
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3390537479883828,
      "learning_rate": 6.745248663532351e-06,
      "loss": 0.19,
      "step": 5661
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4671708164590929,
      "learning_rate": 6.744163071100378e-06,
      "loss": 0.2003,
      "step": 5662
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2791333480531302,
      "learning_rate": 6.743077385048662e-06,
      "loss": 0.2,
      "step": 5663
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3189994027736311,
      "learning_rate": 6.741991605435475e-06,
      "loss": 0.2095,
      "step": 5664
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2622877259411815,
      "learning_rate": 6.7409057323191e-06,
      "loss": 0.2167,
      "step": 5665
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2798733050476965,
      "learning_rate": 6.7398197657578224e-06,
      "loss": 0.1964,
      "step": 5666
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3298721564993314,
      "learning_rate": 6.738733705809932e-06,
      "loss": 0.2021,
      "step": 5667
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4185326127884075,
      "learning_rate": 6.737647552533722e-06,
      "loss": 0.2189,
      "step": 5668
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.825427042928148,
      "learning_rate": 6.736561305987495e-06,
      "loss": 0.6682,
      "step": 5669
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4556261443779515,
      "learning_rate": 6.735474966229558e-06,
      "loss": 0.253,
      "step": 5670
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2609450123145123,
      "learning_rate": 6.734388533318219e-06,
      "loss": 0.1584,
      "step": 5671
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1865214803102657,
      "learning_rate": 6.733302007311792e-06,
      "loss": 0.1879,
      "step": 5672
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3307829453079547,
      "learning_rate": 6.7322153882686e-06,
      "loss": 0.1891,
      "step": 5673
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1971321588825528,
      "learning_rate": 6.7311286762469695e-06,
      "loss": 0.1632,
      "step": 5674
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.306003062447459,
      "learning_rate": 6.730041871305227e-06,
      "loss": 0.2139,
      "step": 5675
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5119779248431573,
      "learning_rate": 6.72895497350171e-06,
      "loss": 0.2152,
      "step": 5676
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3437741442558313,
      "learning_rate": 6.72786798289476e-06,
      "loss": 0.2398,
      "step": 5677
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3986115793524223,
      "learning_rate": 6.726780899542721e-06,
      "loss": 0.1945,
      "step": 5678
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4754978141769395,
      "learning_rate": 6.725693723503943e-06,
      "loss": 0.2211,
      "step": 5679
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.289503148716925,
      "learning_rate": 6.724606454836782e-06,
      "loss": 0.1946,
      "step": 5680
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2160744511196684,
      "learning_rate": 6.723519093599597e-06,
      "loss": 0.1879,
      "step": 5681
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3508942105590303,
      "learning_rate": 6.722431639850755e-06,
      "loss": 0.1852,
      "step": 5682
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4369667744563324,
      "learning_rate": 6.721344093648625e-06,
      "loss": 0.2123,
      "step": 5683
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5170552888975881,
      "learning_rate": 6.720256455051583e-06,
      "loss": 0.227,
      "step": 5684
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4636048633477217,
      "learning_rate": 6.719168724118008e-06,
      "loss": 0.1767,
      "step": 5685
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2595801185377093,
      "learning_rate": 6.718080900906288e-06,
      "loss": 0.1974,
      "step": 5686
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2092727413352022,
      "learning_rate": 6.716992985474808e-06,
      "loss": 0.1826,
      "step": 5687
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3672481777544003,
      "learning_rate": 6.715904977881968e-06,
      "loss": 0.2308,
      "step": 5688
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1813708068878224,
      "learning_rate": 6.714816878186164e-06,
      "loss": 0.185,
      "step": 5689
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5669579216180372,
      "learning_rate": 6.713728686445804e-06,
      "loss": 0.2299,
      "step": 5690
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4903699159583168,
      "learning_rate": 6.712640402719296e-06,
      "loss": 0.2031,
      "step": 5691
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2527154817357304,
      "learning_rate": 6.7115520270650555e-06,
      "loss": 0.2142,
      "step": 5692
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3861229545733509,
      "learning_rate": 6.7104635595415005e-06,
      "loss": 0.2038,
      "step": 5693
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3582388168366353,
      "learning_rate": 6.709375000207058e-06,
      "loss": 0.2402,
      "step": 5694
    },
    {
      "epoch": 0.41,
      "grad_norm": 10.378207215998074,
      "learning_rate": 6.708286349120157e-06,
      "loss": 0.7471,
      "step": 5695
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3097393518637221,
      "learning_rate": 6.707197606339231e-06,
      "loss": 0.229,
      "step": 5696
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4454186492776733,
      "learning_rate": 6.706108771922718e-06,
      "loss": 0.1992,
      "step": 5697
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4415416407224941,
      "learning_rate": 6.7050198459290665e-06,
      "loss": 0.1991,
      "step": 5698
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3622344353971196,
      "learning_rate": 6.7039308284167215e-06,
      "loss": 0.1746,
      "step": 5699
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5682298830567587,
      "learning_rate": 6.702841719444141e-06,
      "loss": 0.1946,
      "step": 5700
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4176130913484848,
      "learning_rate": 6.70175251906978e-06,
      "loss": 0.1868,
      "step": 5701
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5217291404470525,
      "learning_rate": 6.7006632273521054e-06,
      "loss": 0.2398,
      "step": 5702
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3771729268140283,
      "learning_rate": 6.699573844349585e-06,
      "loss": 0.1914,
      "step": 5703
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.298716872644517,
      "learning_rate": 6.698484370120694e-06,
      "loss": 0.2126,
      "step": 5704
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3101969439906083,
      "learning_rate": 6.697394804723909e-06,
      "loss": 0.1904,
      "step": 5705
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3257733140367038,
      "learning_rate": 6.696305148217712e-06,
      "loss": 0.234,
      "step": 5706
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3260888015245609,
      "learning_rate": 6.695215400660597e-06,
      "loss": 0.2436,
      "step": 5707
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.293972584864335,
      "learning_rate": 6.694125562111052e-06,
      "loss": 0.1883,
      "step": 5708
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.301660574142384,
      "learning_rate": 6.693035632627578e-06,
      "loss": 0.2161,
      "step": 5709
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4970658983587233,
      "learning_rate": 6.6919456122686764e-06,
      "loss": 0.1842,
      "step": 5710
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2451956593961513,
      "learning_rate": 6.690855501092858e-06,
      "loss": 0.1904,
      "step": 5711
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2251760310389341,
      "learning_rate": 6.689765299158633e-06,
      "loss": 0.1505,
      "step": 5712
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3432077675301248,
      "learning_rate": 6.688675006524518e-06,
      "loss": 0.2385,
      "step": 5713
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.70178491501225,
      "learning_rate": 6.6875846232490395e-06,
      "loss": 0.5519,
      "step": 5714
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.803092414746553,
      "learning_rate": 6.686494149390721e-06,
      "loss": 0.6494,
      "step": 5715
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3847034323477567,
      "learning_rate": 6.685403585008098e-06,
      "loss": 0.1738,
      "step": 5716
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5130012561071726,
      "learning_rate": 6.684312930159706e-06,
      "loss": 0.1969,
      "step": 5717
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3469290735108792,
      "learning_rate": 6.683222184904088e-06,
      "loss": 0.2216,
      "step": 5718
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.552237356884769,
      "learning_rate": 6.682131349299789e-06,
      "loss": 0.2605,
      "step": 5719
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3784043334802591,
      "learning_rate": 6.681040423405363e-06,
      "loss": 0.2073,
      "step": 5720
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1621381142808658,
      "learning_rate": 6.6799494072793646e-06,
      "loss": 0.156,
      "step": 5721
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3307904565468147,
      "learning_rate": 6.678858300980356e-06,
      "loss": 0.169,
      "step": 5722
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2790254033946793,
      "learning_rate": 6.677767104566904e-06,
      "loss": 0.1715,
      "step": 5723
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3824441325089205,
      "learning_rate": 6.67667581809758e-06,
      "loss": 0.1919,
      "step": 5724
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.280187317096119,
      "learning_rate": 6.675584441630958e-06,
      "loss": 0.1869,
      "step": 5725
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4456989285244566,
      "learning_rate": 6.6744929752256215e-06,
      "loss": 0.2211,
      "step": 5726
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2500894170526489,
      "learning_rate": 6.673401418940152e-06,
      "loss": 0.1908,
      "step": 5727
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.357103916117228,
      "learning_rate": 6.672309772833144e-06,
      "loss": 0.2045,
      "step": 5728
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2699251316660984,
      "learning_rate": 6.671218036963192e-06,
      "loss": 0.209,
      "step": 5729
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3158350854769476,
      "learning_rate": 6.670126211388894e-06,
      "loss": 0.197,
      "step": 5730
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4958896619074076,
      "learning_rate": 6.669034296168855e-06,
      "loss": 0.1965,
      "step": 5731
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4591919793204364,
      "learning_rate": 6.667942291361687e-06,
      "loss": 0.2558,
      "step": 5732
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4072871061940093,
      "learning_rate": 6.666850197026003e-06,
      "loss": 0.1704,
      "step": 5733
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.110998594507066,
      "learning_rate": 6.665758013220422e-06,
      "loss": 0.18,
      "step": 5734
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2012612616825202,
      "learning_rate": 6.66466574000357e-06,
      "loss": 0.1495,
      "step": 5735
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3198798354609544,
      "learning_rate": 6.663573377434074e-06,
      "loss": 0.2195,
      "step": 5736
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2347107647484115,
      "learning_rate": 6.662480925570567e-06,
      "loss": 0.1923,
      "step": 5737
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2981680362855919,
      "learning_rate": 6.6613883844716894e-06,
      "loss": 0.1684,
      "step": 5738
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3114960686925083,
      "learning_rate": 6.660295754196082e-06,
      "loss": 0.193,
      "step": 5739
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4874291814930212,
      "learning_rate": 6.659203034802397e-06,
      "loss": 0.261,
      "step": 5740
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2083904418057052,
      "learning_rate": 6.658110226349284e-06,
      "loss": 0.1605,
      "step": 5741
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.362892995611294,
      "learning_rate": 6.657017328895401e-06,
      "loss": 0.1785,
      "step": 5742
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.306355006571541,
      "learning_rate": 6.65592434249941e-06,
      "loss": 0.1919,
      "step": 5743
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4577068238341349,
      "learning_rate": 6.6548312672199815e-06,
      "loss": 0.243,
      "step": 5744
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4651108479679416,
      "learning_rate": 6.653738103115781e-06,
      "loss": 0.2136,
      "step": 5745
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.259625944798431,
      "learning_rate": 6.6526448502454925e-06,
      "loss": 0.1883,
      "step": 5746
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5029829739662313,
      "learning_rate": 6.651551508667793e-06,
      "loss": 0.2011,
      "step": 5747
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4200453788687677,
      "learning_rate": 6.650458078441368e-06,
      "loss": 0.227,
      "step": 5748
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3398797259999171,
      "learning_rate": 6.649364559624912e-06,
      "loss": 0.2113,
      "step": 5749
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2671121825600271,
      "learning_rate": 6.6482709522771196e-06,
      "loss": 0.1869,
      "step": 5750
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3590177332533206,
      "learning_rate": 6.647177256456691e-06,
      "loss": 0.2686,
      "step": 5751
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3245832044176813,
      "learning_rate": 6.646083472222329e-06,
      "loss": 0.198,
      "step": 5752
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.457116661497489,
      "learning_rate": 6.644989599632747e-06,
      "loss": 0.1844,
      "step": 5753
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4003388410510098,
      "learning_rate": 6.643895638746658e-06,
      "loss": 0.2018,
      "step": 5754
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4306379403620073,
      "learning_rate": 6.6428015896227826e-06,
      "loss": 0.2297,
      "step": 5755
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2520424959493313,
      "learning_rate": 6.641707452319845e-06,
      "loss": 0.1669,
      "step": 5756
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.458495847628931,
      "learning_rate": 6.6406132268965716e-06,
      "loss": 0.1905,
      "step": 5757
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4327577804334437,
      "learning_rate": 6.6395189134117e-06,
      "loss": 0.2105,
      "step": 5758
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5175703626853256,
      "learning_rate": 6.638424511923965e-06,
      "loss": 0.2459,
      "step": 5759
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.30495049377198,
      "learning_rate": 6.637330022492112e-06,
      "loss": 0.1746,
      "step": 5760
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4175972507024075,
      "learning_rate": 6.6362354451748876e-06,
      "loss": 0.2281,
      "step": 5761
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6614805020605328,
      "learning_rate": 6.635140780031047e-06,
      "loss": 0.2368,
      "step": 5762
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.860031903398821,
      "learning_rate": 6.634046027119343e-06,
      "loss": 0.7272,
      "step": 5763
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1892387029159839,
      "learning_rate": 6.632951186498542e-06,
      "loss": 0.168,
      "step": 5764
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.354577102683608,
      "learning_rate": 6.631856258227406e-06,
      "loss": 0.2307,
      "step": 5765
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.434940027389757,
      "learning_rate": 6.63076124236471e-06,
      "loss": 0.2427,
      "step": 5766
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6738306611269618,
      "learning_rate": 6.629666138969231e-06,
      "loss": 0.2289,
      "step": 5767
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2407976139071275,
      "learning_rate": 6.628570948099746e-06,
      "loss": 0.1705,
      "step": 5768
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2711675218171379,
      "learning_rate": 6.627475669815043e-06,
      "loss": 0.1932,
      "step": 5769
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2918210848770273,
      "learning_rate": 6.626380304173913e-06,
      "loss": 0.2035,
      "step": 5770
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3993038185093003,
      "learning_rate": 6.6252848512351475e-06,
      "loss": 0.2013,
      "step": 5771
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.481145230863057,
      "learning_rate": 6.62418931105755e-06,
      "loss": 0.2196,
      "step": 5772
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.566701943275558,
      "learning_rate": 6.623093683699921e-06,
      "loss": 0.2832,
      "step": 5773
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3180289505129423,
      "learning_rate": 6.621997969221072e-06,
      "loss": 0.1873,
      "step": 5774
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2334155726627893,
      "learning_rate": 6.620902167679817e-06,
      "loss": 0.2202,
      "step": 5775
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3969341302125762,
      "learning_rate": 6.6198062791349715e-06,
      "loss": 0.235,
      "step": 5776
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4117737649966786,
      "learning_rate": 6.618710303645361e-06,
      "loss": 0.2173,
      "step": 5777
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.339795476748654,
      "learning_rate": 6.617614241269811e-06,
      "loss": 0.2019,
      "step": 5778
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2367667482086984,
      "learning_rate": 6.616518092067155e-06,
      "loss": 0.1569,
      "step": 5779
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4824745373651798,
      "learning_rate": 6.615421856096231e-06,
      "loss": 0.2413,
      "step": 5780
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5265366193080834,
      "learning_rate": 6.614325533415879e-06,
      "loss": 0.218,
      "step": 5781
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.1946566010934205,
      "learning_rate": 6.613229124084947e-06,
      "loss": 0.5893,
      "step": 5782
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2337097864822355,
      "learning_rate": 6.612132628162283e-06,
      "loss": 0.1676,
      "step": 5783
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4905616945451448,
      "learning_rate": 6.611036045706746e-06,
      "loss": 0.2212,
      "step": 5784
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.125834108847531,
      "learning_rate": 6.609939376777192e-06,
      "loss": 0.5499,
      "step": 5785
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2510642605702738,
      "learning_rate": 6.60884262143249e-06,
      "loss": 0.1854,
      "step": 5786
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2998781303620168,
      "learning_rate": 6.607745779731507e-06,
      "loss": 0.1792,
      "step": 5787
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2044164390816559,
      "learning_rate": 6.606648851733119e-06,
      "loss": 0.2096,
      "step": 5788
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4204717989247857,
      "learning_rate": 6.6055518374962025e-06,
      "loss": 0.1722,
      "step": 5789
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.37736387519301,
      "learning_rate": 6.604454737079639e-06,
      "loss": 0.1922,
      "step": 5790
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4526356734509236,
      "learning_rate": 6.603357550542324e-06,
      "loss": 0.1842,
      "step": 5791
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.206607547875254,
      "learning_rate": 6.602260277943143e-06,
      "loss": 0.1975,
      "step": 5792
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2891778931473696,
      "learning_rate": 6.601162919340998e-06,
      "loss": 0.2219,
      "step": 5793
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3345675469457485,
      "learning_rate": 6.600065474794786e-06,
      "loss": 0.226,
      "step": 5794
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3385439372006158,
      "learning_rate": 6.5989679443634165e-06,
      "loss": 0.1717,
      "step": 5795
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2162891827977884,
      "learning_rate": 6.597870328105801e-06,
      "loss": 0.1972,
      "step": 5796
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3114315083343318,
      "learning_rate": 6.596772626080854e-06,
      "loss": 0.1994,
      "step": 5797
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3799082848016855,
      "learning_rate": 6.595674838347496e-06,
      "loss": 0.2021,
      "step": 5798
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3339837257477711,
      "learning_rate": 6.594576964964652e-06,
      "loss": 0.2069,
      "step": 5799
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.373644131742913,
      "learning_rate": 6.593479005991251e-06,
      "loss": 0.498,
      "step": 5800
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4698861948278468,
      "learning_rate": 6.592380961486228e-06,
      "loss": 0.2246,
      "step": 5801
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4385713560968467,
      "learning_rate": 6.591282831508521e-06,
      "loss": 0.1958,
      "step": 5802
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4490409354008162,
      "learning_rate": 6.590184616117073e-06,
      "loss": 0.2473,
      "step": 5803
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3637450520681709,
      "learning_rate": 6.5890863153708325e-06,
      "loss": 0.1981,
      "step": 5804
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1372947335089096,
      "learning_rate": 6.587987929328753e-06,
      "loss": 0.1718,
      "step": 5805
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2763116248491544,
      "learning_rate": 6.586889458049789e-06,
      "loss": 0.1752,
      "step": 5806
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3777730365982443,
      "learning_rate": 6.585790901592903e-06,
      "loss": 0.187,
      "step": 5807
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.400989115968126,
      "learning_rate": 6.584692260017062e-06,
      "loss": 0.2244,
      "step": 5808
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.455140337920268,
      "learning_rate": 6.583593533381235e-06,
      "loss": 0.2154,
      "step": 5809
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.028481766449438,
      "learning_rate": 6.582494721744399e-06,
      "loss": 0.7127,
      "step": 5810
    },
    {
      "epoch": 0.42,
      "grad_norm": 3.828653839167092,
      "learning_rate": 6.581395825165534e-06,
      "loss": 0.4907,
      "step": 5811
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2944491358521215,
      "learning_rate": 6.580296843703623e-06,
      "loss": 0.1676,
      "step": 5812
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.288376814289655,
      "learning_rate": 6.579197777417656e-06,
      "loss": 0.1917,
      "step": 5813
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4308751371931998,
      "learning_rate": 6.578098626366627e-06,
      "loss": 0.2248,
      "step": 5814
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3307813109395805,
      "learning_rate": 6.576999390609532e-06,
      "loss": 0.1887,
      "step": 5815
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.404066257320106,
      "learning_rate": 6.575900070205375e-06,
      "loss": 0.2622,
      "step": 5816
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.363946171834274,
      "learning_rate": 6.574800665213162e-06,
      "loss": 0.4892,
      "step": 5817
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1342565562742273,
      "learning_rate": 6.5737011756919065e-06,
      "loss": 0.1559,
      "step": 5818
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3390629374145653,
      "learning_rate": 6.572601601700626e-06,
      "loss": 0.2015,
      "step": 5819
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3552363009522392,
      "learning_rate": 6.571501943298335e-06,
      "loss": 0.2059,
      "step": 5820
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.413615809833346,
      "learning_rate": 6.570402200544066e-06,
      "loss": 0.2244,
      "step": 5821
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.726952335686051,
      "learning_rate": 6.569302373496844e-06,
      "loss": 0.2532,
      "step": 5822
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.385709602737147,
      "learning_rate": 6.568202462215707e-06,
      "loss": 0.2194,
      "step": 5823
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3777678397484279,
      "learning_rate": 6.567102466759689e-06,
      "loss": 0.1682,
      "step": 5824
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4669834801301234,
      "learning_rate": 6.5660023871878395e-06,
      "loss": 0.1959,
      "step": 5825
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.437592375258112,
      "learning_rate": 6.5649022235592016e-06,
      "loss": 0.5079,
      "step": 5826
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3516275588362223,
      "learning_rate": 6.563801975932831e-06,
      "loss": 0.2358,
      "step": 5827
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3645295875037928,
      "learning_rate": 6.56270164436778e-06,
      "loss": 0.2029,
      "step": 5828
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4078522411267933,
      "learning_rate": 6.561601228923116e-06,
      "loss": 0.2348,
      "step": 5829
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3644431828728858,
      "learning_rate": 6.5605007296579e-06,
      "loss": 0.214,
      "step": 5830
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2750978268468776,
      "learning_rate": 6.559400146631205e-06,
      "loss": 0.2012,
      "step": 5831
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3505032187284072,
      "learning_rate": 6.558299479902105e-06,
      "loss": 0.2188,
      "step": 5832
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1041574253780873,
      "learning_rate": 6.55719872952968e-06,
      "loss": 0.1777,
      "step": 5833
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3513393518954229,
      "learning_rate": 6.556097895573015e-06,
      "loss": 0.2246,
      "step": 5834
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.522512949439286,
      "learning_rate": 6.554996978091195e-06,
      "loss": 0.1811,
      "step": 5835
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2674060529948155,
      "learning_rate": 6.553895977143316e-06,
      "loss": 0.22,
      "step": 5836
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.168520078217555,
      "learning_rate": 6.552794892788473e-06,
      "loss": 0.18,
      "step": 5837
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4865238283399682,
      "learning_rate": 6.55169372508577e-06,
      "loss": 0.237,
      "step": 5838
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4676534859000676,
      "learning_rate": 6.550592474094313e-06,
      "loss": 0.2596,
      "step": 5839
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1856485615316945,
      "learning_rate": 6.549491139873211e-06,
      "loss": 0.1789,
      "step": 5840
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1907948669397783,
      "learning_rate": 6.548389722481579e-06,
      "loss": 0.1832,
      "step": 5841
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2804324033510017,
      "learning_rate": 6.54728822197854e-06,
      "loss": 0.2482,
      "step": 5842
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.395626716107591,
      "learning_rate": 6.546186638423218e-06,
      "loss": 0.2046,
      "step": 5843
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3416283520985277,
      "learning_rate": 6.545084971874738e-06,
      "loss": 0.208,
      "step": 5844
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2694107596100175,
      "learning_rate": 6.543983222392234e-06,
      "loss": 0.1903,
      "step": 5845
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1308645061090474,
      "learning_rate": 6.542881390034846e-06,
      "loss": 0.176,
      "step": 5846
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2070393017771712,
      "learning_rate": 6.541779474861715e-06,
      "loss": 0.1758,
      "step": 5847
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2109682595934381,
      "learning_rate": 6.540677476931987e-06,
      "loss": 0.1598,
      "step": 5848
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3516036121659398,
      "learning_rate": 6.53957539630481e-06,
      "loss": 0.1809,
      "step": 5849
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.263059345262111,
      "learning_rate": 6.5384732330393455e-06,
      "loss": 0.204,
      "step": 5850
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.737474395910261,
      "learning_rate": 6.537370987194749e-06,
      "loss": 0.4954,
      "step": 5851
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3643204377707794,
      "learning_rate": 6.536268658830185e-06,
      "loss": 0.2165,
      "step": 5852
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.396008763518398,
      "learning_rate": 6.535166248004821e-06,
      "loss": 0.1951,
      "step": 5853
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1608353835904177,
      "learning_rate": 6.534063754777835e-06,
      "loss": 0.1839,
      "step": 5854
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3736640493188175,
      "learning_rate": 6.532961179208399e-06,
      "loss": 0.2163,
      "step": 5855
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4782450295109213,
      "learning_rate": 6.531858521355699e-06,
      "loss": 0.1798,
      "step": 5856
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4002589477576388,
      "learning_rate": 6.530755781278918e-06,
      "loss": 0.2173,
      "step": 5857
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.793937893111318,
      "learning_rate": 6.529652959037247e-06,
      "loss": 0.567,
      "step": 5858
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2277525259011342,
      "learning_rate": 6.528550054689884e-06,
      "loss": 0.1657,
      "step": 5859
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4775171224782773,
      "learning_rate": 6.527447068296026e-06,
      "loss": 0.2176,
      "step": 5860
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3015551534841745,
      "learning_rate": 6.526343999914877e-06,
      "loss": 0.2016,
      "step": 5861
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5000661439196052,
      "learning_rate": 6.525240849605646e-06,
      "loss": 0.1896,
      "step": 5862
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2988463326285618,
      "learning_rate": 6.524137617427546e-06,
      "loss": 0.2214,
      "step": 5863
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.463932839366527,
      "learning_rate": 6.523034303439794e-06,
      "loss": 0.2005,
      "step": 5864
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4553030842357515,
      "learning_rate": 6.521930907701612e-06,
      "loss": 0.2342,
      "step": 5865
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.7450200298018315,
      "learning_rate": 6.520827430272223e-06,
      "loss": 0.625,
      "step": 5866
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.179887560028545,
      "learning_rate": 6.5197238712108616e-06,
      "loss": 0.1577,
      "step": 5867
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5002743010074722,
      "learning_rate": 6.51862023057676e-06,
      "loss": 0.2469,
      "step": 5868
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3344591606642755,
      "learning_rate": 6.517516508429157e-06,
      "loss": 0.1892,
      "step": 5869
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2166895454585263,
      "learning_rate": 6.5164127048272966e-06,
      "loss": 0.1681,
      "step": 5870
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.361521785623413,
      "learning_rate": 6.515308819830426e-06,
      "loss": 0.2114,
      "step": 5871
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.154625262597219,
      "learning_rate": 6.514204853497799e-06,
      "loss": 0.6385,
      "step": 5872
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2436883536414283,
      "learning_rate": 6.51310080588867e-06,
      "loss": 0.1713,
      "step": 5873
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.38674664660239,
      "learning_rate": 6.511996677062302e-06,
      "loss": 0.1874,
      "step": 5874
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2431111317813046,
      "learning_rate": 6.510892467077958e-06,
      "loss": 0.1966,
      "step": 5875
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3116195008222193,
      "learning_rate": 6.5097881759949096e-06,
      "loss": 0.177,
      "step": 5876
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.291090736803952,
      "learning_rate": 6.508683803872431e-06,
      "loss": 0.1968,
      "step": 5877
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5268017496508894,
      "learning_rate": 6.5075793507698e-06,
      "loss": 0.2454,
      "step": 5878
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4881959677282703,
      "learning_rate": 6.506474816746296e-06,
      "loss": 0.2634,
      "step": 5879
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.366364070181591,
      "learning_rate": 6.50537020186121e-06,
      "loss": 0.2164,
      "step": 5880
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.327254732488619,
      "learning_rate": 6.504265506173832e-06,
      "loss": 0.2157,
      "step": 5881
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5096418726092107,
      "learning_rate": 6.503160729743459e-06,
      "loss": 0.2195,
      "step": 5882
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.344942739348234,
      "learning_rate": 6.502055872629387e-06,
      "loss": 0.1703,
      "step": 5883
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6419840901517802,
      "learning_rate": 6.500950934890925e-06,
      "loss": 0.2398,
      "step": 5884
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4539572219982062,
      "learning_rate": 6.499845916587379e-06,
      "loss": 0.2198,
      "step": 5885
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.479323759121893,
      "learning_rate": 6.4987408177780644e-06,
      "loss": 0.2044,
      "step": 5886
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4279422505801262,
      "learning_rate": 6.497635638522295e-06,
      "loss": 0.211,
      "step": 5887
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2626392367876296,
      "learning_rate": 6.496530378879396e-06,
      "loss": 0.1684,
      "step": 5888
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.426438435200346,
      "learning_rate": 6.495425038908691e-06,
      "loss": 0.5226,
      "step": 5889
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3809045050154767,
      "learning_rate": 6.494319618669511e-06,
      "loss": 0.2114,
      "step": 5890
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1858061325324465,
      "learning_rate": 6.49321411822119e-06,
      "loss": 0.1758,
      "step": 5891
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6024961671804752,
      "learning_rate": 6.492108537623067e-06,
      "loss": 0.2311,
      "step": 5892
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1899911538664314,
      "learning_rate": 6.491002876934487e-06,
      "loss": 0.1906,
      "step": 5893
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3188653112616997,
      "learning_rate": 6.489897136214796e-06,
      "loss": 0.2242,
      "step": 5894
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3973454325963925,
      "learning_rate": 6.488791315523343e-06,
      "loss": 0.2072,
      "step": 5895
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4019710859358498,
      "learning_rate": 6.487685414919489e-06,
      "loss": 0.2449,
      "step": 5896
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4006508241178328,
      "learning_rate": 6.486579434462592e-06,
      "loss": 0.2021,
      "step": 5897
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3857691016848122,
      "learning_rate": 6.485473374212017e-06,
      "loss": 0.2022,
      "step": 5898
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2222022273123703,
      "learning_rate": 6.4843672342271315e-06,
      "loss": 0.2194,
      "step": 5899
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2047049380664667,
      "learning_rate": 6.483261014567311e-06,
      "loss": 0.1688,
      "step": 5900
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.227358422999814,
      "learning_rate": 6.482154715291933e-06,
      "loss": 0.2142,
      "step": 5901
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3329649469811904,
      "learning_rate": 6.481048336460377e-06,
      "loss": 0.2104,
      "step": 5902
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3695443403163936,
      "learning_rate": 6.47994187813203e-06,
      "loss": 0.1804,
      "step": 5903
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3203471656623678,
      "learning_rate": 6.478835340366283e-06,
      "loss": 0.1864,
      "step": 5904
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2227725143288974,
      "learning_rate": 6.477728723222528e-06,
      "loss": 0.1873,
      "step": 5905
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4888314018696718,
      "learning_rate": 6.476622026760169e-06,
      "loss": 0.202,
      "step": 5906
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3135846646724263,
      "learning_rate": 6.475515251038604e-06,
      "loss": 0.1802,
      "step": 5907
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4025241675095252,
      "learning_rate": 6.4744083961172435e-06,
      "loss": 0.2086,
      "step": 5908
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.448966500528644,
      "learning_rate": 6.473301462055496e-06,
      "loss": 0.5136,
      "step": 5909
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2523462139898633,
      "learning_rate": 6.4721944489127795e-06,
      "loss": 0.1761,
      "step": 5910
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5759270520669146,
      "learning_rate": 6.4710873567485154e-06,
      "loss": 0.21,
      "step": 5911
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.440067354064942,
      "learning_rate": 6.469980185622126e-06,
      "loss": 0.2248,
      "step": 5912
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.784287688240427,
      "learning_rate": 6.468872935593038e-06,
      "loss": 0.5357,
      "step": 5913
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2234199934316157,
      "learning_rate": 6.467765606720689e-06,
      "loss": 0.2182,
      "step": 5914
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3489742132603246,
      "learning_rate": 6.466658199064512e-06,
      "loss": 0.1849,
      "step": 5915
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2976974628135605,
      "learning_rate": 6.465550712683949e-06,
      "loss": 0.2062,
      "step": 5916
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.173882356417723,
      "learning_rate": 6.464443147638446e-06,
      "loss": 0.4705,
      "step": 5917
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4845354867133451,
      "learning_rate": 6.463335503987453e-06,
      "loss": 0.2432,
      "step": 5918
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.646708847517072,
      "learning_rate": 6.462227781790425e-06,
      "loss": 0.6802,
      "step": 5919
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.457177906182861,
      "learning_rate": 6.4611199811068196e-06,
      "loss": 0.2446,
      "step": 5920
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.428018628917624,
      "learning_rate": 6.460012101996095e-06,
      "loss": 0.1994,
      "step": 5921
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.326422509604255,
      "learning_rate": 6.458904144517723e-06,
      "loss": 0.1991,
      "step": 5922
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5753864955116637,
      "learning_rate": 6.457796108731173e-06,
      "loss": 0.2182,
      "step": 5923
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4946401195482777,
      "learning_rate": 6.456687994695918e-06,
      "loss": 0.2192,
      "step": 5924
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4298647431191143,
      "learning_rate": 6.455579802471439e-06,
      "loss": 0.2256,
      "step": 5925
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.441258178748346,
      "learning_rate": 6.4544715321172195e-06,
      "loss": 0.1882,
      "step": 5926
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.543330786728296,
      "learning_rate": 6.453363183692747e-06,
      "loss": 0.6414,
      "step": 5927
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.410805166736668,
      "learning_rate": 6.4522547572575126e-06,
      "loss": 0.2366,
      "step": 5928
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3467241157606815,
      "learning_rate": 6.45114625287101e-06,
      "loss": 0.197,
      "step": 5929
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5686855818906367,
      "learning_rate": 6.450037670592744e-06,
      "loss": 0.1962,
      "step": 5930
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2514219546097685,
      "learning_rate": 6.448929010482215e-06,
      "loss": 0.1641,
      "step": 5931
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3434494673188755,
      "learning_rate": 6.447820272598935e-06,
      "loss": 0.2298,
      "step": 5932
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.192009584164867,
      "learning_rate": 6.446711457002412e-06,
      "loss": 0.1944,
      "step": 5933
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4215603166585253,
      "learning_rate": 6.445602563752167e-06,
      "loss": 0.2271,
      "step": 5934
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2040112411460901,
      "learning_rate": 6.444493592907719e-06,
      "loss": 0.1744,
      "step": 5935
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.294529589000353,
      "learning_rate": 6.443384544528595e-06,
      "loss": 0.1771,
      "step": 5936
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.542482111671868,
      "learning_rate": 6.442275418674321e-06,
      "loss": 0.2107,
      "step": 5937
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4649221220498023,
      "learning_rate": 6.441166215404432e-06,
      "loss": 0.253,
      "step": 5938
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4157801939047272,
      "learning_rate": 6.440056934778468e-06,
      "loss": 0.2253,
      "step": 5939
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.402721685432459,
      "learning_rate": 6.4389475768559675e-06,
      "loss": 0.2394,
      "step": 5940
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2252310072851156,
      "learning_rate": 6.4378381416964785e-06,
      "loss": 0.1749,
      "step": 5941
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4038690190908032,
      "learning_rate": 6.4367286293595495e-06,
      "loss": 0.2193,
      "step": 5942
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.048032419441468,
      "learning_rate": 6.435619039904737e-06,
      "loss": 0.6367,
      "step": 5943
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3364933807635349,
      "learning_rate": 6.434509373391596e-06,
      "loss": 0.2181,
      "step": 5944
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2656920943798071,
      "learning_rate": 6.433399629879693e-06,
      "loss": 0.2042,
      "step": 5945
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3399756767361302,
      "learning_rate": 6.432289809428589e-06,
      "loss": 0.218,
      "step": 5946
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1010395013712686,
      "learning_rate": 6.4311799120978614e-06,
      "loss": 0.1406,
      "step": 5947
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2604562260316823,
      "learning_rate": 6.430069937947081e-06,
      "loss": 0.1512,
      "step": 5948
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3259867563499246,
      "learning_rate": 6.428959887035828e-06,
      "loss": 0.185,
      "step": 5949
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2893931797495104,
      "learning_rate": 6.427849759423684e-06,
      "loss": 0.2268,
      "step": 5950
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2955594250160696,
      "learning_rate": 6.426739555170239e-06,
      "loss": 0.2291,
      "step": 5951
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2122340386448756,
      "learning_rate": 6.425629274335081e-06,
      "loss": 0.1915,
      "step": 5952
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2990168098314965,
      "learning_rate": 6.4245189169778086e-06,
      "loss": 0.1912,
      "step": 5953
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2669679137435874,
      "learning_rate": 6.4234084831580205e-06,
      "loss": 0.1654,
      "step": 5954
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3821924819216012,
      "learning_rate": 6.4222979729353185e-06,
      "loss": 0.2187,
      "step": 5955
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4880833905000703,
      "learning_rate": 6.421187386369313e-06,
      "loss": 0.3013,
      "step": 5956
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2788119359227417,
      "learning_rate": 6.420076723519615e-06,
      "loss": 0.2234,
      "step": 5957
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4379247842890428,
      "learning_rate": 6.4189659844458385e-06,
      "loss": 0.226,
      "step": 5958
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.135747896149007,
      "learning_rate": 6.4178551692076056e-06,
      "loss": 0.1781,
      "step": 5959
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1950873021654478,
      "learning_rate": 6.416744277864541e-06,
      "loss": 0.1814,
      "step": 5960
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3915629970627863,
      "learning_rate": 6.415633310476271e-06,
      "loss": 0.201,
      "step": 5961
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1916977228600398,
      "learning_rate": 6.41452226710243e-06,
      "loss": 0.1858,
      "step": 5962
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.347095480190013,
      "learning_rate": 6.413411147802652e-06,
      "loss": 0.1762,
      "step": 5963
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3051788588075717,
      "learning_rate": 6.41229995263658e-06,
      "loss": 0.1982,
      "step": 5964
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2144848564630306,
      "learning_rate": 6.411188681663857e-06,
      "loss": 0.2122,
      "step": 5965
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2269403931337466,
      "learning_rate": 6.410077334944132e-06,
      "loss": 0.2016,
      "step": 5966
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2310353015325373,
      "learning_rate": 6.408965912537056e-06,
      "loss": 0.1701,
      "step": 5967
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3501885632216366,
      "learning_rate": 6.407854414502289e-06,
      "loss": 0.1841,
      "step": 5968
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.266179408155581,
      "learning_rate": 6.406742840899489e-06,
      "loss": 0.1945,
      "step": 5969
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3041677477887543,
      "learning_rate": 6.405631191788323e-06,
      "loss": 0.2229,
      "step": 5970
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4609152780374801,
      "learning_rate": 6.404519467228458e-06,
      "loss": 0.2135,
      "step": 5971
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1262734460938701,
      "learning_rate": 6.403407667279569e-06,
      "loss": 0.1441,
      "step": 5972
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3759527012886266,
      "learning_rate": 6.402295792001332e-06,
      "loss": 0.1999,
      "step": 5973
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3399992536264542,
      "learning_rate": 6.401183841453427e-06,
      "loss": 0.2162,
      "step": 5974
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1887821337767206,
      "learning_rate": 6.400071815695541e-06,
      "loss": 0.1759,
      "step": 5975
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1884530497356576,
      "learning_rate": 6.3989597147873606e-06,
      "loss": 0.1474,
      "step": 5976
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2275994458889348,
      "learning_rate": 6.397847538788581e-06,
      "loss": 0.164,
      "step": 5977
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2987352699326156,
      "learning_rate": 6.3967352877589e-06,
      "loss": 0.1868,
      "step": 5978
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.266851966614568,
      "learning_rate": 6.3956229617580165e-06,
      "loss": 0.1767,
      "step": 5979
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2813758808086664,
      "learning_rate": 6.394510560845637e-06,
      "loss": 0.2219,
      "step": 5980
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3552096037884132,
      "learning_rate": 6.39339808508147e-06,
      "loss": 0.177,
      "step": 5981
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4254453694214366,
      "learning_rate": 6.392285534525231e-06,
      "loss": 0.219,
      "step": 5982
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.206071177567004,
      "learning_rate": 6.391172909236636e-06,
      "loss": 0.1727,
      "step": 5983
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3689629623887476,
      "learning_rate": 6.3900602092754034e-06,
      "loss": 0.1949,
      "step": 5984
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3283682590450712,
      "learning_rate": 6.388947434701265e-06,
      "loss": 0.223,
      "step": 5985
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4366286017005132,
      "learning_rate": 6.387834585573944e-06,
      "loss": 0.2099,
      "step": 5986
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.364550249291559,
      "learning_rate": 6.386721661953177e-06,
      "loss": 0.2109,
      "step": 5987
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.300147407512584,
      "learning_rate": 6.3856086638986995e-06,
      "loss": 0.2132,
      "step": 5988
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4007614831246213,
      "learning_rate": 6.3844955914702534e-06,
      "loss": 0.2088,
      "step": 5989
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6843347474779486,
      "learning_rate": 6.383382444727585e-06,
      "loss": 0.1897,
      "step": 5990
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.52461273967348,
      "learning_rate": 6.382269223730443e-06,
      "loss": 0.5295,
      "step": 5991
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3506725592827098,
      "learning_rate": 6.381155928538579e-06,
      "loss": 0.1582,
      "step": 5992
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3187605434614629,
      "learning_rate": 6.380042559211753e-06,
      "loss": 0.1788,
      "step": 5993
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4852857438599727,
      "learning_rate": 6.378929115809725e-06,
      "loss": 0.1937,
      "step": 5994
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3559121462875783,
      "learning_rate": 6.37781559839226e-06,
      "loss": 0.1909,
      "step": 5995
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3457770800742648,
      "learning_rate": 6.376702007019127e-06,
      "loss": 0.1756,
      "step": 5996
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.54786787786061,
      "learning_rate": 6.3755883417501006e-06,
      "loss": 0.4815,
      "step": 5997
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1066497192571227,
      "learning_rate": 6.374474602644958e-06,
      "loss": 0.1448,
      "step": 5998
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2577465797862053,
      "learning_rate": 6.373360789763477e-06,
      "loss": 0.189,
      "step": 5999
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2788970309672827,
      "learning_rate": 6.372246903165445e-06,
      "loss": 0.2042,
      "step": 6000
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3972279532730183,
      "learning_rate": 6.371132942910652e-06,
      "loss": 0.2302,
      "step": 6001
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.503823317905543,
      "learning_rate": 6.370018909058889e-06,
      "loss": 0.2363,
      "step": 6002
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.370060516102682,
      "learning_rate": 6.368904801669955e-06,
      "loss": 0.1895,
      "step": 6003
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7386443131796494,
      "learning_rate": 6.3677906208036485e-06,
      "loss": 0.2125,
      "step": 6004
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.039571077501851,
      "learning_rate": 6.366676366519775e-06,
      "loss": 0.6157,
      "step": 6005
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.665452647494954,
      "learning_rate": 6.365562038878145e-06,
      "loss": 0.5618,
      "step": 6006
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5390271453761908,
      "learning_rate": 6.364447637938569e-06,
      "loss": 0.2201,
      "step": 6007
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3523826531813135,
      "learning_rate": 6.363333163760865e-06,
      "loss": 0.2091,
      "step": 6008
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4781381140183742,
      "learning_rate": 6.362218616404854e-06,
      "loss": 0.2064,
      "step": 6009
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1729211448378916,
      "learning_rate": 6.361103995930357e-06,
      "loss": 0.1736,
      "step": 6010
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.8514395666538705,
      "learning_rate": 6.3599893023972056e-06,
      "loss": 0.5345,
      "step": 6011
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2529744461132255,
      "learning_rate": 6.358874535865233e-06,
      "loss": 0.2249,
      "step": 6012
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2752063477226305,
      "learning_rate": 6.3577596963942725e-06,
      "loss": 0.182,
      "step": 6013
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3158678118337033,
      "learning_rate": 6.356644784044164e-06,
      "loss": 0.16,
      "step": 6014
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3945931081818745,
      "learning_rate": 6.355529798874757e-06,
      "loss": 0.1989,
      "step": 6015
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3473817810635123,
      "learning_rate": 6.354414740945894e-06,
      "loss": 0.2294,
      "step": 6016
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2162395444267329,
      "learning_rate": 6.353299610317427e-06,
      "loss": 0.1779,
      "step": 6017
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4712114430649834,
      "learning_rate": 6.352184407049214e-06,
      "loss": 0.2113,
      "step": 6018
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4251986494226967,
      "learning_rate": 6.351069131201115e-06,
      "loss": 0.2152,
      "step": 6019
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3055438377445368,
      "learning_rate": 6.349953782832991e-06,
      "loss": 0.1581,
      "step": 6020
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1280743247758633,
      "learning_rate": 6.348838362004713e-06,
      "loss": 0.1541,
      "step": 6021
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3732847305084301,
      "learning_rate": 6.347722868776148e-06,
      "loss": 0.1814,
      "step": 6022
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3943803096345815,
      "learning_rate": 6.346607303207176e-06,
      "loss": 0.2308,
      "step": 6023
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2452937287808699,
      "learning_rate": 6.345491665357673e-06,
      "loss": 0.1704,
      "step": 6024
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3581319366717806,
      "learning_rate": 6.344375955287523e-06,
      "loss": 0.2247,
      "step": 6025
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.491029662219368,
      "learning_rate": 6.343260173056612e-06,
      "loss": 0.2472,
      "step": 6026
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.309496566040985,
      "learning_rate": 6.34214431872483e-06,
      "loss": 0.2111,
      "step": 6027
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3164834825764118,
      "learning_rate": 6.341028392352076e-06,
      "loss": 0.1957,
      "step": 6028
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.8440888161367095,
      "learning_rate": 6.339912393998244e-06,
      "loss": 0.6598,
      "step": 6029
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.272016747790002,
      "learning_rate": 6.338796323723236e-06,
      "loss": 0.1969,
      "step": 6030
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3960289170959554,
      "learning_rate": 6.337680181586963e-06,
      "loss": 0.2064,
      "step": 6031
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4520561618171652,
      "learning_rate": 6.336563967649332e-06,
      "loss": 0.237,
      "step": 6032
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.35469442577264,
      "learning_rate": 6.335447681970255e-06,
      "loss": 0.6443,
      "step": 6033
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4395958848140078,
      "learning_rate": 6.334331324609654e-06,
      "loss": 0.2034,
      "step": 6034
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.367556922101954,
      "learning_rate": 6.3332148956274466e-06,
      "loss": 0.1539,
      "step": 6035
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.206265415641196,
      "learning_rate": 6.332098395083562e-06,
      "loss": 0.1566,
      "step": 6036
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1895328825014628,
      "learning_rate": 6.330981823037926e-06,
      "loss": 0.1875,
      "step": 6037
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3440759949805046,
      "learning_rate": 6.329865179550475e-06,
      "loss": 0.1783,
      "step": 6038
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.376585951856802,
      "learning_rate": 6.328748464681143e-06,
      "loss": 0.1949,
      "step": 6039
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1629048279951808,
      "learning_rate": 6.327631678489874e-06,
      "loss": 0.153,
      "step": 6040
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3623472418213103,
      "learning_rate": 6.32651482103661e-06,
      "loss": 0.2241,
      "step": 6041
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2903805178978347,
      "learning_rate": 6.325397892381301e-06,
      "loss": 0.1918,
      "step": 6042
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3277787323291077,
      "learning_rate": 6.324280892583896e-06,
      "loss": 0.1947,
      "step": 6043
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.188570791239475,
      "learning_rate": 6.323163821704357e-06,
      "loss": 0.1688,
      "step": 6044
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3774756657643852,
      "learning_rate": 6.322046679802639e-06,
      "loss": 0.2389,
      "step": 6045
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5206647968224447,
      "learning_rate": 6.320929466938709e-06,
      "loss": 0.2466,
      "step": 6046
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5675209041491782,
      "learning_rate": 6.319812183172531e-06,
      "loss": 0.2493,
      "step": 6047
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5112712767834573,
      "learning_rate": 6.3186948285640795e-06,
      "loss": 0.1908,
      "step": 6048
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.346116175191913,
      "learning_rate": 6.317577403173328e-06,
      "loss": 0.1958,
      "step": 6049
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.479828707900571,
      "learning_rate": 6.316459907060257e-06,
      "loss": 0.7712,
      "step": 6050
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3213781213243325,
      "learning_rate": 6.315342340284847e-06,
      "loss": 0.2467,
      "step": 6051
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5808386537163082,
      "learning_rate": 6.3142247029070855e-06,
      "loss": 0.1988,
      "step": 6052
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3449799290707891,
      "learning_rate": 6.313106994986965e-06,
      "loss": 0.174,
      "step": 6053
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5131614861257754,
      "learning_rate": 6.311989216584476e-06,
      "loss": 0.2302,
      "step": 6054
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.145758856149474,
      "learning_rate": 6.310871367759619e-06,
      "loss": 0.1351,
      "step": 6055
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3241803729449049,
      "learning_rate": 6.309753448572394e-06,
      "loss": 0.2086,
      "step": 6056
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3726814861839929,
      "learning_rate": 6.308635459082808e-06,
      "loss": 0.2234,
      "step": 6057
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3933392060361727,
      "learning_rate": 6.3075173993508684e-06,
      "loss": 0.2267,
      "step": 6058
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2764532389853296,
      "learning_rate": 6.306399269436591e-06,
      "loss": 0.2342,
      "step": 6059
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.168912319889708,
      "learning_rate": 6.305281069399989e-06,
      "loss": 0.8818,
      "step": 6060
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3707749990702702,
      "learning_rate": 6.3041627993010855e-06,
      "loss": 0.2187,
      "step": 6061
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4589728663563595,
      "learning_rate": 6.303044459199905e-06,
      "loss": 0.27,
      "step": 6062
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4452196504646069,
      "learning_rate": 6.301926049156475e-06,
      "loss": 0.2253,
      "step": 6063
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4641511553846807,
      "learning_rate": 6.300807569230825e-06,
      "loss": 0.1943,
      "step": 6064
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.137920974039288,
      "learning_rate": 6.299689019482993e-06,
      "loss": 0.5305,
      "step": 6065
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.217361603368205,
      "learning_rate": 6.298570399973017e-06,
      "loss": 0.4738,
      "step": 6066
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.758242731517963,
      "learning_rate": 6.297451710760944e-06,
      "loss": 0.2013,
      "step": 6067
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.158738996363695,
      "learning_rate": 6.296332951906815e-06,
      "loss": 0.7065,
      "step": 6068
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4595291321970978,
      "learning_rate": 6.2952141234706835e-06,
      "loss": 0.2189,
      "step": 6069
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2934768057145665,
      "learning_rate": 6.294095225512604e-06,
      "loss": 0.2,
      "step": 6070
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4605990474584731,
      "learning_rate": 6.292976258092635e-06,
      "loss": 0.1858,
      "step": 6071
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4418245200721458,
      "learning_rate": 6.291857221270836e-06,
      "loss": 0.2087,
      "step": 6072
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3168402606578098,
      "learning_rate": 6.290738115107273e-06,
      "loss": 0.2103,
      "step": 6073
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.390301597819154,
      "learning_rate": 6.2896189396620165e-06,
      "loss": 0.2359,
      "step": 6074
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.658191379068836,
      "learning_rate": 6.288499694995139e-06,
      "loss": 0.2555,
      "step": 6075
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4754928777384115,
      "learning_rate": 6.2873803811667164e-06,
      "loss": 0.1822,
      "step": 6076
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.480901366231782,
      "learning_rate": 6.286260998236829e-06,
      "loss": 0.2364,
      "step": 6077
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3909388935933629,
      "learning_rate": 6.2851415462655625e-06,
      "loss": 0.2456,
      "step": 6078
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5305797781282102,
      "learning_rate": 6.284022025313003e-06,
      "loss": 0.2374,
      "step": 6079
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.457662784771119,
      "learning_rate": 6.282902435439242e-06,
      "loss": 0.2442,
      "step": 6080
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3925048699359455,
      "learning_rate": 6.281782776704375e-06,
      "loss": 0.2241,
      "step": 6081
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.7416384937177707,
      "learning_rate": 6.2806630491685e-06,
      "loss": 0.7541,
      "step": 6082
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.408815079810787,
      "learning_rate": 6.279543252891722e-06,
      "loss": 0.2092,
      "step": 6083
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.396629402432976,
      "learning_rate": 6.278423387934145e-06,
      "loss": 0.2411,
      "step": 6084
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2451831326037093,
      "learning_rate": 6.277303454355878e-06,
      "loss": 0.1866,
      "step": 6085
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0819483604290343,
      "learning_rate": 6.276183452217038e-06,
      "loss": 0.1477,
      "step": 6086
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5559081183368948,
      "learning_rate": 6.2750633815777395e-06,
      "loss": 0.2522,
      "step": 6087
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4025796562908368,
      "learning_rate": 6.273943242498105e-06,
      "loss": 0.2384,
      "step": 6088
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3885648019350323,
      "learning_rate": 6.272823035038257e-06,
      "loss": 0.2022,
      "step": 6089
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4706824483661423,
      "learning_rate": 6.2717027592583255e-06,
      "loss": 0.2,
      "step": 6090
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4267343312064267,
      "learning_rate": 6.270582415218442e-06,
      "loss": 0.1925,
      "step": 6091
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.172331016913087,
      "learning_rate": 6.269462002978743e-06,
      "loss": 0.1802,
      "step": 6092
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.261669885250662,
      "learning_rate": 6.268341522599366e-06,
      "loss": 0.1751,
      "step": 6093
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1600762773277573,
      "learning_rate": 6.267220974140456e-06,
      "loss": 0.1419,
      "step": 6094
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4545003234020253,
      "learning_rate": 6.266100357662158e-06,
      "loss": 0.2117,
      "step": 6095
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.126817450825488,
      "learning_rate": 6.2649796732246226e-06,
      "loss": 0.6336,
      "step": 6096
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.981721024362856,
      "learning_rate": 6.263858920888003e-06,
      "loss": 0.6032,
      "step": 6097
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3231552146704568,
      "learning_rate": 6.262738100712458e-06,
      "loss": 0.1853,
      "step": 6098
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4078309506930746,
      "learning_rate": 6.261617212758148e-06,
      "loss": 0.1914,
      "step": 6099
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4980321479181489,
      "learning_rate": 6.26049625708524e-06,
      "loss": 0.2265,
      "step": 6100
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3244830573180066,
      "learning_rate": 6.259375233753901e-06,
      "loss": 0.2024,
      "step": 6101
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.679345222315573,
      "learning_rate": 6.258254142824299e-06,
      "loss": 0.2545,
      "step": 6102
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3277166982723225,
      "learning_rate": 6.257132984356616e-06,
      "loss": 0.1807,
      "step": 6103
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1656771841958482,
      "learning_rate": 6.256011758411029e-06,
      "loss": 0.1536,
      "step": 6104
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4484315046107583,
      "learning_rate": 6.254890465047721e-06,
      "loss": 0.2322,
      "step": 6105
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.333120438361033,
      "learning_rate": 6.253769104326877e-06,
      "loss": 0.1772,
      "step": 6106
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3073745304785707,
      "learning_rate": 6.252647676308687e-06,
      "loss": 0.1582,
      "step": 6107
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4194973406503713,
      "learning_rate": 6.251526181053349e-06,
      "loss": 0.2508,
      "step": 6108
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.40369981239689,
      "learning_rate": 6.250404618621058e-06,
      "loss": 0.2187,
      "step": 6109
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.359033950143654,
      "learning_rate": 6.249282989072013e-06,
      "loss": 0.1966,
      "step": 6110
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3879305257460344,
      "learning_rate": 6.248161292466419e-06,
      "loss": 0.2077,
      "step": 6111
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2953763457714846,
      "learning_rate": 6.247039528864488e-06,
      "loss": 0.1887,
      "step": 6112
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.301919732019173,
      "learning_rate": 6.245917698326429e-06,
      "loss": 0.1953,
      "step": 6113
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2076293467702066,
      "learning_rate": 6.244795800912458e-06,
      "loss": 0.189,
      "step": 6114
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.350264703496325,
      "learning_rate": 6.243673836682791e-06,
      "loss": 0.5526,
      "step": 6115
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2103954099887813,
      "learning_rate": 6.242551805697655e-06,
      "loss": 0.1607,
      "step": 6116
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.329998748756925,
      "learning_rate": 6.241429708017276e-06,
      "loss": 0.1592,
      "step": 6117
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.425673407956938,
      "learning_rate": 6.2403075437018804e-06,
      "loss": 0.2315,
      "step": 6118
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2871733217204477,
      "learning_rate": 6.239185312811703e-06,
      "loss": 0.1977,
      "step": 6119
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3576584400873155,
      "learning_rate": 6.238063015406982e-06,
      "loss": 0.2047,
      "step": 6120
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.646365886454608,
      "learning_rate": 6.236940651547957e-06,
      "loss": 0.4971,
      "step": 6121
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.673272147120256,
      "learning_rate": 6.235818221294871e-06,
      "loss": 0.6871,
      "step": 6122
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3710451409448157,
      "learning_rate": 6.234695724707973e-06,
      "loss": 0.215,
      "step": 6123
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.24275454631224,
      "learning_rate": 6.233573161847515e-06,
      "loss": 0.6136,
      "step": 6124
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3173447842887431,
      "learning_rate": 6.232450532773749e-06,
      "loss": 0.2116,
      "step": 6125
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.223449819969378,
      "learning_rate": 6.231327837546935e-06,
      "loss": 0.1865,
      "step": 6126
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.206800843035456,
      "learning_rate": 6.230205076227334e-06,
      "loss": 0.572,
      "step": 6127
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.546131691032404,
      "learning_rate": 6.229082248875212e-06,
      "loss": 0.2109,
      "step": 6128
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2287647396310375,
      "learning_rate": 6.227959355550838e-06,
      "loss": 0.1658,
      "step": 6129
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3238274378728416,
      "learning_rate": 6.226836396314484e-06,
      "loss": 0.1658,
      "step": 6130
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8058051582682093,
      "learning_rate": 6.225713371226427e-06,
      "loss": 0.254,
      "step": 6131
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5235175658254225,
      "learning_rate": 6.224590280346944e-06,
      "loss": 0.2128,
      "step": 6132
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5450292786024251,
      "learning_rate": 6.223467123736321e-06,
      "loss": 0.2081,
      "step": 6133
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4204993746034242,
      "learning_rate": 6.222343901454844e-06,
      "loss": 0.2162,
      "step": 6134
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2087701624518392,
      "learning_rate": 6.221220613562802e-06,
      "loss": 0.1741,
      "step": 6135
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3617109027969951,
      "learning_rate": 6.220097260120488e-06,
      "loss": 0.1869,
      "step": 6136
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.289692806054038,
      "learning_rate": 6.218973841188202e-06,
      "loss": 0.1996,
      "step": 6137
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.21172292445162,
      "learning_rate": 6.217850356826242e-06,
      "loss": 0.1946,
      "step": 6138
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4087152832187702,
      "learning_rate": 6.216726807094914e-06,
      "loss": 0.2207,
      "step": 6139
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.362987376426746,
      "learning_rate": 6.215603192054523e-06,
      "loss": 0.1787,
      "step": 6140
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3603964348432152,
      "learning_rate": 6.2144795117653835e-06,
      "loss": 0.2139,
      "step": 6141
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3756707903231142,
      "learning_rate": 6.213355766287808e-06,
      "loss": 0.171,
      "step": 6142
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1973063848281436,
      "learning_rate": 6.212231955682115e-06,
      "loss": 0.1818,
      "step": 6143
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5103461390608002,
      "learning_rate": 6.211108080008627e-06,
      "loss": 0.2559,
      "step": 6144
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.954328025773266,
      "learning_rate": 6.20998413932767e-06,
      "loss": 0.6369,
      "step": 6145
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.765131265325706,
      "learning_rate": 6.2088601336995705e-06,
      "loss": 0.5469,
      "step": 6146
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2826158108561603,
      "learning_rate": 6.207736063184661e-06,
      "loss": 0.1886,
      "step": 6147
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.443263937670081,
      "learning_rate": 6.206611927843277e-06,
      "loss": 0.6505,
      "step": 6148
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3189733136217026,
      "learning_rate": 6.205487727735759e-06,
      "loss": 0.2324,
      "step": 6149
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.241002664653362,
      "learning_rate": 6.20436346292245e-06,
      "loss": 0.1907,
      "step": 6150
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8471270765250267,
      "learning_rate": 6.203239133463695e-06,
      "loss": 0.1875,
      "step": 6151
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4086430969684023,
      "learning_rate": 6.202114739419843e-06,
      "loss": 0.2146,
      "step": 6152
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1660663473024138,
      "learning_rate": 6.200990280851247e-06,
      "loss": 0.1915,
      "step": 6153
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3683259942529886,
      "learning_rate": 6.199865757818264e-06,
      "loss": 0.203,
      "step": 6154
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1616775809018711,
      "learning_rate": 6.198741170381255e-06,
      "loss": 0.1775,
      "step": 6155
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4802348407405437,
      "learning_rate": 6.1976165186005825e-06,
      "loss": 0.1963,
      "step": 6156
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.405109664397268,
      "learning_rate": 6.19649180253661e-06,
      "loss": 0.199,
      "step": 6157
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1960728828482798,
      "learning_rate": 6.195367022249714e-06,
      "loss": 0.1956,
      "step": 6158
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.424632469178128,
      "learning_rate": 6.194242177800265e-06,
      "loss": 0.2227,
      "step": 6159
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3099422770056406,
      "learning_rate": 6.1931172692486405e-06,
      "loss": 0.1807,
      "step": 6160
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3783965786067665,
      "learning_rate": 6.191992296655219e-06,
      "loss": 0.1773,
      "step": 6161
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4164351520790772,
      "learning_rate": 6.190867260080387e-06,
      "loss": 0.2208,
      "step": 6162
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2331184457132873,
      "learning_rate": 6.189742159584533e-06,
      "loss": 0.1925,
      "step": 6163
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2115048258332202,
      "learning_rate": 6.188616995228046e-06,
      "loss": 0.2144,
      "step": 6164
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2972706704868169,
      "learning_rate": 6.187491767071321e-06,
      "loss": 0.2259,
      "step": 6165
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.403564026534955,
      "learning_rate": 6.186366475174754e-06,
      "loss": 0.2168,
      "step": 6166
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.36688384725241,
      "learning_rate": 6.185241119598749e-06,
      "loss": 0.1912,
      "step": 6167
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5636020618240198,
      "learning_rate": 6.18411570040371e-06,
      "loss": 0.2447,
      "step": 6168
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4533881277320841,
      "learning_rate": 6.182990217650045e-06,
      "loss": 0.2063,
      "step": 6169
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4702121720775534,
      "learning_rate": 6.1818646713981626e-06,
      "loss": 0.1927,
      "step": 6170
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4412977528362718,
      "learning_rate": 6.180739061708481e-06,
      "loss": 0.2275,
      "step": 6171
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3760067997333865,
      "learning_rate": 6.179613388641419e-06,
      "loss": 0.2098,
      "step": 6172
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3991209815623915,
      "learning_rate": 6.178487652257395e-06,
      "loss": 0.2252,
      "step": 6173
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.851941936343868,
      "learning_rate": 6.177361852616834e-06,
      "loss": 0.5485,
      "step": 6174
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.036157039773243,
      "learning_rate": 6.176235989780169e-06,
      "loss": 0.5849,
      "step": 6175
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3042836319967661,
      "learning_rate": 6.175110063807829e-06,
      "loss": 0.2073,
      "step": 6176
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2279721196528364,
      "learning_rate": 6.173984074760249e-06,
      "loss": 0.2118,
      "step": 6177
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4152578089982346,
      "learning_rate": 6.1728580226978656e-06,
      "loss": 0.1845,
      "step": 6178
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2402171604683756,
      "learning_rate": 6.171731907681124e-06,
      "loss": 0.1738,
      "step": 6179
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.358663751974876,
      "learning_rate": 6.17060572977047e-06,
      "loss": 0.2181,
      "step": 6180
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2218877796487795,
      "learning_rate": 6.169479489026351e-06,
      "loss": 0.1481,
      "step": 6181
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3014885595741474,
      "learning_rate": 6.168353185509217e-06,
      "loss": 0.199,
      "step": 6182
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3369567669595106,
      "learning_rate": 6.1672268192795285e-06,
      "loss": 0.1955,
      "step": 6183
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2670428442030366,
      "learning_rate": 6.1661003903977405e-06,
      "loss": 0.1896,
      "step": 6184
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4355153720215206,
      "learning_rate": 6.164973898924315e-06,
      "loss": 0.1974,
      "step": 6185
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.324647936278329,
      "learning_rate": 6.1638473449197185e-06,
      "loss": 0.2015,
      "step": 6186
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.20759406732928,
      "learning_rate": 6.162720728444422e-06,
      "loss": 0.1575,
      "step": 6187
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2511800565545457,
      "learning_rate": 6.1615940495588955e-06,
      "loss": 0.2126,
      "step": 6188
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5014285751933938,
      "learning_rate": 6.160467308323616e-06,
      "loss": 0.2076,
      "step": 6189
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.340871336596143,
      "learning_rate": 6.159340504799059e-06,
      "loss": 0.2168,
      "step": 6190
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2616066255345741,
      "learning_rate": 6.158213639045712e-06,
      "loss": 0.2553,
      "step": 6191
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3510305597489163,
      "learning_rate": 6.157086711124057e-06,
      "loss": 0.1799,
      "step": 6192
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4910693978490932,
      "learning_rate": 6.155959721094585e-06,
      "loss": 0.1985,
      "step": 6193
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.379994434569168,
      "learning_rate": 6.154832669017787e-06,
      "loss": 0.245,
      "step": 6194
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4038194192086932,
      "learning_rate": 6.153705554954158e-06,
      "loss": 0.2156,
      "step": 6195
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3806045759435384,
      "learning_rate": 6.1525783789642e-06,
      "loss": 0.2084,
      "step": 6196
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3250138340182083,
      "learning_rate": 6.1514511411084134e-06,
      "loss": 0.1962,
      "step": 6197
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4094631144210183,
      "learning_rate": 6.150323841447304e-06,
      "loss": 0.2415,
      "step": 6198
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.421841611223444,
      "learning_rate": 6.1491964800413795e-06,
      "loss": 0.2101,
      "step": 6199
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3832341057798752,
      "learning_rate": 6.1480690569511545e-06,
      "loss": 0.2275,
      "step": 6200
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.392889850897278,
      "learning_rate": 6.146941572237144e-06,
      "loss": 0.2147,
      "step": 6201
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.217405760422023,
      "learning_rate": 6.145814025959866e-06,
      "loss": 0.1822,
      "step": 6202
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5699487203119262,
      "learning_rate": 6.144686418179844e-06,
      "loss": 0.2452,
      "step": 6203
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.366077704621606,
      "learning_rate": 6.143558748957601e-06,
      "loss": 0.2108,
      "step": 6204
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4732859544911876,
      "learning_rate": 6.14243101835367e-06,
      "loss": 0.2477,
      "step": 6205
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3315202482635133,
      "learning_rate": 6.141303226428579e-06,
      "loss": 0.191,
      "step": 6206
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2482294060093437,
      "learning_rate": 6.140175373242865e-06,
      "loss": 0.1996,
      "step": 6207
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3569112411065822,
      "learning_rate": 6.139047458857066e-06,
      "loss": 0.1976,
      "step": 6208
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4052322355071327,
      "learning_rate": 6.137919483331725e-06,
      "loss": 0.1721,
      "step": 6209
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.263342237411041,
      "learning_rate": 6.136791446727387e-06,
      "loss": 0.1834,
      "step": 6210
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.311912214696059,
      "learning_rate": 6.135663349104601e-06,
      "loss": 0.1904,
      "step": 6211
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2401868278387478,
      "learning_rate": 6.134535190523917e-06,
      "loss": 0.1895,
      "step": 6212
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.32775518689021,
      "learning_rate": 6.133406971045892e-06,
      "loss": 0.5603,
      "step": 6213
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.382364531410594,
      "learning_rate": 6.132278690731084e-06,
      "loss": 0.2349,
      "step": 6214
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0386007285000776,
      "learning_rate": 6.131150349640053e-06,
      "loss": 0.1463,
      "step": 6215
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3906146274093631,
      "learning_rate": 6.130021947833364e-06,
      "loss": 0.1851,
      "step": 6216
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.630223903710757,
      "learning_rate": 6.128893485371589e-06,
      "loss": 0.2271,
      "step": 6217
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2084793956096727,
      "learning_rate": 6.127764962315296e-06,
      "loss": 0.1484,
      "step": 6218
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4312937297665498,
      "learning_rate": 6.12663637872506e-06,
      "loss": 0.2123,
      "step": 6219
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3638214268796884,
      "learning_rate": 6.125507734661458e-06,
      "loss": 0.2116,
      "step": 6220
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2938095651235335,
      "learning_rate": 6.124379030185072e-06,
      "loss": 0.2341,
      "step": 6221
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.885239796801945,
      "learning_rate": 6.123250265356489e-06,
      "loss": 0.7492,
      "step": 6222
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1906012301671285,
      "learning_rate": 6.122121440236292e-06,
      "loss": 0.1739,
      "step": 6223
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2731473919044556,
      "learning_rate": 6.1209925548850734e-06,
      "loss": 0.1658,
      "step": 6224
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2877739378374993,
      "learning_rate": 6.119863609363429e-06,
      "loss": 0.1921,
      "step": 6225
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3621938534785047,
      "learning_rate": 6.118734603731955e-06,
      "loss": 0.1817,
      "step": 6226
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3648345183442678,
      "learning_rate": 6.117605538051252e-06,
      "loss": 0.2033,
      "step": 6227
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.138755521519348,
      "learning_rate": 6.116476412381926e-06,
      "loss": 0.1467,
      "step": 6228
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1088150409775865,
      "learning_rate": 6.115347226784578e-06,
      "loss": 0.1651,
      "step": 6229
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.330296853056125,
      "learning_rate": 6.114217981319823e-06,
      "loss": 0.6135,
      "step": 6230
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2721725444224916,
      "learning_rate": 6.113088676048274e-06,
      "loss": 0.1946,
      "step": 6231
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5357647293942507,
      "learning_rate": 6.111959311030548e-06,
      "loss": 0.2226,
      "step": 6232
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3967966416842132,
      "learning_rate": 6.1108298863272616e-06,
      "loss": 0.2105,
      "step": 6233
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.076899093344789,
      "learning_rate": 6.109700401999042e-06,
      "loss": 0.55,
      "step": 6234
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2311866535673874,
      "learning_rate": 6.108570858106512e-06,
      "loss": 0.1161,
      "step": 6235
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3166317583633051,
      "learning_rate": 6.1074412547103025e-06,
      "loss": 0.1784,
      "step": 6236
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2704777627249038,
      "learning_rate": 6.106311591871046e-06,
      "loss": 0.1776,
      "step": 6237
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.19983985382999,
      "learning_rate": 6.105181869649379e-06,
      "loss": 0.2002,
      "step": 6238
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4677889375537017,
      "learning_rate": 6.10405208810594e-06,
      "loss": 0.1827,
      "step": 6239
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.390272710189704,
      "learning_rate": 6.1029222473013705e-06,
      "loss": 0.5498,
      "step": 6240
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.293239891737135,
      "learning_rate": 6.101792347296316e-06,
      "loss": 0.2385,
      "step": 6241
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.443916025669607,
      "learning_rate": 6.100662388151427e-06,
      "loss": 0.7363,
      "step": 6242
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5122152655578487,
      "learning_rate": 6.099532369927353e-06,
      "loss": 0.2285,
      "step": 6243
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.336912865585975,
      "learning_rate": 6.09840229268475e-06,
      "loss": 0.1723,
      "step": 6244
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2734663966435864,
      "learning_rate": 6.097272156484274e-06,
      "loss": 0.1879,
      "step": 6245
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.448569829089818,
      "learning_rate": 6.096141961386589e-06,
      "loss": 0.212,
      "step": 6246
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.614573599409491,
      "learning_rate": 6.095011707452358e-06,
      "loss": 0.6934,
      "step": 6247
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4237615740702574,
      "learning_rate": 6.09388139474225e-06,
      "loss": 0.1955,
      "step": 6248
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2836320754134385,
      "learning_rate": 6.092751023316933e-06,
      "loss": 0.1798,
      "step": 6249
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.243663407298675,
      "learning_rate": 6.091620593237083e-06,
      "loss": 0.2014,
      "step": 6250
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2637810131072553,
      "learning_rate": 6.090490104563377e-06,
      "loss": 0.1807,
      "step": 6251
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2507542335233655,
      "learning_rate": 6.0893595573564935e-06,
      "loss": 0.186,
      "step": 6252
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2434477247734192,
      "learning_rate": 6.088228951677117e-06,
      "loss": 0.1597,
      "step": 6253
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3929016069730167,
      "learning_rate": 6.087098287585933e-06,
      "loss": 0.2329,
      "step": 6254
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.574675666920025,
      "learning_rate": 6.085967565143632e-06,
      "loss": 0.2048,
      "step": 6255
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.392765196853635,
      "learning_rate": 6.084836784410908e-06,
      "loss": 0.1806,
      "step": 6256
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.2100092271209295,
      "learning_rate": 6.083705945448453e-06,
      "loss": 0.5043,
      "step": 6257
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3240589149003132,
      "learning_rate": 6.08257504831697e-06,
      "loss": 0.179,
      "step": 6258
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4255817514024363,
      "learning_rate": 6.081444093077157e-06,
      "loss": 0.1835,
      "step": 6259
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.374730721722942,
      "learning_rate": 6.080313079789723e-06,
      "loss": 0.194,
      "step": 6260
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4282025574862633,
      "learning_rate": 6.079182008515376e-06,
      "loss": 0.2235,
      "step": 6261
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5389624388632812,
      "learning_rate": 6.078050879314824e-06,
      "loss": 0.2101,
      "step": 6262
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.192906115255055,
      "learning_rate": 6.076919692248784e-06,
      "loss": 0.6677,
      "step": 6263
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2621826272550847,
      "learning_rate": 6.075788447377974e-06,
      "loss": 0.1909,
      "step": 6264
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3613137196387608,
      "learning_rate": 6.074657144763114e-06,
      "loss": 0.1984,
      "step": 6265
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5738135038837997,
      "learning_rate": 6.073525784464928e-06,
      "loss": 0.2065,
      "step": 6266
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5311593565023167,
      "learning_rate": 6.072394366544141e-06,
      "loss": 0.2259,
      "step": 6267
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2303994063072423,
      "learning_rate": 6.071262891061487e-06,
      "loss": 0.2171,
      "step": 6268
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.149972000965288,
      "learning_rate": 6.070131358077695e-06,
      "loss": 0.5325,
      "step": 6269
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2961106972910443,
      "learning_rate": 6.068999767653505e-06,
      "loss": 0.1794,
      "step": 6270
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.5074673461155017,
      "learning_rate": 6.067868119849654e-06,
      "loss": 0.5808,
      "step": 6271
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5227394000854955,
      "learning_rate": 6.066736414726884e-06,
      "loss": 0.1943,
      "step": 6272
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4007077133053332,
      "learning_rate": 6.065604652345942e-06,
      "loss": 0.2008,
      "step": 6273
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3788116505384536,
      "learning_rate": 6.064472832767575e-06,
      "loss": 0.215,
      "step": 6274
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2167009064908902,
      "learning_rate": 6.063340956052535e-06,
      "loss": 0.1746,
      "step": 6275
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3587374269621821,
      "learning_rate": 6.062209022261577e-06,
      "loss": 0.1911,
      "step": 6276
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.302926753246791,
      "learning_rate": 6.06107703145546e-06,
      "loss": 0.2233,
      "step": 6277
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3926403236691507,
      "learning_rate": 6.059944983694942e-06,
      "loss": 0.2215,
      "step": 6278
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4597420568222197,
      "learning_rate": 6.058812879040788e-06,
      "loss": 0.2114,
      "step": 6279
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.370211512995902,
      "learning_rate": 6.0576807175537654e-06,
      "loss": 0.2237,
      "step": 6280
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.500930703748208,
      "learning_rate": 6.056548499294646e-06,
      "loss": 0.4779,
      "step": 6281
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4633484928762333,
      "learning_rate": 6.055416224324198e-06,
      "loss": 0.2114,
      "step": 6282
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4013170605867908,
      "learning_rate": 6.0542838927032e-06,
      "loss": 0.2321,
      "step": 6283
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3518412157419437,
      "learning_rate": 6.053151504492431e-06,
      "loss": 0.1731,
      "step": 6284
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2443941602769666,
      "learning_rate": 6.052019059752675e-06,
      "loss": 0.1609,
      "step": 6285
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2183093941072516,
      "learning_rate": 6.0508865585447156e-06,
      "loss": 0.1587,
      "step": 6286
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2435163329780377,
      "learning_rate": 6.0497540009293375e-06,
      "loss": 0.1905,
      "step": 6287
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.236734889156336,
      "learning_rate": 6.048621386967337e-06,
      "loss": 0.2169,
      "step": 6288
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2936016621117368,
      "learning_rate": 6.0474887167195065e-06,
      "loss": 0.1714,
      "step": 6289
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3117060899185808,
      "learning_rate": 6.0463559902466425e-06,
      "loss": 0.1963,
      "step": 6290
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.441185049728224,
      "learning_rate": 6.045223207609547e-06,
      "loss": 0.2242,
      "step": 6291
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4909393207291666,
      "learning_rate": 6.044090368869021e-06,
      "loss": 0.1856,
      "step": 6292
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4453673900207722,
      "learning_rate": 6.042957474085873e-06,
      "loss": 0.2356,
      "step": 6293
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1441891228733112,
      "learning_rate": 6.041824523320911e-06,
      "loss": 0.1875,
      "step": 6294
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.568945226444918,
      "learning_rate": 6.0406915166349475e-06,
      "loss": 0.6374,
      "step": 6295
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5429572550344108,
      "learning_rate": 6.039558454088796e-06,
      "loss": 0.2178,
      "step": 6296
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3936526079781395,
      "learning_rate": 6.038425335743279e-06,
      "loss": 0.1977,
      "step": 6297
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3292338981763416,
      "learning_rate": 6.037292161659215e-06,
      "loss": 0.2133,
      "step": 6298
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4828141189545134,
      "learning_rate": 6.036158931897429e-06,
      "loss": 0.2228,
      "step": 6299
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4020798827656242,
      "learning_rate": 6.035025646518747e-06,
      "loss": 0.2271,
      "step": 6300
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.131310637340404,
      "learning_rate": 6.033892305584001e-06,
      "loss": 0.7952,
      "step": 6301
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3079993540816803,
      "learning_rate": 6.032758909154023e-06,
      "loss": 0.1964,
      "step": 6302
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.237006692797655,
      "learning_rate": 6.031625457289649e-06,
      "loss": 0.1886,
      "step": 6303
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.338943704898319,
      "learning_rate": 6.0304919500517214e-06,
      "loss": 0.223,
      "step": 6304
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3342799236949077,
      "learning_rate": 6.029358387501076e-06,
      "loss": 0.2181,
      "step": 6305
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3623897368278908,
      "learning_rate": 6.0282247696985655e-06,
      "loss": 0.1875,
      "step": 6306
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5052215264138904,
      "learning_rate": 6.027091096705032e-06,
      "loss": 0.2224,
      "step": 6307
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3118463019171482,
      "learning_rate": 6.02595736858133e-06,
      "loss": 0.1778,
      "step": 6308
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2686771197138902,
      "learning_rate": 6.024823585388311e-06,
      "loss": 0.2018,
      "step": 6309
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1575823445280766,
      "learning_rate": 6.023689747186835e-06,
      "loss": 0.1493,
      "step": 6310
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4153949398860801,
      "learning_rate": 6.022555854037759e-06,
      "loss": 0.1896,
      "step": 6311
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.982957771772166,
      "learning_rate": 6.021421906001948e-06,
      "loss": 0.7722,
      "step": 6312
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2853598883875035,
      "learning_rate": 6.020287903140266e-06,
      "loss": 0.1959,
      "step": 6313
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.315768973307454,
      "learning_rate": 6.019153845513583e-06,
      "loss": 0.2492,
      "step": 6314
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3328433004418445,
      "learning_rate": 6.018019733182772e-06,
      "loss": 0.1857,
      "step": 6315
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5841571644699906,
      "learning_rate": 6.016885566208706e-06,
      "loss": 0.2486,
      "step": 6316
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1870924265019922,
      "learning_rate": 6.01575134465226e-06,
      "loss": 0.1482,
      "step": 6317
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.488639556326491,
      "learning_rate": 6.0146170685743185e-06,
      "loss": 0.5919,
      "step": 6318
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.336588295500303,
      "learning_rate": 6.013482738035765e-06,
      "loss": 0.1923,
      "step": 6319
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.374509516732174,
      "learning_rate": 6.012348353097484e-06,
      "loss": 0.2108,
      "step": 6320
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4314863576343633,
      "learning_rate": 6.011213913820364e-06,
      "loss": 0.1782,
      "step": 6321
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.483039803388748,
      "learning_rate": 6.010079420265298e-06,
      "loss": 0.2066,
      "step": 6322
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2134817967379525,
      "learning_rate": 6.0089448724931836e-06,
      "loss": 0.1716,
      "step": 6323
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2544916123895111,
      "learning_rate": 6.007810270564916e-06,
      "loss": 0.1593,
      "step": 6324
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3369852528936066,
      "learning_rate": 6.006675614541397e-06,
      "loss": 0.2108,
      "step": 6325
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3928096542961756,
      "learning_rate": 6.005540904483529e-06,
      "loss": 0.1911,
      "step": 6326
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.488883859381624,
      "learning_rate": 6.004406140452221e-06,
      "loss": 0.7346,
      "step": 6327
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.289322667574908,
      "learning_rate": 6.003271322508381e-06,
      "loss": 0.1857,
      "step": 6328
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.352329087691462,
      "learning_rate": 6.002136450712924e-06,
      "loss": 0.2203,
      "step": 6329
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3021521239738627,
      "learning_rate": 6.001001525126762e-06,
      "loss": 0.2233,
      "step": 6330
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2048483133621708,
      "learning_rate": 5.9998665458108155e-06,
      "loss": 0.2254,
      "step": 6331
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2440829804628666,
      "learning_rate": 5.998731512826006e-06,
      "loss": 0.1708,
      "step": 6332
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.737432064359035,
      "learning_rate": 5.9975964262332554e-06,
      "loss": 0.4908,
      "step": 6333
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4185981452451342,
      "learning_rate": 5.996461286093491e-06,
      "loss": 0.1882,
      "step": 6334
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5086443231035676,
      "learning_rate": 5.9953260924676446e-06,
      "loss": 0.2472,
      "step": 6335
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.310336954815626,
      "learning_rate": 5.994190845416649e-06,
      "loss": 0.2043,
      "step": 6336
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.388052397772874,
      "learning_rate": 5.993055545001437e-06,
      "loss": 0.1975,
      "step": 6337
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4282144729464996,
      "learning_rate": 5.991920191282948e-06,
      "loss": 0.1752,
      "step": 6338
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4173446922685864,
      "learning_rate": 5.990784784322126e-06,
      "loss": 0.1945,
      "step": 6339
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.400586578613333,
      "learning_rate": 5.9896493241799115e-06,
      "loss": 0.2114,
      "step": 6340
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3621881375185332,
      "learning_rate": 5.988513810917254e-06,
      "loss": 0.2026,
      "step": 6341
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1921816035416763,
      "learning_rate": 5.9873782445951e-06,
      "loss": 0.2189,
      "step": 6342
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.323916164001303,
      "learning_rate": 5.986242625274407e-06,
      "loss": 0.2143,
      "step": 6343
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2273759572975567,
      "learning_rate": 5.9851069530161265e-06,
      "loss": 0.2084,
      "step": 6344
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6739554039408258,
      "learning_rate": 5.983971227881219e-06,
      "loss": 0.2101,
      "step": 6345
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3402818777437433,
      "learning_rate": 5.982835449930644e-06,
      "loss": 0.1547,
      "step": 6346
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3433283134082625,
      "learning_rate": 5.981699619225368e-06,
      "loss": 0.1898,
      "step": 6347
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4108050089036783,
      "learning_rate": 5.980563735826355e-06,
      "loss": 0.2186,
      "step": 6348
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1332072821247325,
      "learning_rate": 5.979427799794579e-06,
      "loss": 0.1605,
      "step": 6349
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.847725651219658,
      "learning_rate": 5.9782918111910085e-06,
      "loss": 0.5338,
      "step": 6350
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2091786517248881,
      "learning_rate": 5.977155770076618e-06,
      "loss": 0.2137,
      "step": 6351
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1839504407795458,
      "learning_rate": 5.97601967651239e-06,
      "loss": 0.1899,
      "step": 6352
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1579553877660862,
      "learning_rate": 5.974883530559302e-06,
      "loss": 0.1595,
      "step": 6353
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1934046565649217,
      "learning_rate": 5.97374733227834e-06,
      "loss": 0.1813,
      "step": 6354
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.522164194237041,
      "learning_rate": 5.97261108173049e-06,
      "loss": 0.5392,
      "step": 6355
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.769575816204579,
      "learning_rate": 5.97147477897674e-06,
      "loss": 0.6495,
      "step": 6356
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2036694679804498,
      "learning_rate": 5.970338424078084e-06,
      "loss": 0.1724,
      "step": 6357
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2452484179167644,
      "learning_rate": 5.969202017095517e-06,
      "loss": 0.1861,
      "step": 6358
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2018314197124143,
      "learning_rate": 5.968065558090036e-06,
      "loss": 0.1867,
      "step": 6359
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.356294461046957,
      "learning_rate": 5.966929047122641e-06,
      "loss": 0.18,
      "step": 6360
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3255124937647749,
      "learning_rate": 5.965792484254336e-06,
      "loss": 0.2095,
      "step": 6361
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.2302897521075975,
      "learning_rate": 5.964655869546128e-06,
      "loss": 0.3118,
      "step": 6362
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.291048859009271,
      "learning_rate": 5.963519203059025e-06,
      "loss": 0.1627,
      "step": 6363
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2439628346359013,
      "learning_rate": 5.962382484854038e-06,
      "loss": 0.1975,
      "step": 6364
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1378770205241788,
      "learning_rate": 5.961245714992184e-06,
      "loss": 0.1457,
      "step": 6365
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1367603185012025,
      "learning_rate": 5.9601088935344785e-06,
      "loss": 0.1675,
      "step": 6366
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2476079926971129,
      "learning_rate": 5.9589720205419425e-06,
      "loss": 0.1429,
      "step": 6367
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4171734097597941,
      "learning_rate": 5.957835096075597e-06,
      "loss": 0.222,
      "step": 6368
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.176774758143685,
      "learning_rate": 5.956698120196469e-06,
      "loss": 0.1813,
      "step": 6369
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.382647891989676,
      "learning_rate": 5.955561092965588e-06,
      "loss": 0.2267,
      "step": 6370
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3247987986788528,
      "learning_rate": 5.9544240144439835e-06,
      "loss": 0.228,
      "step": 6371
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3269420621580488,
      "learning_rate": 5.953286884692688e-06,
      "loss": 0.2056,
      "step": 6372
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4553216205781438,
      "learning_rate": 5.95214970377274e-06,
      "loss": 0.2063,
      "step": 6373
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3425910116135131,
      "learning_rate": 5.951012471745181e-06,
      "loss": 0.2347,
      "step": 6374
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3962498087779134,
      "learning_rate": 5.949875188671049e-06,
      "loss": 0.2357,
      "step": 6375
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.442870179887074,
      "learning_rate": 5.948737854611389e-06,
      "loss": 0.1731,
      "step": 6376
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5360169247097335,
      "learning_rate": 5.947600469627253e-06,
      "loss": 0.2285,
      "step": 6377
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1387305531954393,
      "learning_rate": 5.946463033779687e-06,
      "loss": 0.1554,
      "step": 6378
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.824312749942107,
      "learning_rate": 5.945325547129745e-06,
      "loss": 0.65,
      "step": 6379
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3442747503456838,
      "learning_rate": 5.944188009738483e-06,
      "loss": 0.2047,
      "step": 6380
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5116092199632625,
      "learning_rate": 5.943050421666962e-06,
      "loss": 0.2073,
      "step": 6381
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3126078406247952,
      "learning_rate": 5.941912782976239e-06,
      "loss": 0.2028,
      "step": 6382
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2668469679867702,
      "learning_rate": 5.94077509372738e-06,
      "loss": 0.1658,
      "step": 6383
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2331608070574458,
      "learning_rate": 5.939637353981451e-06,
      "loss": 0.174,
      "step": 6384
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3844469717760677,
      "learning_rate": 5.938499563799521e-06,
      "loss": 0.2148,
      "step": 6385
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3783196853090651,
      "learning_rate": 5.937361723242665e-06,
      "loss": 0.1829,
      "step": 6386
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3181039909223589,
      "learning_rate": 5.936223832371956e-06,
      "loss": 0.1985,
      "step": 6387
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4296927282769185,
      "learning_rate": 5.93508589124847e-06,
      "loss": 0.2021,
      "step": 6388
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4129004479727896,
      "learning_rate": 5.9339478999332874e-06,
      "loss": 0.2301,
      "step": 6389
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.24118835363352,
      "learning_rate": 5.932809858487493e-06,
      "loss": 0.1598,
      "step": 6390
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1243158269688118,
      "learning_rate": 5.931671766972172e-06,
      "loss": 0.1635,
      "step": 6391
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4752454909314463,
      "learning_rate": 5.9305336254484105e-06,
      "loss": 0.2243,
      "step": 6392
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2603702706787374,
      "learning_rate": 5.929395433977301e-06,
      "loss": 0.2454,
      "step": 6393
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1428968817615206,
      "learning_rate": 5.928257192619937e-06,
      "loss": 0.1684,
      "step": 6394
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.387450367807098,
      "learning_rate": 5.927118901437416e-06,
      "loss": 0.2049,
      "step": 6395
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.423292449416241,
      "learning_rate": 5.925980560490834e-06,
      "loss": 0.2093,
      "step": 6396
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2822988742082448,
      "learning_rate": 5.924842169841294e-06,
      "loss": 0.1937,
      "step": 6397
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3155132368791596,
      "learning_rate": 5.923703729549901e-06,
      "loss": 0.1819,
      "step": 6398
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5626142861090788,
      "learning_rate": 5.922565239677762e-06,
      "loss": 0.2641,
      "step": 6399
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.308320508740407,
      "learning_rate": 5.921426700285986e-06,
      "loss": 0.1748,
      "step": 6400
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2876023378190788,
      "learning_rate": 5.920288111435685e-06,
      "loss": 0.2238,
      "step": 6401
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2336669247018972,
      "learning_rate": 5.9191494731879736e-06,
      "loss": 0.1851,
      "step": 6402
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.1882990397510005,
      "learning_rate": 5.918010785603971e-06,
      "loss": 0.7286,
      "step": 6403
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3885533454512806,
      "learning_rate": 5.916872048744797e-06,
      "loss": 0.1919,
      "step": 6404
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2347547274853918,
      "learning_rate": 5.915733262671573e-06,
      "loss": 0.2041,
      "step": 6405
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4240502757995508,
      "learning_rate": 5.914594427445426e-06,
      "loss": 0.2152,
      "step": 6406
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2365012189078137,
      "learning_rate": 5.913455543127484e-06,
      "loss": 0.1982,
      "step": 6407
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2404379751449344,
      "learning_rate": 5.912316609778876e-06,
      "loss": 0.2051,
      "step": 6408
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3768777308964077,
      "learning_rate": 5.911177627460739e-06,
      "loss": 0.2185,
      "step": 6409
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3143196388988154,
      "learning_rate": 5.910038596234205e-06,
      "loss": 0.1495,
      "step": 6410
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2938354765102151,
      "learning_rate": 5.908899516160414e-06,
      "loss": 0.168,
      "step": 6411
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4387654948685327,
      "learning_rate": 5.9077603873005095e-06,
      "loss": 0.1971,
      "step": 6412
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2633516864995231,
      "learning_rate": 5.906621209715634e-06,
      "loss": 0.1665,
      "step": 6413
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3178462276839193,
      "learning_rate": 5.905481983466932e-06,
      "loss": 0.196,
      "step": 6414
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.434962758979205,
      "learning_rate": 5.904342708615556e-06,
      "loss": 0.5096,
      "step": 6415
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.3323461858176815,
      "learning_rate": 5.903203385222656e-06,
      "loss": 0.6354,
      "step": 6416
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3684449014669653,
      "learning_rate": 5.902064013349385e-06,
      "loss": 0.2103,
      "step": 6417
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3605920854798437,
      "learning_rate": 5.900924593056902e-06,
      "loss": 0.2166,
      "step": 6418
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3185792753387833,
      "learning_rate": 5.899785124406366e-06,
      "loss": 0.1688,
      "step": 6419
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.535848228670038,
      "learning_rate": 5.898645607458941e-06,
      "loss": 0.1933,
      "step": 6420
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3999551608078555,
      "learning_rate": 5.897506042275789e-06,
      "loss": 0.183,
      "step": 6421
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2242529210332216,
      "learning_rate": 5.896366428918078e-06,
      "loss": 0.1419,
      "step": 6422
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4836206187691554,
      "learning_rate": 5.895226767446978e-06,
      "loss": 0.2209,
      "step": 6423
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3573684868770453,
      "learning_rate": 5.894087057923661e-06,
      "loss": 0.1915,
      "step": 6424
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.494923519387455,
      "learning_rate": 5.892947300409304e-06,
      "loss": 0.2321,
      "step": 6425
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2687737264928087,
      "learning_rate": 5.891807494965083e-06,
      "loss": 0.1679,
      "step": 6426
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4795524925076713,
      "learning_rate": 5.890667641652178e-06,
      "loss": 0.1795,
      "step": 6427
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.312366472918141,
      "learning_rate": 5.889527740531773e-06,
      "loss": 0.1524,
      "step": 6428
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.005535262918279,
      "learning_rate": 5.8883877916650535e-06,
      "loss": 0.4854,
      "step": 6429
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.412068786406388,
      "learning_rate": 5.887247795113206e-06,
      "loss": 0.2051,
      "step": 6430
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3485782267298134,
      "learning_rate": 5.886107750937421e-06,
      "loss": 0.1849,
      "step": 6431
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.0660343015209515,
      "learning_rate": 5.884967659198893e-06,
      "loss": 0.6338,
      "step": 6432
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.283252719212704,
      "learning_rate": 5.883827519958817e-06,
      "loss": 0.1945,
      "step": 6433
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.383096483958068,
      "learning_rate": 5.882687333278391e-06,
      "loss": 0.201,
      "step": 6434
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.304577357921389,
      "learning_rate": 5.881547099218815e-06,
      "loss": 0.1616,
      "step": 6435
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2478015838822425,
      "learning_rate": 5.8804068178412936e-06,
      "loss": 0.1688,
      "step": 6436
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.323507064769264,
      "learning_rate": 5.8792664892070314e-06,
      "loss": 0.2196,
      "step": 6437
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3980612651983149,
      "learning_rate": 5.878126113377239e-06,
      "loss": 0.1881,
      "step": 6438
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3814027750282512,
      "learning_rate": 5.876985690413122e-06,
      "loss": 0.2173,
      "step": 6439
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.295807002156964,
      "learning_rate": 5.8758452203758995e-06,
      "loss": 0.1659,
      "step": 6440
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.578053999684828,
      "learning_rate": 5.874704703326786e-06,
      "loss": 0.204,
      "step": 6441
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3166363777251926,
      "learning_rate": 5.8735641393269975e-06,
      "loss": 0.1593,
      "step": 6442
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4103537034339437,
      "learning_rate": 5.8724235284377565e-06,
      "loss": 0.2189,
      "step": 6443
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3600438782692628,
      "learning_rate": 5.871282870720286e-06,
      "loss": 0.2375,
      "step": 6444
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.371705733346872,
      "learning_rate": 5.870142166235814e-06,
      "loss": 0.2158,
      "step": 6445
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2388547779423673,
      "learning_rate": 5.869001415045567e-06,
      "loss": 0.1721,
      "step": 6446
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3007379947510882,
      "learning_rate": 5.867860617210775e-06,
      "loss": 0.176,
      "step": 6447
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2484623578651421,
      "learning_rate": 5.866719772792674e-06,
      "loss": 0.1837,
      "step": 6448
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.155756294626106,
      "learning_rate": 5.865578881852498e-06,
      "loss": 0.1617,
      "step": 6449
    },
    {
      "epoch": 0.46,
      "grad_norm": 3.198978912643704,
      "learning_rate": 5.864437944451488e-06,
      "loss": 0.431,
      "step": 6450
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3208647477769846,
      "learning_rate": 5.8632969606508815e-06,
      "loss": 0.195,
      "step": 6451
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4377743757926795,
      "learning_rate": 5.862155930511924e-06,
      "loss": 0.2322,
      "step": 6452
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2712935847919,
      "learning_rate": 5.861014854095863e-06,
      "loss": 0.2069,
      "step": 6453
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4941703778333344,
      "learning_rate": 5.859873731463945e-06,
      "loss": 0.2027,
      "step": 6454
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3204050486221552,
      "learning_rate": 5.85873256267742e-06,
      "loss": 0.174,
      "step": 6455
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4286429509496164,
      "learning_rate": 5.857591347797545e-06,
      "loss": 0.2206,
      "step": 6456
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.051661885850453,
      "learning_rate": 5.85645008688557e-06,
      "loss": 0.1427,
      "step": 6457
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3302224970221388,
      "learning_rate": 5.8553087800027585e-06,
      "loss": 0.184,
      "step": 6458
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2557404886319494,
      "learning_rate": 5.85416742721037e-06,
      "loss": 0.1575,
      "step": 6459
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4176027702175606,
      "learning_rate": 5.8530260285696674e-06,
      "loss": 0.2094,
      "step": 6460
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.140488907024391,
      "learning_rate": 5.851884584141916e-06,
      "loss": 0.1615,
      "step": 6461
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4692581979450128,
      "learning_rate": 5.850743093988385e-06,
      "loss": 0.2118,
      "step": 6462
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3168929573000587,
      "learning_rate": 5.849601558170343e-06,
      "loss": 0.1726,
      "step": 6463
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2374304751719958,
      "learning_rate": 5.8484599767490665e-06,
      "loss": 0.2145,
      "step": 6464
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.327025197307282,
      "learning_rate": 5.847318349785829e-06,
      "loss": 0.1798,
      "step": 6465
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2473517908221359,
      "learning_rate": 5.846176677341908e-06,
      "loss": 0.1788,
      "step": 6466
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4175127651531947,
      "learning_rate": 5.845034959478585e-06,
      "loss": 0.2146,
      "step": 6467
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1074090292812042,
      "learning_rate": 5.8438931962571435e-06,
      "loss": 0.1606,
      "step": 6468
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5104794569956375,
      "learning_rate": 5.842751387738865e-06,
      "loss": 0.223,
      "step": 6469
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2197224016780748,
      "learning_rate": 5.841609533985042e-06,
      "loss": 0.2253,
      "step": 6470
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1662703105674344,
      "learning_rate": 5.840467635056964e-06,
      "loss": 0.1654,
      "step": 6471
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.281069834174271,
      "learning_rate": 5.83932569101592e-06,
      "loss": 0.1994,
      "step": 6472
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3414520922450812,
      "learning_rate": 5.838183701923208e-06,
      "loss": 0.2052,
      "step": 6473
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2867824577668747,
      "learning_rate": 5.837041667840125e-06,
      "loss": 0.1953,
      "step": 6474
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.274664089341185,
      "learning_rate": 5.835899588827971e-06,
      "loss": 0.1897,
      "step": 6475
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.247652038364852,
      "learning_rate": 5.834757464948049e-06,
      "loss": 0.1978,
      "step": 6476
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.111805810154478,
      "learning_rate": 5.83361529626166e-06,
      "loss": 0.5317,
      "step": 6477
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3386962283526556,
      "learning_rate": 5.832473082830116e-06,
      "loss": 0.1698,
      "step": 6478
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4294715056398621,
      "learning_rate": 5.8313308247147226e-06,
      "loss": 0.1985,
      "step": 6479
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3744150466120613,
      "learning_rate": 5.830188521976794e-06,
      "loss": 0.2055,
      "step": 6480
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5797107604026737,
      "learning_rate": 5.829046174677643e-06,
      "loss": 0.2563,
      "step": 6481
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3072715248108897,
      "learning_rate": 5.827903782878587e-06,
      "loss": 0.1862,
      "step": 6482
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3226750607331805,
      "learning_rate": 5.826761346640946e-06,
      "loss": 0.2098,
      "step": 6483
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5039041127818567,
      "learning_rate": 5.82561886602604e-06,
      "loss": 0.2061,
      "step": 6484
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3639965472850113,
      "learning_rate": 5.824476341095193e-06,
      "loss": 0.2079,
      "step": 6485
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3268587125144686,
      "learning_rate": 5.823333771909731e-06,
      "loss": 0.1853,
      "step": 6486
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3486487300756063,
      "learning_rate": 5.822191158530983e-06,
      "loss": 0.214,
      "step": 6487
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3417341654137487,
      "learning_rate": 5.8210485010202825e-06,
      "loss": 0.1677,
      "step": 6488
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2829756437540412,
      "learning_rate": 5.819905799438958e-06,
      "loss": 0.2039,
      "step": 6489
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2608536338619447,
      "learning_rate": 5.818763053848347e-06,
      "loss": 0.1934,
      "step": 6490
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6073396735482781,
      "learning_rate": 5.81762026430979e-06,
      "loss": 0.2208,
      "step": 6491
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2741602671703798,
      "learning_rate": 5.816477430884625e-06,
      "loss": 0.156,
      "step": 6492
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2968001085064529,
      "learning_rate": 5.815334553634194e-06,
      "loss": 0.2001,
      "step": 6493
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.437122287301326,
      "learning_rate": 5.814191632619843e-06,
      "loss": 0.194,
      "step": 6494
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6382713465825602,
      "learning_rate": 5.8130486679029216e-06,
      "loss": 0.2679,
      "step": 6495
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.46536819673083,
      "learning_rate": 5.811905659544777e-06,
      "loss": 0.414,
      "step": 6496
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.391566702455359,
      "learning_rate": 5.8107626076067635e-06,
      "loss": 0.186,
      "step": 6497
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3325702424872043,
      "learning_rate": 5.809619512150232e-06,
      "loss": 0.208,
      "step": 6498
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2099104541239016,
      "learning_rate": 5.8084763732365425e-06,
      "loss": 0.1571,
      "step": 6499
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5583130284005118,
      "learning_rate": 5.807333190927054e-06,
      "loss": 0.2273,
      "step": 6500
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3537215103713915,
      "learning_rate": 5.806189965283125e-06,
      "loss": 0.2143,
      "step": 6501
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3913161315149758,
      "learning_rate": 5.8050466963661235e-06,
      "loss": 0.2378,
      "step": 6502
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2780095081635476,
      "learning_rate": 5.803903384237412e-06,
      "loss": 0.1583,
      "step": 6503
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3197158481725357,
      "learning_rate": 5.802760028958362e-06,
      "loss": 0.1951,
      "step": 6504
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3062029036510774,
      "learning_rate": 5.801616630590342e-06,
      "loss": 0.1549,
      "step": 6505
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5011822094458633,
      "learning_rate": 5.800473189194727e-06,
      "loss": 0.1997,
      "step": 6506
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.280410093862145,
      "learning_rate": 5.79932970483289e-06,
      "loss": 0.1766,
      "step": 6507
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.324915390740993,
      "learning_rate": 5.798186177566211e-06,
      "loss": 0.1946,
      "step": 6508
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2260447282313718,
      "learning_rate": 5.797042607456068e-06,
      "loss": 0.2061,
      "step": 6509
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3943975520058736,
      "learning_rate": 5.7958989945638455e-06,
      "loss": 0.1993,
      "step": 6510
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.382449279495089,
      "learning_rate": 5.794755338950926e-06,
      "loss": 0.207,
      "step": 6511
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3461185249876617,
      "learning_rate": 5.793611640678697e-06,
      "loss": 0.207,
      "step": 6512
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.362568649297016,
      "learning_rate": 5.792467899808549e-06,
      "loss": 0.1988,
      "step": 6513
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3492548212186104,
      "learning_rate": 5.791324116401871e-06,
      "loss": 0.179,
      "step": 6514
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3090815463032532,
      "learning_rate": 5.79018029052006e-06,
      "loss": 0.1872,
      "step": 6515
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5097649231636865,
      "learning_rate": 5.789036422224508e-06,
      "loss": 0.1959,
      "step": 6516
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2995178332281412,
      "learning_rate": 5.787892511576617e-06,
      "loss": 0.178,
      "step": 6517
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3567649910791422,
      "learning_rate": 5.786748558637787e-06,
      "loss": 0.186,
      "step": 6518
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2600833120858175,
      "learning_rate": 5.785604563469419e-06,
      "loss": 0.155,
      "step": 6519
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5433940458167406,
      "learning_rate": 5.784460526132918e-06,
      "loss": 0.1966,
      "step": 6520
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4848559604976856,
      "learning_rate": 5.783316446689694e-06,
      "loss": 0.182,
      "step": 6521
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3022873825124526,
      "learning_rate": 5.782172325201155e-06,
      "loss": 0.2118,
      "step": 6522
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3458818653693136,
      "learning_rate": 5.781028161728714e-06,
      "loss": 0.1834,
      "step": 6523
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2499798888822795,
      "learning_rate": 5.77988395633378e-06,
      "loss": 0.1805,
      "step": 6524
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3723680168278265,
      "learning_rate": 5.778739709077777e-06,
      "loss": 0.2063,
      "step": 6525
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5032534056326532,
      "learning_rate": 5.7775954200221206e-06,
      "loss": 0.2136,
      "step": 6526
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5117166346725868,
      "learning_rate": 5.77645108922823e-06,
      "loss": 0.2272,
      "step": 6527
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2499605548875892,
      "learning_rate": 5.775306716757529e-06,
      "loss": 0.1723,
      "step": 6528
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2172999915430145,
      "learning_rate": 5.774162302671444e-06,
      "loss": 0.2073,
      "step": 6529
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3838191904059878,
      "learning_rate": 5.773017847031403e-06,
      "loss": 0.1605,
      "step": 6530
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2005138673340463,
      "learning_rate": 5.771873349898835e-06,
      "loss": 0.1747,
      "step": 6531
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2074800256054974,
      "learning_rate": 5.770728811335171e-06,
      "loss": 0.1542,
      "step": 6532
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4086544387965998,
      "learning_rate": 5.769584231401847e-06,
      "loss": 0.1723,
      "step": 6533
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4471187224720532,
      "learning_rate": 5.768439610160299e-06,
      "loss": 0.1801,
      "step": 6534
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4321299981649063,
      "learning_rate": 5.767294947671967e-06,
      "loss": 0.2127,
      "step": 6535
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.395255678701889,
      "learning_rate": 5.766150243998288e-06,
      "loss": 0.2329,
      "step": 6536
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.207618633862699,
      "learning_rate": 5.7650054992007074e-06,
      "loss": 0.2048,
      "step": 6537
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1460654610544796,
      "learning_rate": 5.7638607133406725e-06,
      "loss": 0.1565,
      "step": 6538
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3829574436665721,
      "learning_rate": 5.762715886479629e-06,
      "loss": 0.228,
      "step": 6539
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2864625718187568,
      "learning_rate": 5.761571018679025e-06,
      "loss": 0.1876,
      "step": 6540
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4898457334327468,
      "learning_rate": 5.760426110000317e-06,
      "loss": 0.2363,
      "step": 6541
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.362393062545094,
      "learning_rate": 5.759281160504955e-06,
      "loss": 0.1571,
      "step": 6542
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4242618532825995,
      "learning_rate": 5.758136170254397e-06,
      "loss": 0.2033,
      "step": 6543
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.245624717784355,
      "learning_rate": 5.756991139310101e-06,
      "loss": 0.2083,
      "step": 6544
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3930637078334769,
      "learning_rate": 5.755846067733527e-06,
      "loss": 0.2049,
      "step": 6545
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.454082084339547,
      "learning_rate": 5.754700955586141e-06,
      "loss": 0.2181,
      "step": 6546
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.326006153677617,
      "learning_rate": 5.753555802929405e-06,
      "loss": 0.2113,
      "step": 6547
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2632819620314486,
      "learning_rate": 5.752410609824788e-06,
      "loss": 0.1682,
      "step": 6548
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.781802814930023,
      "learning_rate": 5.7512653763337565e-06,
      "loss": 0.5712,
      "step": 6549
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3219037170566765,
      "learning_rate": 5.750120102517787e-06,
      "loss": 0.2033,
      "step": 6550
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.474044600550226,
      "learning_rate": 5.748974788438349e-06,
      "loss": 0.1974,
      "step": 6551
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5234361024017575,
      "learning_rate": 5.74782943415692e-06,
      "loss": 0.2367,
      "step": 6552
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4046211089672842,
      "learning_rate": 5.746684039734976e-06,
      "loss": 0.2021,
      "step": 6553
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4051430098137947,
      "learning_rate": 5.745538605234003e-06,
      "loss": 0.2246,
      "step": 6554
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.202109561029295,
      "learning_rate": 5.744393130715478e-06,
      "loss": 0.1746,
      "step": 6555
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.392074710822377,
      "learning_rate": 5.7432476162408854e-06,
      "loss": 0.2,
      "step": 6556
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6035897987328618,
      "learning_rate": 5.742102061871716e-06,
      "loss": 0.197,
      "step": 6557
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2301113007281486,
      "learning_rate": 5.740956467669454e-06,
      "loss": 0.1687,
      "step": 6558
    },
    {
      "epoch": 0.47,
      "grad_norm": 12.496434456399223,
      "learning_rate": 5.739810833695594e-06,
      "loss": 0.5074,
      "step": 6559
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3882148171153095,
      "learning_rate": 5.738665160011627e-06,
      "loss": 0.2081,
      "step": 6560
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.493534877243792,
      "learning_rate": 5.7375194466790495e-06,
      "loss": 0.207,
      "step": 6561
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3523957007748022,
      "learning_rate": 5.736373693759357e-06,
      "loss": 0.1435,
      "step": 6562
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.203909406807871,
      "learning_rate": 5.73522790131405e-06,
      "loss": 0.6458,
      "step": 6563
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.302314193502512,
      "learning_rate": 5.734082069404631e-06,
      "loss": 0.1939,
      "step": 6564
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3508130527036732,
      "learning_rate": 5.732936198092603e-06,
      "loss": 0.2098,
      "step": 6565
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.021243716312969,
      "learning_rate": 5.7317902874394694e-06,
      "loss": 0.6822,
      "step": 6566
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3980451685602266,
      "learning_rate": 5.730644337506743e-06,
      "loss": 0.1566,
      "step": 6567
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4976865161057045,
      "learning_rate": 5.729498348355931e-06,
      "loss": 0.2238,
      "step": 6568
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.906146507194042,
      "learning_rate": 5.728352320048545e-06,
      "loss": 0.6332,
      "step": 6569
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4098158614209872,
      "learning_rate": 5.7272062526461e-06,
      "loss": 0.2076,
      "step": 6570
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.317533695252874,
      "learning_rate": 5.726060146210113e-06,
      "loss": 0.1774,
      "step": 6571
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2906586447583197,
      "learning_rate": 5.724914000802102e-06,
      "loss": 0.1951,
      "step": 6572
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.414917475059135,
      "learning_rate": 5.723767816483587e-06,
      "loss": 0.1741,
      "step": 6573
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5828806073634103,
      "learning_rate": 5.72262159331609e-06,
      "loss": 0.191,
      "step": 6574
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.529656325700105,
      "learning_rate": 5.721475331361137e-06,
      "loss": 0.2253,
      "step": 6575
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3483928013740227,
      "learning_rate": 5.720329030680256e-06,
      "loss": 0.1583,
      "step": 6576
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.356213511425819,
      "learning_rate": 5.719182691334972e-06,
      "loss": 0.6198,
      "step": 6577
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4159356903603015,
      "learning_rate": 5.718036313386821e-06,
      "loss": 0.184,
      "step": 6578
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.298559359192132,
      "learning_rate": 5.716889896897331e-06,
      "loss": 0.1663,
      "step": 6579
    },
    {
      "epoch": 0.47,
      "grad_norm": 11.235786677172939,
      "learning_rate": 5.715743441928041e-06,
      "loss": 0.7425,
      "step": 6580
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2350710716526738,
      "learning_rate": 5.7145969485404865e-06,
      "loss": 0.1666,
      "step": 6581
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4163438485808915,
      "learning_rate": 5.713450416796206e-06,
      "loss": 0.204,
      "step": 6582
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.354082948875915,
      "learning_rate": 5.71230384675674e-06,
      "loss": 0.1989,
      "step": 6583
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4846696665257617,
      "learning_rate": 5.711157238483635e-06,
      "loss": 0.2071,
      "step": 6584
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4616357037774714,
      "learning_rate": 5.7100105920384365e-06,
      "loss": 0.2028,
      "step": 6585
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5201786333393195,
      "learning_rate": 5.708863907482688e-06,
      "loss": 0.2288,
      "step": 6586
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.237476820167331,
      "learning_rate": 5.707717184877941e-06,
      "loss": 0.1873,
      "step": 6587
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4882630525263791,
      "learning_rate": 5.706570424285747e-06,
      "loss": 0.2235,
      "step": 6588
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5193640327634763,
      "learning_rate": 5.705423625767661e-06,
      "loss": 0.1962,
      "step": 6589
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.488793686929965,
      "learning_rate": 5.704276789385237e-06,
      "loss": 0.2321,
      "step": 6590
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.211678914986396,
      "learning_rate": 5.703129915200032e-06,
      "loss": 0.1306,
      "step": 6591
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4452476328485626,
      "learning_rate": 5.701983003273607e-06,
      "loss": 0.2044,
      "step": 6592
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.941880668226676,
      "learning_rate": 5.700836053667524e-06,
      "loss": 0.6017,
      "step": 6593
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3334533712039534,
      "learning_rate": 5.6996890664433444e-06,
      "loss": 0.1733,
      "step": 6594
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4157998865403114,
      "learning_rate": 5.698542041662636e-06,
      "loss": 0.1813,
      "step": 6595
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3426195883272436,
      "learning_rate": 5.6973949793869665e-06,
      "loss": 0.2308,
      "step": 6596
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3432609085054619,
      "learning_rate": 5.696247879677905e-06,
      "loss": 0.1976,
      "step": 6597
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.465047681545308,
      "learning_rate": 5.695100742597023e-06,
      "loss": 0.2107,
      "step": 6598
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.351271092964941,
      "learning_rate": 5.693953568205894e-06,
      "loss": 0.2075,
      "step": 6599
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4097710561011356,
      "learning_rate": 5.6928063565660955e-06,
      "loss": 0.1871,
      "step": 6600
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3803266064713453,
      "learning_rate": 5.691659107739205e-06,
      "loss": 0.2166,
      "step": 6601
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.301797639221462,
      "learning_rate": 5.690511821786799e-06,
      "loss": 0.6561,
      "step": 6602
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3030832555992136,
      "learning_rate": 5.689364498770464e-06,
      "loss": 0.1911,
      "step": 6603
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2376309669081882,
      "learning_rate": 5.688217138751779e-06,
      "loss": 0.1801,
      "step": 6604
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.063643900205768,
      "learning_rate": 5.687069741792334e-06,
      "loss": 0.5719,
      "step": 6605
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1588282811234394,
      "learning_rate": 5.685922307953714e-06,
      "loss": 0.1588,
      "step": 6606
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2998763576833958,
      "learning_rate": 5.684774837297511e-06,
      "loss": 0.2183,
      "step": 6607
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2433683507958764,
      "learning_rate": 5.6836273298853125e-06,
      "loss": 0.2105,
      "step": 6608
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3256238480455282,
      "learning_rate": 5.682479785778716e-06,
      "loss": 0.1579,
      "step": 6609
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.416507335296526,
      "learning_rate": 5.681332205039316e-06,
      "loss": 0.2062,
      "step": 6610
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1959167561202941,
      "learning_rate": 5.680184587728711e-06,
      "loss": 0.1261,
      "step": 6611
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3383093427406567,
      "learning_rate": 5.6790369339085e-06,
      "loss": 0.2549,
      "step": 6612
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3447069054150522,
      "learning_rate": 5.677889243640281e-06,
      "loss": 0.1911,
      "step": 6613
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3073870274351875,
      "learning_rate": 5.676741516985662e-06,
      "loss": 0.2171,
      "step": 6614
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2179887755105836,
      "learning_rate": 5.675593754006248e-06,
      "loss": 0.2135,
      "step": 6615
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4375101013690217,
      "learning_rate": 5.674445954763646e-06,
      "loss": 0.2053,
      "step": 6616
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4484395951201523,
      "learning_rate": 5.673298119319461e-06,
      "loss": 0.1374,
      "step": 6617
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5721007717575335,
      "learning_rate": 5.672150247735311e-06,
      "loss": 0.2196,
      "step": 6618
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5482159371704185,
      "learning_rate": 5.671002340072805e-06,
      "loss": 0.2191,
      "step": 6619
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.375198427149108,
      "learning_rate": 5.669854396393559e-06,
      "loss": 0.2374,
      "step": 6620
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3284557340254821,
      "learning_rate": 5.66870641675919e-06,
      "loss": 0.241,
      "step": 6621
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5456241729458258,
      "learning_rate": 5.667558401231318e-06,
      "loss": 0.1554,
      "step": 6622
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.514184569796053,
      "learning_rate": 5.666410349871563e-06,
      "loss": 0.1735,
      "step": 6623
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.709034238782593,
      "learning_rate": 5.665262262741548e-06,
      "loss": 0.7653,
      "step": 6624
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3817859441044498,
      "learning_rate": 5.664114139902897e-06,
      "loss": 0.1734,
      "step": 6625
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4096745830497825,
      "learning_rate": 5.6629659814172365e-06,
      "loss": 0.1697,
      "step": 6626
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3168424168755262,
      "learning_rate": 5.661817787346198e-06,
      "loss": 0.2086,
      "step": 6627
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3622785181946422,
      "learning_rate": 5.660669557751409e-06,
      "loss": 0.2058,
      "step": 6628
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.2315474394610164,
      "learning_rate": 5.659521292694502e-06,
      "loss": 0.3605,
      "step": 6629
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2399285585743323,
      "learning_rate": 5.6583729922371135e-06,
      "loss": 0.1819,
      "step": 6630
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4703244971557394,
      "learning_rate": 5.657224656440878e-06,
      "loss": 0.2211,
      "step": 6631
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4120333186098457,
      "learning_rate": 5.656076285367434e-06,
      "loss": 0.1898,
      "step": 6632
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3875502650553995,
      "learning_rate": 5.65492787907842e-06,
      "loss": 0.2044,
      "step": 6633
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3045646050508035,
      "learning_rate": 5.653779437635481e-06,
      "loss": 0.1763,
      "step": 6634
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4478781748780905,
      "learning_rate": 5.65263096110026e-06,
      "loss": 0.2167,
      "step": 6635
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.393015761074465,
      "learning_rate": 5.6514824495344e-06,
      "loss": 0.1909,
      "step": 6636
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1692813776182762,
      "learning_rate": 5.65033390299955e-06,
      "loss": 0.15,
      "step": 6637
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5013667525119645,
      "learning_rate": 5.64918532155736e-06,
      "loss": 0.2549,
      "step": 6638
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1807121184803353,
      "learning_rate": 5.6480367052694814e-06,
      "loss": 0.1872,
      "step": 6639
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.447065956867901,
      "learning_rate": 5.646888054197568e-06,
      "loss": 0.1996,
      "step": 6640
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.540889333290801,
      "learning_rate": 5.645739368403272e-06,
      "loss": 0.2144,
      "step": 6641
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4663105084313814,
      "learning_rate": 5.644590647948252e-06,
      "loss": 0.2349,
      "step": 6642
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4120831257870394,
      "learning_rate": 5.643441892894168e-06,
      "loss": 0.2338,
      "step": 6643
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.423531552999105,
      "learning_rate": 5.642293103302681e-06,
      "loss": 0.2014,
      "step": 6644
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3576783681898044,
      "learning_rate": 5.64114427923545e-06,
      "loss": 0.1745,
      "step": 6645
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2271112875772985,
      "learning_rate": 5.639995420754141e-06,
      "loss": 0.1794,
      "step": 6646
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.284576286012348,
      "learning_rate": 5.638846527920421e-06,
      "loss": 0.2049,
      "step": 6647
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.735140367055312,
      "learning_rate": 5.637697600795957e-06,
      "loss": 0.5307,
      "step": 6648
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5159081046847358,
      "learning_rate": 5.63654863944242e-06,
      "loss": 0.1797,
      "step": 6649
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.23526647937402,
      "learning_rate": 5.635399643921479e-06,
      "loss": 0.5596,
      "step": 6650
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.467072851912935,
      "learning_rate": 5.634250614294811e-06,
      "loss": 0.2073,
      "step": 6651
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3815219896405229,
      "learning_rate": 5.633101550624089e-06,
      "loss": 0.2072,
      "step": 6652
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.260196760246533,
      "learning_rate": 5.631952452970992e-06,
      "loss": 0.1917,
      "step": 6653
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4720139244592239,
      "learning_rate": 5.630803321397198e-06,
      "loss": 0.2407,
      "step": 6654
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3286458796307956,
      "learning_rate": 5.6296541559643854e-06,
      "loss": 0.1979,
      "step": 6655
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.306094194100853,
      "learning_rate": 5.628504956734242e-06,
      "loss": 0.2366,
      "step": 6656
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.319116897884429,
      "learning_rate": 5.627355723768449e-06,
      "loss": 0.2167,
      "step": 6657
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.272662775051772,
      "learning_rate": 5.626206457128692e-06,
      "loss": 0.1859,
      "step": 6658
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2047439538386733,
      "learning_rate": 5.62505715687666e-06,
      "loss": 0.1743,
      "step": 6659
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2344994679699708,
      "learning_rate": 5.623907823074044e-06,
      "loss": 0.1813,
      "step": 6660
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3093303281663546,
      "learning_rate": 5.622758455782534e-06,
      "loss": 0.1957,
      "step": 6661
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3582004498814646,
      "learning_rate": 5.621609055063826e-06,
      "loss": 0.1726,
      "step": 6662
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.176726358488571,
      "learning_rate": 5.62045962097961e-06,
      "loss": 0.1608,
      "step": 6663
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2183205294078407,
      "learning_rate": 5.6193101535915894e-06,
      "loss": 0.1742,
      "step": 6664
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.265218178169771,
      "learning_rate": 5.618160652961459e-06,
      "loss": 0.1311,
      "step": 6665
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2435751219275146,
      "learning_rate": 5.617011119150922e-06,
      "loss": 0.1777,
      "step": 6666
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.914470400501476,
      "learning_rate": 5.615861552221677e-06,
      "loss": 0.7304,
      "step": 6667
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1829092287141503,
      "learning_rate": 5.614711952235432e-06,
      "loss": 0.2057,
      "step": 6668
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2986090537159352,
      "learning_rate": 5.6135623192538915e-06,
      "loss": 0.1649,
      "step": 6669
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3058567665023229,
      "learning_rate": 5.612412653338763e-06,
      "loss": 0.1851,
      "step": 6670
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4467702225431087,
      "learning_rate": 5.6112629545517565e-06,
      "loss": 0.2026,
      "step": 6671
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2249025798770863,
      "learning_rate": 5.610113222954583e-06,
      "loss": 0.1815,
      "step": 6672
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2688443562901717,
      "learning_rate": 5.608963458608953e-06,
      "loss": 0.1762,
      "step": 6673
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.985408878215837,
      "learning_rate": 5.607813661576587e-06,
      "loss": 0.649,
      "step": 6674
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2681637172408171,
      "learning_rate": 5.606663831919198e-06,
      "loss": 0.1903,
      "step": 6675
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8481052922301722,
      "learning_rate": 5.605513969698504e-06,
      "loss": 0.21,
      "step": 6676
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2842823473237996,
      "learning_rate": 5.604364074976227e-06,
      "loss": 0.1769,
      "step": 6677
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3543061435568093,
      "learning_rate": 5.603214147814086e-06,
      "loss": 0.2237,
      "step": 6678
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4139731288564583,
      "learning_rate": 5.602064188273806e-06,
      "loss": 0.2254,
      "step": 6679
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4527351608229153,
      "learning_rate": 5.600914196417112e-06,
      "loss": 0.2305,
      "step": 6680
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2851283704251248,
      "learning_rate": 5.599764172305732e-06,
      "loss": 0.17,
      "step": 6681
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3838102091405438,
      "learning_rate": 5.598614116001394e-06,
      "loss": 0.1657,
      "step": 6682
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5208474089354278,
      "learning_rate": 5.5974640275658284e-06,
      "loss": 0.1863,
      "step": 6683
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4431061962676712,
      "learning_rate": 5.596313907060766e-06,
      "loss": 0.2286,
      "step": 6684
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.301210709072611,
      "learning_rate": 5.595163754547945e-06,
      "loss": 0.1946,
      "step": 6685
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3528715438233607,
      "learning_rate": 5.594013570089096e-06,
      "loss": 0.1714,
      "step": 6686
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.863473383864885,
      "learning_rate": 5.592863353745959e-06,
      "loss": 0.6037,
      "step": 6687
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5296372419376616,
      "learning_rate": 5.591713105580272e-06,
      "loss": 0.2185,
      "step": 6688
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3475353144390334,
      "learning_rate": 5.590562825653777e-06,
      "loss": 0.1708,
      "step": 6689
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3621160200499347,
      "learning_rate": 5.589412514028215e-06,
      "loss": 0.2149,
      "step": 6690
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2343770963578025,
      "learning_rate": 5.588262170765332e-06,
      "loss": 0.1475,
      "step": 6691
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.226253359094388,
      "learning_rate": 5.587111795926871e-06,
      "loss": 0.1955,
      "step": 6692
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.665249904996852,
      "learning_rate": 5.585961389574582e-06,
      "loss": 0.6404,
      "step": 6693
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4663346383017628,
      "learning_rate": 5.584810951770215e-06,
      "loss": 0.2036,
      "step": 6694
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4705378305401333,
      "learning_rate": 5.5836604825755165e-06,
      "loss": 0.2304,
      "step": 6695
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.328203328595461,
      "learning_rate": 5.582509982052244e-06,
      "loss": 0.1689,
      "step": 6696
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4327974602094515,
      "learning_rate": 5.58135945026215e-06,
      "loss": 0.1966,
      "step": 6697
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5586132079801607,
      "learning_rate": 5.580208887266989e-06,
      "loss": 0.2036,
      "step": 6698
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2520196199064095,
      "learning_rate": 5.579058293128522e-06,
      "loss": 0.1665,
      "step": 6699
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3531001361487087,
      "learning_rate": 5.577907667908505e-06,
      "loss": 0.1832,
      "step": 6700
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2592648229797585,
      "learning_rate": 5.5767570116687e-06,
      "loss": 0.1551,
      "step": 6701
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.391433623850729,
      "learning_rate": 5.57560632447087e-06,
      "loss": 0.2143,
      "step": 6702
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.289281399178544,
      "learning_rate": 5.5744556063767804e-06,
      "loss": 0.2076,
      "step": 6703
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.466612044841786,
      "learning_rate": 5.573304857448195e-06,
      "loss": 0.5673,
      "step": 6704
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.322530237319094,
      "learning_rate": 5.572154077746882e-06,
      "loss": 0.1739,
      "step": 6705
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2205176424784763,
      "learning_rate": 5.571003267334612e-06,
      "loss": 0.1744,
      "step": 6706
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3967933641228538,
      "learning_rate": 5.569852426273154e-06,
      "loss": 0.228,
      "step": 6707
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2940529143495834,
      "learning_rate": 5.568701554624284e-06,
      "loss": 0.2075,
      "step": 6708
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3132503436465324,
      "learning_rate": 5.567550652449773e-06,
      "loss": 0.1854,
      "step": 6709
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1810895547714022,
      "learning_rate": 5.566399719811396e-06,
      "loss": 0.1499,
      "step": 6710
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3570662722160278,
      "learning_rate": 5.565248756770934e-06,
      "loss": 0.2078,
      "step": 6711
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.132891668764976,
      "learning_rate": 5.564097763390164e-06,
      "loss": 0.5219,
      "step": 6712
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4552138400814318,
      "learning_rate": 5.5629467397308666e-06,
      "loss": 0.2003,
      "step": 6713
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4386516292199851,
      "learning_rate": 5.561795685854825e-06,
      "loss": 0.1968,
      "step": 6714
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.4208210831262718,
      "learning_rate": 5.560644601823823e-06,
      "loss": 0.5845,
      "step": 6715
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2243283906933147,
      "learning_rate": 5.5594934876996455e-06,
      "loss": 0.1686,
      "step": 6716
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5637156771632874,
      "learning_rate": 5.558342343544081e-06,
      "loss": 0.2178,
      "step": 6717
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.023920571981453,
      "learning_rate": 5.5571911694189175e-06,
      "loss": 0.404,
      "step": 6718
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.475071302654246,
      "learning_rate": 5.556039965385945e-06,
      "loss": 0.2102,
      "step": 6719
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.7532609839625657,
      "learning_rate": 5.5548887315069575e-06,
      "loss": 0.5653,
      "step": 6720
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3775939142712832,
      "learning_rate": 5.553737467843747e-06,
      "loss": 0.202,
      "step": 6721
    },
    {
      "epoch": 0.48,
      "grad_norm": 10.951403106549975,
      "learning_rate": 5.552586174458108e-06,
      "loss": 0.6949,
      "step": 6722
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3748090969676616,
      "learning_rate": 5.5514348514118395e-06,
      "loss": 0.1988,
      "step": 6723
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3024851074778925,
      "learning_rate": 5.5502834987667386e-06,
      "loss": 0.1384,
      "step": 6724
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5291363573830403,
      "learning_rate": 5.549132116584606e-06,
      "loss": 0.1884,
      "step": 6725
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.45020599412656,
      "learning_rate": 5.547980704927241e-06,
      "loss": 0.2131,
      "step": 6726
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4983111458375615,
      "learning_rate": 5.546829263856451e-06,
      "loss": 0.2094,
      "step": 6727
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2516208235667108,
      "learning_rate": 5.545677793434039e-06,
      "loss": 0.1676,
      "step": 6728
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5275846375975488,
      "learning_rate": 5.5445262937218106e-06,
      "loss": 0.253,
      "step": 6729
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.445212578479364,
      "learning_rate": 5.543374764781571e-06,
      "loss": 0.6099,
      "step": 6730
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4068861590880137,
      "learning_rate": 5.542223206675137e-06,
      "loss": 0.1808,
      "step": 6731
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2113251939943535,
      "learning_rate": 5.541071619464314e-06,
      "loss": 0.1632,
      "step": 6732
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2824024101668492,
      "learning_rate": 5.5399200032109145e-06,
      "loss": 0.1804,
      "step": 6733
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2874919486512382,
      "learning_rate": 5.538768357976755e-06,
      "loss": 0.2087,
      "step": 6734
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.729622441398244,
      "learning_rate": 5.537616683823651e-06,
      "loss": 0.5009,
      "step": 6735
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1965131187862261,
      "learning_rate": 5.536464980813418e-06,
      "loss": 0.157,
      "step": 6736
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.31904448568832,
      "learning_rate": 5.535313249007878e-06,
      "loss": 0.1714,
      "step": 6737
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3645902524097202,
      "learning_rate": 5.5341614884688475e-06,
      "loss": 0.2174,
      "step": 6738
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4646830743059327,
      "learning_rate": 5.533009699258152e-06,
      "loss": 0.2073,
      "step": 6739
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4699739579948878,
      "learning_rate": 5.531857881437612e-06,
      "loss": 0.1898,
      "step": 6740
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.049197321903939,
      "learning_rate": 5.5307060350690554e-06,
      "loss": 0.3951,
      "step": 6741
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.387503532780477,
      "learning_rate": 5.529554160214306e-06,
      "loss": 0.1745,
      "step": 6742
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3764467731007737,
      "learning_rate": 5.528402256935194e-06,
      "loss": 0.2002,
      "step": 6743
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.136569721889636,
      "learning_rate": 5.5272503252935474e-06,
      "loss": 0.1671,
      "step": 6744
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.253399986912368,
      "learning_rate": 5.526098365351198e-06,
      "loss": 0.1576,
      "step": 6745
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.835596509172331,
      "learning_rate": 5.524946377169979e-06,
      "loss": 0.6145,
      "step": 6746
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2477404081621708,
      "learning_rate": 5.523794360811723e-06,
      "loss": 0.1937,
      "step": 6747
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3502021143467564,
      "learning_rate": 5.522642316338268e-06,
      "loss": 0.1722,
      "step": 6748
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1249319624186798,
      "learning_rate": 5.521490243811449e-06,
      "loss": 0.1437,
      "step": 6749
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4134308181069786,
      "learning_rate": 5.520338143293106e-06,
      "loss": 0.1702,
      "step": 6750
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5058031024348788,
      "learning_rate": 5.519186014845077e-06,
      "loss": 0.2021,
      "step": 6751
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3451481541968922,
      "learning_rate": 5.518033858529207e-06,
      "loss": 0.1839,
      "step": 6752
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2538438229051385,
      "learning_rate": 5.516881674407337e-06,
      "loss": 0.1942,
      "step": 6753
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1579813727682793,
      "learning_rate": 5.515729462541312e-06,
      "loss": 0.158,
      "step": 6754
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.331516309303722,
      "learning_rate": 5.51457722299298e-06,
      "loss": 0.1993,
      "step": 6755
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.279591499033453,
      "learning_rate": 5.513424955824185e-06,
      "loss": 0.1928,
      "step": 6756
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1928069592518773,
      "learning_rate": 5.512272661096779e-06,
      "loss": 0.1674,
      "step": 6757
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4181095732284703,
      "learning_rate": 5.511120338872611e-06,
      "loss": 0.2564,
      "step": 6758
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.310888891244863,
      "learning_rate": 5.509967989213535e-06,
      "loss": 0.1885,
      "step": 6759
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3813179269948987,
      "learning_rate": 5.508815612181401e-06,
      "loss": 0.1641,
      "step": 6760
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5016381790547977,
      "learning_rate": 5.507663207838069e-06,
      "loss": 0.2606,
      "step": 6761
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.564429076716475,
      "learning_rate": 5.506510776245392e-06,
      "loss": 0.2573,
      "step": 6762
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.469400090289019,
      "learning_rate": 5.505358317465228e-06,
      "loss": 0.2177,
      "step": 6763
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5118141701378751,
      "learning_rate": 5.504205831559437e-06,
      "loss": 0.2163,
      "step": 6764
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.337947196177064,
      "learning_rate": 5.5030533185898815e-06,
      "loss": 0.1411,
      "step": 6765
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3890752819568728,
      "learning_rate": 5.501900778618422e-06,
      "loss": 0.1775,
      "step": 6766
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.290413323372114,
      "learning_rate": 5.500748211706922e-06,
      "loss": 0.1754,
      "step": 6767
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2812686906137314,
      "learning_rate": 5.499595617917247e-06,
      "loss": 0.1961,
      "step": 6768
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3361976560245312,
      "learning_rate": 5.498442997311264e-06,
      "loss": 0.1819,
      "step": 6769
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4551005583589307,
      "learning_rate": 5.497290349950842e-06,
      "loss": 0.251,
      "step": 6770
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1775914443063273,
      "learning_rate": 5.49613767589785e-06,
      "loss": 0.1828,
      "step": 6771
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3957315529181746,
      "learning_rate": 5.494984975214158e-06,
      "loss": 0.2393,
      "step": 6772
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2548407835443514,
      "learning_rate": 5.493832247961638e-06,
      "loss": 0.1992,
      "step": 6773
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1888279045100618,
      "learning_rate": 5.492679494202167e-06,
      "loss": 0.1977,
      "step": 6774
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.229916681884861,
      "learning_rate": 5.491526713997617e-06,
      "loss": 0.1912,
      "step": 6775
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.483293591571265,
      "learning_rate": 5.490373907409866e-06,
      "loss": 0.2101,
      "step": 6776
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3174297839789983,
      "learning_rate": 5.48922107450079e-06,
      "loss": 0.1762,
      "step": 6777
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4859410262083592,
      "learning_rate": 5.488068215332273e-06,
      "loss": 0.2503,
      "step": 6778
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3900691398471186,
      "learning_rate": 5.486915329966192e-06,
      "loss": 0.2292,
      "step": 6779
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.338193212251261,
      "learning_rate": 5.48576241846443e-06,
      "loss": 0.2071,
      "step": 6780
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3424836903990778,
      "learning_rate": 5.484609480888872e-06,
      "loss": 0.2058,
      "step": 6781
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3353165304076882,
      "learning_rate": 5.483456517301401e-06,
      "loss": 0.193,
      "step": 6782
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.281846483294925,
      "learning_rate": 5.482303527763906e-06,
      "loss": 0.1815,
      "step": 6783
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5155167360178363,
      "learning_rate": 5.481150512338275e-06,
      "loss": 0.1975,
      "step": 6784
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2260338966621829,
      "learning_rate": 5.479997471086394e-06,
      "loss": 0.1657,
      "step": 6785
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2471540592974437,
      "learning_rate": 5.478844404070155e-06,
      "loss": 0.2368,
      "step": 6786
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1602960211615858,
      "learning_rate": 5.477691311351453e-06,
      "loss": 0.1769,
      "step": 6787
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5165083929741874,
      "learning_rate": 5.476538192992178e-06,
      "loss": 0.2244,
      "step": 6788
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3051802629646676,
      "learning_rate": 5.475385049054225e-06,
      "loss": 0.1849,
      "step": 6789
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2823513404076694,
      "learning_rate": 5.474231879599493e-06,
      "loss": 0.2212,
      "step": 6790
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3062268530636707,
      "learning_rate": 5.4730786846898765e-06,
      "loss": 0.1611,
      "step": 6791
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.351821537968671,
      "learning_rate": 5.471925464387276e-06,
      "loss": 0.567,
      "step": 6792
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3518153230617713,
      "learning_rate": 5.470772218753592e-06,
      "loss": 0.1883,
      "step": 6793
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.417485558053931,
      "learning_rate": 5.4696189478507255e-06,
      "loss": 0.2116,
      "step": 6794
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3634475709227498,
      "learning_rate": 5.468465651740579e-06,
      "loss": 0.1773,
      "step": 6795
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2240665757903137,
      "learning_rate": 5.467312330485058e-06,
      "loss": 0.1618,
      "step": 6796
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8658802132960561,
      "learning_rate": 5.466158984146066e-06,
      "loss": 0.1846,
      "step": 6797
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4045855747908262,
      "learning_rate": 5.465005612785514e-06,
      "loss": 0.1969,
      "step": 6798
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.155989318442908,
      "learning_rate": 5.463852216465307e-06,
      "loss": 0.1877,
      "step": 6799
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3539566726968721,
      "learning_rate": 5.462698795247357e-06,
      "loss": 0.2181,
      "step": 6800
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.748510638540412,
      "learning_rate": 5.461545349193571e-06,
      "loss": 0.5437,
      "step": 6801
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3186992652514298,
      "learning_rate": 5.460391878365865e-06,
      "loss": 0.1642,
      "step": 6802
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2401163653155394,
      "learning_rate": 5.4592383828261534e-06,
      "loss": 0.1781,
      "step": 6803
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2487002072660474,
      "learning_rate": 5.45808486263635e-06,
      "loss": 0.1845,
      "step": 6804
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3288613206864088,
      "learning_rate": 5.456931317858371e-06,
      "loss": 0.1541,
      "step": 6805
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2787722773924046,
      "learning_rate": 5.455777748554134e-06,
      "loss": 0.158,
      "step": 6806
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.193116551333469,
      "learning_rate": 5.454624154785557e-06,
      "loss": 0.1942,
      "step": 6807
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3163562959430897,
      "learning_rate": 5.4534705366145624e-06,
      "loss": 0.2113,
      "step": 6808
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3153432053486518,
      "learning_rate": 5.452316894103071e-06,
      "loss": 0.1727,
      "step": 6809
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.166680895264921,
      "learning_rate": 5.451163227313006e-06,
      "loss": 0.6397,
      "step": 6810
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2726459140397102,
      "learning_rate": 5.45000953630629e-06,
      "loss": 0.1813,
      "step": 6811
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3199381108460457,
      "learning_rate": 5.448855821144851e-06,
      "loss": 0.1934,
      "step": 6812
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2631007381920603,
      "learning_rate": 5.447702081890614e-06,
      "loss": 0.2148,
      "step": 6813
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2805316866182044,
      "learning_rate": 5.446548318605509e-06,
      "loss": 0.1934,
      "step": 6814
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6003247685395632,
      "learning_rate": 5.445394531351463e-06,
      "loss": 0.2113,
      "step": 6815
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0813379121308828,
      "learning_rate": 5.444240720190409e-06,
      "loss": 0.1212,
      "step": 6816
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1900432749014551,
      "learning_rate": 5.443086885184279e-06,
      "loss": 0.1821,
      "step": 6817
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.852085547568391,
      "learning_rate": 5.441933026395003e-06,
      "loss": 0.6501,
      "step": 6818
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2468105134590544,
      "learning_rate": 5.440779143884518e-06,
      "loss": 0.1545,
      "step": 6819
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4776502847286042,
      "learning_rate": 5.4396252377147615e-06,
      "loss": 0.236,
      "step": 6820
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5009812213307152,
      "learning_rate": 5.438471307947668e-06,
      "loss": 0.225,
      "step": 6821
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4038094267466799,
      "learning_rate": 5.437317354645177e-06,
      "loss": 0.1819,
      "step": 6822
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1683216835146955,
      "learning_rate": 5.436163377869225e-06,
      "loss": 0.134,
      "step": 6823
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2999795014448476,
      "learning_rate": 5.435009377681759e-06,
      "loss": 0.1737,
      "step": 6824
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3251914868544477,
      "learning_rate": 5.433855354144717e-06,
      "loss": 0.2024,
      "step": 6825
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.426699190203773,
      "learning_rate": 5.432701307320043e-06,
      "loss": 0.4185,
      "step": 6826
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.335181480267974,
      "learning_rate": 5.431547237269681e-06,
      "loss": 0.2167,
      "step": 6827
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.912729742178636,
      "learning_rate": 5.430393144055579e-06,
      "loss": 0.4096,
      "step": 6828
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.996352912267826,
      "learning_rate": 5.429239027739683e-06,
      "loss": 0.4474,
      "step": 6829
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2818577794654593,
      "learning_rate": 5.428084888383941e-06,
      "loss": 0.1916,
      "step": 6830
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.368241629993963,
      "learning_rate": 5.426930726050301e-06,
      "loss": 0.1853,
      "step": 6831
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4627782578378192,
      "learning_rate": 5.425776540800718e-06,
      "loss": 0.2162,
      "step": 6832
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4066559397733949,
      "learning_rate": 5.42462233269714e-06,
      "loss": 0.2014,
      "step": 6833
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4254521665605342,
      "learning_rate": 5.423468101801524e-06,
      "loss": 0.2122,
      "step": 6834
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4675478161898972,
      "learning_rate": 5.422313848175822e-06,
      "loss": 0.2148,
      "step": 6835
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1566236627923878,
      "learning_rate": 5.42115957188199e-06,
      "loss": 0.1408,
      "step": 6836
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3084436902699248,
      "learning_rate": 5.420005272981985e-06,
      "loss": 0.1894,
      "step": 6837
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.459146106449613,
      "learning_rate": 5.418850951537766e-06,
      "loss": 0.1978,
      "step": 6838
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1352391865180778,
      "learning_rate": 5.417696607611293e-06,
      "loss": 0.1332,
      "step": 6839
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3241699706533734,
      "learning_rate": 5.416542241264524e-06,
      "loss": 0.1756,
      "step": 6840
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4338074572884234,
      "learning_rate": 5.415387852559423e-06,
      "loss": 0.2379,
      "step": 6841
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4625409549017454,
      "learning_rate": 5.414233441557952e-06,
      "loss": 0.231,
      "step": 6842
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.821278361849323,
      "learning_rate": 5.413079008322077e-06,
      "loss": 0.5212,
      "step": 6843
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.394559279673393,
      "learning_rate": 5.41192455291376e-06,
      "loss": 0.2031,
      "step": 6844
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3980311142225494,
      "learning_rate": 5.41077007539497e-06,
      "loss": 0.1725,
      "step": 6845
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4455303301307307,
      "learning_rate": 5.409615575827675e-06,
      "loss": 0.1838,
      "step": 6846
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3000556821255083,
      "learning_rate": 5.408461054273843e-06,
      "loss": 0.1493,
      "step": 6847
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.262575224506312,
      "learning_rate": 5.407306510795443e-06,
      "loss": 0.1784,
      "step": 6848
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.323117468141364,
      "learning_rate": 5.406151945454449e-06,
      "loss": 0.1752,
      "step": 6849
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2848735054760394,
      "learning_rate": 5.404997358312831e-06,
      "loss": 0.1995,
      "step": 6850
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.492760115222576,
      "learning_rate": 5.403842749432565e-06,
      "loss": 0.1844,
      "step": 6851
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4224989711900202,
      "learning_rate": 5.402688118875624e-06,
      "loss": 0.1973,
      "step": 6852
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.352826740035777,
      "learning_rate": 5.401533466703984e-06,
      "loss": 0.206,
      "step": 6853
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.360718796004018,
      "learning_rate": 5.400378792979623e-06,
      "loss": 0.223,
      "step": 6854
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2441070913734704,
      "learning_rate": 5.3992240977645205e-06,
      "loss": 0.1958,
      "step": 6855
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3684218888807818,
      "learning_rate": 5.398069381120654e-06,
      "loss": 0.2145,
      "step": 6856
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.9437922931801097,
      "learning_rate": 5.396914643110003e-06,
      "loss": 0.5525,
      "step": 6857
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4578418294020825,
      "learning_rate": 5.395759883794552e-06,
      "loss": 0.2529,
      "step": 6858
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3429557470919053,
      "learning_rate": 5.394605103236282e-06,
      "loss": 0.2009,
      "step": 6859
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3486119063612603,
      "learning_rate": 5.39345030149718e-06,
      "loss": 0.194,
      "step": 6860
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2981974073177407,
      "learning_rate": 5.392295478639226e-06,
      "loss": 0.1854,
      "step": 6861
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1880723233403436,
      "learning_rate": 5.39114063472441e-06,
      "loss": 0.1407,
      "step": 6862
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.184610553066995,
      "learning_rate": 5.38998576981472e-06,
      "loss": 0.2021,
      "step": 6863
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.850625546616521,
      "learning_rate": 5.388830883972142e-06,
      "loss": 0.6059,
      "step": 6864
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2536190233682296,
      "learning_rate": 5.387675977258667e-06,
      "loss": 0.1758,
      "step": 6865
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3615340293904863,
      "learning_rate": 5.386521049736284e-06,
      "loss": 0.1969,
      "step": 6866
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2418907660541736,
      "learning_rate": 5.385366101466989e-06,
      "loss": 0.1481,
      "step": 6867
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.199297659947035,
      "learning_rate": 5.384211132512772e-06,
      "loss": 0.2011,
      "step": 6868
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3400687511898048,
      "learning_rate": 5.383056142935627e-06,
      "loss": 0.2109,
      "step": 6869
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3918692645164261,
      "learning_rate": 5.381901132797551e-06,
      "loss": 0.2039,
      "step": 6870
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.694929636188387,
      "learning_rate": 5.3807461021605404e-06,
      "loss": 0.5441,
      "step": 6871
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4078895202731236,
      "learning_rate": 5.3795910510865904e-06,
      "loss": 0.2142,
      "step": 6872
    },
    {
      "epoch": 0.49,
      "grad_norm": 3.065199216341531,
      "learning_rate": 5.378435979637701e-06,
      "loss": 0.3409,
      "step": 6873
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1539222015472128,
      "learning_rate": 5.377280887875871e-06,
      "loss": 0.1562,
      "step": 6874
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.592742425504455,
      "learning_rate": 5.376125775863102e-06,
      "loss": 0.4414,
      "step": 6875
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.344416840980456,
      "learning_rate": 5.374970643661397e-06,
      "loss": 0.1899,
      "step": 6876
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.725309334488556,
      "learning_rate": 5.3738154913327566e-06,
      "loss": 0.6448,
      "step": 6877
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.204632970642909,
      "learning_rate": 5.372660318939185e-06,
      "loss": 0.2217,
      "step": 6878
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4521894238228432,
      "learning_rate": 5.371505126542688e-06,
      "loss": 0.2271,
      "step": 6879
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3690083233734283,
      "learning_rate": 5.370349914205273e-06,
      "loss": 0.2101,
      "step": 6880
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2626558170234456,
      "learning_rate": 5.369194681988946e-06,
      "loss": 0.2326,
      "step": 6881
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3904146465182194,
      "learning_rate": 5.368039429955714e-06,
      "loss": 0.2108,
      "step": 6882
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.752489433358283,
      "learning_rate": 5.366884158167588e-06,
      "loss": 0.746,
      "step": 6883
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3799612514551323,
      "learning_rate": 5.365728866686578e-06,
      "loss": 0.1798,
      "step": 6884
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2959051595514177,
      "learning_rate": 5.364573555574696e-06,
      "loss": 0.159,
      "step": 6885
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4767672741729594,
      "learning_rate": 5.363418224893953e-06,
      "loss": 0.2122,
      "step": 6886
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3577137054221575,
      "learning_rate": 5.362262874706363e-06,
      "loss": 0.2116,
      "step": 6887
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5070744292955682,
      "learning_rate": 5.361107505073942e-06,
      "loss": 0.1797,
      "step": 6888
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3814161611560107,
      "learning_rate": 5.359952116058705e-06,
      "loss": 0.2478,
      "step": 6889
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2749135343725408,
      "learning_rate": 5.358796707722668e-06,
      "loss": 0.1836,
      "step": 6890
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.290265885306727,
      "learning_rate": 5.357641280127849e-06,
      "loss": 0.1746,
      "step": 6891
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3955214226735753,
      "learning_rate": 5.356485833336267e-06,
      "loss": 0.1964,
      "step": 6892
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3305364451393777,
      "learning_rate": 5.355330367409942e-06,
      "loss": 0.2232,
      "step": 6893
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2208846958712158,
      "learning_rate": 5.3541748824108945e-06,
      "loss": 0.1893,
      "step": 6894
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3603845632106566,
      "learning_rate": 5.3530193784011455e-06,
      "loss": 0.2293,
      "step": 6895
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.800738808515453,
      "learning_rate": 5.351863855442719e-06,
      "loss": 0.7261,
      "step": 6896
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2834949531715418,
      "learning_rate": 5.350708313597639e-06,
      "loss": 0.1975,
      "step": 6897
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3520939986193152,
      "learning_rate": 5.34955275292793e-06,
      "loss": 0.1537,
      "step": 6898
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3723173599808838,
      "learning_rate": 5.348397173495618e-06,
      "loss": 0.1846,
      "step": 6899
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1476418856636816,
      "learning_rate": 5.347241575362729e-06,
      "loss": 0.1404,
      "step": 6900
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2702689076544331,
      "learning_rate": 5.346085958591294e-06,
      "loss": 0.1821,
      "step": 6901
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3500271506732693,
      "learning_rate": 5.3449303232433384e-06,
      "loss": 0.2059,
      "step": 6902
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2900776653928712,
      "learning_rate": 5.343774669380894e-06,
      "loss": 0.1758,
      "step": 6903
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1916150179732805,
      "learning_rate": 5.342618997065988e-06,
      "loss": 0.1657,
      "step": 6904
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.484467974452162,
      "learning_rate": 5.341463306360658e-06,
      "loss": 0.6843,
      "step": 6905
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4559229356402914,
      "learning_rate": 5.340307597326935e-06,
      "loss": 0.2419,
      "step": 6906
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4558134333716426,
      "learning_rate": 5.339151870026851e-06,
      "loss": 0.2036,
      "step": 6907
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3192070564716802,
      "learning_rate": 5.337996124522441e-06,
      "loss": 0.1911,
      "step": 6908
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4431564962778856,
      "learning_rate": 5.336840360875743e-06,
      "loss": 0.191,
      "step": 6909
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.4956812204791055,
      "learning_rate": 5.335684579148793e-06,
      "loss": 0.6524,
      "step": 6910
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3257691893566066,
      "learning_rate": 5.334528779403628e-06,
      "loss": 0.2044,
      "step": 6911
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3651487555684232,
      "learning_rate": 5.333372961702287e-06,
      "loss": 0.2291,
      "step": 6912
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3676981633956835,
      "learning_rate": 5.33221712610681e-06,
      "loss": 0.2122,
      "step": 6913
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3421180363013006,
      "learning_rate": 5.331061272679238e-06,
      "loss": 0.1786,
      "step": 6914
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3973527291268202,
      "learning_rate": 5.329905401481612e-06,
      "loss": 0.2173,
      "step": 6915
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.396112863515519,
      "learning_rate": 5.3287495125759734e-06,
      "loss": 0.1914,
      "step": 6916
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4514155692232678,
      "learning_rate": 5.327593606024369e-06,
      "loss": 0.2066,
      "step": 6917
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.128054667522911,
      "learning_rate": 5.32643768188884e-06,
      "loss": 0.1616,
      "step": 6918
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.299898500672018,
      "learning_rate": 5.325281740231433e-06,
      "loss": 0.2013,
      "step": 6919
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.272469977229793,
      "learning_rate": 5.324125781114193e-06,
      "loss": 0.5872,
      "step": 6920
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2542762514476613,
      "learning_rate": 5.3229698045991715e-06,
      "loss": 0.203,
      "step": 6921
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.170224660158486,
      "learning_rate": 5.321813810748413e-06,
      "loss": 0.5297,
      "step": 6922
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3475238184530494,
      "learning_rate": 5.320657799623968e-06,
      "loss": 0.1883,
      "step": 6923
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1954081109023742,
      "learning_rate": 5.319501771287885e-06,
      "loss": 0.1599,
      "step": 6924
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3056892781408058,
      "learning_rate": 5.318345725802217e-06,
      "loss": 0.1767,
      "step": 6925
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.666525736250797,
      "learning_rate": 5.317189663229016e-06,
      "loss": 0.5874,
      "step": 6926
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5332223871057418,
      "learning_rate": 5.316033583630333e-06,
      "loss": 0.214,
      "step": 6927
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5811909891696048,
      "learning_rate": 5.314877487068223e-06,
      "loss": 0.2209,
      "step": 6928
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.30206773777624,
      "learning_rate": 5.313721373604739e-06,
      "loss": 0.5329,
      "step": 6929
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.105534250946762,
      "learning_rate": 5.312565243301939e-06,
      "loss": 0.6328,
      "step": 6930
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3536912276683108,
      "learning_rate": 5.31140909622188e-06,
      "loss": 0.1913,
      "step": 6931
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1950031128266467,
      "learning_rate": 5.310252932426617e-06,
      "loss": 0.1731,
      "step": 6932
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.498715990170105,
      "learning_rate": 5.309096751978207e-06,
      "loss": 0.2314,
      "step": 6933
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.260478479196862,
      "learning_rate": 5.307940554938715e-06,
      "loss": 0.2002,
      "step": 6934
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.336780929062254,
      "learning_rate": 5.306784341370196e-06,
      "loss": 0.1989,
      "step": 6935
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.0520937222232885,
      "learning_rate": 5.3056281113347115e-06,
      "loss": 0.3559,
      "step": 6936
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3642359387185994,
      "learning_rate": 5.304471864894324e-06,
      "loss": 0.2,
      "step": 6937
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2210056089928485,
      "learning_rate": 5.303315602111098e-06,
      "loss": 0.1728,
      "step": 6938
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.384300097852221,
      "learning_rate": 5.302159323047095e-06,
      "loss": 0.1813,
      "step": 6939
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3997601349736764,
      "learning_rate": 5.30100302776438e-06,
      "loss": 0.2112,
      "step": 6940
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.301123689449632,
      "learning_rate": 5.299846716325017e-06,
      "loss": 0.1593,
      "step": 6941
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1918243835388032,
      "learning_rate": 5.298690388791075e-06,
      "loss": 0.1741,
      "step": 6942
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.133828752410014,
      "learning_rate": 5.297534045224618e-06,
      "loss": 0.1802,
      "step": 6943
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.219529933202826,
      "learning_rate": 5.296377685687717e-06,
      "loss": 0.1675,
      "step": 6944
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.312904182970714,
      "learning_rate": 5.295221310242438e-06,
      "loss": 0.1953,
      "step": 6945
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2828822917969591,
      "learning_rate": 5.294064918950852e-06,
      "loss": 0.1437,
      "step": 6946
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1921141517366456,
      "learning_rate": 5.2929085118750314e-06,
      "loss": 0.167,
      "step": 6947
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6084642096365014,
      "learning_rate": 5.291752089077044e-06,
      "loss": 0.2363,
      "step": 6948
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3404605033511738,
      "learning_rate": 5.2905956506189624e-06,
      "loss": 0.2265,
      "step": 6949
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1686624605572,
      "learning_rate": 5.289439196562862e-06,
      "loss": 0.1546,
      "step": 6950
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2927540260027461,
      "learning_rate": 5.288282726970816e-06,
      "loss": 0.1807,
      "step": 6951
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3429634620110584,
      "learning_rate": 5.287126241904898e-06,
      "loss": 0.1805,
      "step": 6952
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.099768949498347,
      "learning_rate": 5.285969741427185e-06,
      "loss": 0.1556,
      "step": 6953
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.215707948192235,
      "learning_rate": 5.28481322559975e-06,
      "loss": 0.4682,
      "step": 6954
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4053041347325612,
      "learning_rate": 5.2836566944846745e-06,
      "loss": 0.1862,
      "step": 6955
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2802057930304818,
      "learning_rate": 5.2825001481440345e-06,
      "loss": 0.1709,
      "step": 6956
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2007798873765443,
      "learning_rate": 5.281343586639908e-06,
      "loss": 0.1832,
      "step": 6957
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.087866708648857,
      "learning_rate": 5.280187010034375e-06,
      "loss": 0.2304,
      "step": 6958
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1988287776468347,
      "learning_rate": 5.279030418389517e-06,
      "loss": 0.1647,
      "step": 6959
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.383176448583845,
      "learning_rate": 5.277873811767415e-06,
      "loss": 0.1515,
      "step": 6960
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3608641872017129,
      "learning_rate": 5.27671719023015e-06,
      "loss": 0.1662,
      "step": 6961
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3549062675690828,
      "learning_rate": 5.275560553839807e-06,
      "loss": 0.2089,
      "step": 6962
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3829462886193433,
      "learning_rate": 5.274403902658466e-06,
      "loss": 0.1884,
      "step": 6963
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2074995673104718,
      "learning_rate": 5.273247236748216e-06,
      "loss": 0.1277,
      "step": 6964
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.871830434045196,
      "learning_rate": 5.272090556171139e-06,
      "loss": 0.5336,
      "step": 6965
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.6855446302255834,
      "learning_rate": 5.270933860989322e-06,
      "loss": 0.7287,
      "step": 6966
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3946988138142102,
      "learning_rate": 5.269777151264851e-06,
      "loss": 0.2073,
      "step": 6967
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.357318475599064,
      "learning_rate": 5.268620427059815e-06,
      "loss": 0.1827,
      "step": 6968
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4805490001588364,
      "learning_rate": 5.267463688436302e-06,
      "loss": 0.2241,
      "step": 6969
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.201551357368844,
      "learning_rate": 5.2663069354564e-06,
      "loss": 0.1711,
      "step": 6970
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4327573573096604,
      "learning_rate": 5.265150168182199e-06,
      "loss": 0.1994,
      "step": 6971
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2832779482491845,
      "learning_rate": 5.263993386675792e-06,
      "loss": 0.2089,
      "step": 6972
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3975638935301886,
      "learning_rate": 5.262836590999267e-06,
      "loss": 0.2297,
      "step": 6973
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4179187780677898,
      "learning_rate": 5.2616797812147205e-06,
      "loss": 0.2539,
      "step": 6974
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.398053235881588,
      "learning_rate": 5.260522957384241e-06,
      "loss": 0.5396,
      "step": 6975
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2574885714448814,
      "learning_rate": 5.259366119569924e-06,
      "loss": 0.1864,
      "step": 6976
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.737646301992324,
      "learning_rate": 5.2582092678338656e-06,
      "loss": 0.6419,
      "step": 6977
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1046360742955015,
      "learning_rate": 5.2570524022381585e-06,
      "loss": 0.1492,
      "step": 6978
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.398534291552053,
      "learning_rate": 5.2558955228449e-06,
      "loss": 0.1754,
      "step": 6979
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.352348505318937,
      "learning_rate": 5.254738629716186e-06,
      "loss": 0.194,
      "step": 6980
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.319624161414183,
      "learning_rate": 5.253581722914115e-06,
      "loss": 0.1629,
      "step": 6981
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.373644358565873,
      "learning_rate": 5.252424802500784e-06,
      "loss": 0.1765,
      "step": 6982
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3272613268584128,
      "learning_rate": 5.251267868538292e-06,
      "loss": 0.1958,
      "step": 6983
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.344822600053962,
      "learning_rate": 5.25011092108874e-06,
      "loss": 0.1823,
      "step": 6984
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.37170821096451,
      "learning_rate": 5.248953960214227e-06,
      "loss": 0.2124,
      "step": 6985
    },
    {
      "epoch": 0.5,
      "grad_norm": 19.26135577532696,
      "learning_rate": 5.2477969859768555e-06,
      "loss": 0.5871,
      "step": 6986
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3361000936955896,
      "learning_rate": 5.246639998438724e-06,
      "loss": 0.186,
      "step": 6987
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3360560542393998,
      "learning_rate": 5.245482997661939e-06,
      "loss": 0.1855,
      "step": 6988
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4722810098348993,
      "learning_rate": 5.244325983708601e-06,
      "loss": 0.2202,
      "step": 6989
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3993748967527266,
      "learning_rate": 5.2431689566408165e-06,
      "loss": 0.2029,
      "step": 6990
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3615873275828976,
      "learning_rate": 5.242011916520686e-06,
      "loss": 0.1852,
      "step": 6991
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2744668123237122,
      "learning_rate": 5.240854863410319e-06,
      "loss": 0.1577,
      "step": 6992
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4264109569810484,
      "learning_rate": 5.239697797371819e-06,
      "loss": 0.1703,
      "step": 6993
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2506961273297414,
      "learning_rate": 5.238540718467294e-06,
      "loss": 0.1703,
      "step": 6994
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.265950406133173,
      "learning_rate": 5.237383626758851e-06,
      "loss": 0.191,
      "step": 6995
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2953092896525331,
      "learning_rate": 5.236226522308596e-06,
      "loss": 0.1572,
      "step": 6996
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1351216479714892,
      "learning_rate": 5.235069405178642e-06,
      "loss": 0.1433,
      "step": 6997
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2207629120414003,
      "learning_rate": 5.233912275431096e-06,
      "loss": 0.2124,
      "step": 6998
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3174142058167035,
      "learning_rate": 5.232755133128069e-06,
      "loss": 0.1772,
      "step": 6999
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5438121558763236,
      "learning_rate": 5.231597978331669e-06,
      "loss": 0.2281,
      "step": 7000
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2985162404889248,
      "learning_rate": 5.230440811104012e-06,
      "loss": 0.1713,
      "step": 7001
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2827631169196971,
      "learning_rate": 5.229283631507206e-06,
      "loss": 0.2186,
      "step": 7002
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1565063249181857,
      "learning_rate": 5.228126439603367e-06,
      "loss": 0.1378,
      "step": 7003
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1641651947723026,
      "learning_rate": 5.226969235454606e-06,
      "loss": 0.1754,
      "step": 7004
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3260369658131836,
      "learning_rate": 5.225812019123038e-06,
      "loss": 0.2013,
      "step": 7005
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3861039585408548,
      "learning_rate": 5.224654790670778e-06,
      "loss": 0.1968,
      "step": 7006
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2949928363136323,
      "learning_rate": 5.223497550159942e-06,
      "loss": 0.1988,
      "step": 7007
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2485435084200702,
      "learning_rate": 5.222340297652645e-06,
      "loss": 0.2012,
      "step": 7008
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3913081818240074,
      "learning_rate": 5.221183033211003e-06,
      "loss": 0.2029,
      "step": 7009
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3168410976478224,
      "learning_rate": 5.220025756897136e-06,
      "loss": 0.1806,
      "step": 7010
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.401583369278531,
      "learning_rate": 5.21886846877316e-06,
      "loss": 0.1883,
      "step": 7011
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.317320049903733,
      "learning_rate": 5.217711168901195e-06,
      "loss": 0.2076,
      "step": 7012
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4421780401820896,
      "learning_rate": 5.216553857343358e-06,
      "loss": 0.1944,
      "step": 7013
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.635311612326544,
      "learning_rate": 5.215396534161771e-06,
      "loss": 0.2123,
      "step": 7014
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.516834044415225,
      "learning_rate": 5.214239199418554e-06,
      "loss": 0.2019,
      "step": 7015
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4399606499390736,
      "learning_rate": 5.213081853175827e-06,
      "loss": 0.2138,
      "step": 7016
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1339955683637923,
      "learning_rate": 5.211924495495713e-06,
      "loss": 0.1839,
      "step": 7017
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2956875074473657,
      "learning_rate": 5.210767126440335e-06,
      "loss": 0.1628,
      "step": 7018
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1934055159897403,
      "learning_rate": 5.209609746071813e-06,
      "loss": 0.2093,
      "step": 7019
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1273795600006267,
      "learning_rate": 5.208452354452275e-06,
      "loss": 0.1267,
      "step": 7020
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5470143033170909,
      "learning_rate": 5.20729495164384e-06,
      "loss": 0.2313,
      "step": 7021
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4267961746927944,
      "learning_rate": 5.206137537708637e-06,
      "loss": 0.2218,
      "step": 7022
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2137288029382431,
      "learning_rate": 5.20498011270879e-06,
      "loss": 0.2052,
      "step": 7023
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2822849107152563,
      "learning_rate": 5.203822676706426e-06,
      "loss": 0.1873,
      "step": 7024
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3115825239060672,
      "learning_rate": 5.202665229763668e-06,
      "loss": 0.2035,
      "step": 7025
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4396129248174863,
      "learning_rate": 5.2015077719426465e-06,
      "loss": 0.1905,
      "step": 7026
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4868518646980733,
      "learning_rate": 5.200350303305489e-06,
      "loss": 0.2282,
      "step": 7027
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2709610230821584,
      "learning_rate": 5.199192823914323e-06,
      "loss": 0.1561,
      "step": 7028
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.339942110143628,
      "learning_rate": 5.1980353338312775e-06,
      "loss": 0.1489,
      "step": 7029
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2329534340646224,
      "learning_rate": 5.196877833118481e-06,
      "loss": 0.1856,
      "step": 7030
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1645231827208187,
      "learning_rate": 5.195720321838066e-06,
      "loss": 0.1459,
      "step": 7031
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.764728955111807,
      "learning_rate": 5.194562800052162e-06,
      "loss": 0.4393,
      "step": 7032
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3027468360760934,
      "learning_rate": 5.1934052678229e-06,
      "loss": 0.1875,
      "step": 7033
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3435743058686787,
      "learning_rate": 5.192247725212409e-06,
      "loss": 0.2083,
      "step": 7034
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.335528018493987,
      "learning_rate": 5.191090172282826e-06,
      "loss": 0.1683,
      "step": 7035
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2898255348556065,
      "learning_rate": 5.189932609096281e-06,
      "loss": 0.206,
      "step": 7036
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.490030857060649,
      "learning_rate": 5.188775035714909e-06,
      "loss": 0.1855,
      "step": 7037
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.300653683151527,
      "learning_rate": 5.187617452200842e-06,
      "loss": 0.1765,
      "step": 7038
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.874356721541645,
      "learning_rate": 5.1864598586162166e-06,
      "loss": 0.5751,
      "step": 7039
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3160138141432036,
      "learning_rate": 5.185302255023166e-06,
      "loss": 0.1784,
      "step": 7040
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4020593781714124,
      "learning_rate": 5.1841446414838274e-06,
      "loss": 0.1772,
      "step": 7041
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3068384059456177,
      "learning_rate": 5.182987018060334e-06,
      "loss": 0.1972,
      "step": 7042
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4934326714937047,
      "learning_rate": 5.181829384814826e-06,
      "loss": 0.2209,
      "step": 7043
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3126916178357773,
      "learning_rate": 5.180671741809439e-06,
      "loss": 0.1889,
      "step": 7044
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1300865424387185,
      "learning_rate": 5.17951408910631e-06,
      "loss": 0.1489,
      "step": 7045
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2875430695330707,
      "learning_rate": 5.178356426767578e-06,
      "loss": 0.2111,
      "step": 7046
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2520092563962444,
      "learning_rate": 5.177198754855382e-06,
      "loss": 0.2065,
      "step": 7047
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.269322156897904,
      "learning_rate": 5.17604107343186e-06,
      "loss": 0.2094,
      "step": 7048
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2796420565569904,
      "learning_rate": 5.174883382559154e-06,
      "loss": 0.1892,
      "step": 7049
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2834589851512133,
      "learning_rate": 5.173725682299402e-06,
      "loss": 0.2245,
      "step": 7050
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3795351525805337,
      "learning_rate": 5.172567972714745e-06,
      "loss": 0.1888,
      "step": 7051
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.866081714136038,
      "learning_rate": 5.171410253867325e-06,
      "loss": 0.4929,
      "step": 7052
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.641775236051953,
      "learning_rate": 5.170252525819285e-06,
      "loss": 0.2264,
      "step": 7053
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.25455871205736,
      "learning_rate": 5.1690947886327645e-06,
      "loss": 0.1686,
      "step": 7054
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5455810236713847,
      "learning_rate": 5.167937042369907e-06,
      "loss": 0.1882,
      "step": 7055
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4186789268967792,
      "learning_rate": 5.166779287092857e-06,
      "loss": 0.161,
      "step": 7056
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3049888042387654,
      "learning_rate": 5.165621522863756e-06,
      "loss": 0.1994,
      "step": 7057
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3146143827330297,
      "learning_rate": 5.164463749744751e-06,
      "loss": 0.2199,
      "step": 7058
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3194439567219534,
      "learning_rate": 5.163305967797986e-06,
      "loss": 0.1889,
      "step": 7059
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.343740174539173,
      "learning_rate": 5.162148177085604e-06,
      "loss": 0.1947,
      "step": 7060
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.304934655096607,
      "learning_rate": 5.1609903776697535e-06,
      "loss": 0.2038,
      "step": 7061
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1889128165698772,
      "learning_rate": 5.1598325696125785e-06,
      "loss": 0.1801,
      "step": 7062
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5597980706662804,
      "learning_rate": 5.158674752976226e-06,
      "loss": 0.2298,
      "step": 7063
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2082773574681906,
      "learning_rate": 5.157516927822842e-06,
      "loss": 0.1834,
      "step": 7064
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3492537362478436,
      "learning_rate": 5.156359094214577e-06,
      "loss": 0.1668,
      "step": 7065
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2461915119225135,
      "learning_rate": 5.155201252213576e-06,
      "loss": 0.1865,
      "step": 7066
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2273496009884375,
      "learning_rate": 5.1540434018819895e-06,
      "loss": 0.1813,
      "step": 7067
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4732433039028856,
      "learning_rate": 5.152885543281964e-06,
      "loss": 0.2226,
      "step": 7068
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.342413607381237,
      "learning_rate": 5.15172767647565e-06,
      "loss": 0.1821,
      "step": 7069
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3368474612813426,
      "learning_rate": 5.150569801525197e-06,
      "loss": 0.2137,
      "step": 7070
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3045391482715343,
      "learning_rate": 5.149411918492756e-06,
      "loss": 0.2033,
      "step": 7071
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5681584252122862,
      "learning_rate": 5.1482540274404755e-06,
      "loss": 0.1987,
      "step": 7072
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3403861875833958,
      "learning_rate": 5.1470961284305085e-06,
      "loss": 0.2026,
      "step": 7073
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2343605542430138,
      "learning_rate": 5.145938221525006e-06,
      "loss": 0.1678,
      "step": 7074
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.157920229430441,
      "learning_rate": 5.14478030678612e-06,
      "loss": 0.1879,
      "step": 7075
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5097109678860694,
      "learning_rate": 5.1436223842760015e-06,
      "loss": 0.214,
      "step": 7076
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.283303493299244,
      "learning_rate": 5.142464454056803e-06,
      "loss": 0.206,
      "step": 7077
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.148093922148823,
      "learning_rate": 5.14130651619068e-06,
      "loss": 0.603,
      "step": 7078
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3134949394077244,
      "learning_rate": 5.140148570739786e-06,
      "loss": 0.1713,
      "step": 7079
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3070542790122428,
      "learning_rate": 5.1389906177662705e-06,
      "loss": 0.2009,
      "step": 7080
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4085656614458386,
      "learning_rate": 5.137832657332292e-06,
      "loss": 0.163,
      "step": 7081
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5965888074933976,
      "learning_rate": 5.1366746895000055e-06,
      "loss": 0.2107,
      "step": 7082
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.689250647979224,
      "learning_rate": 5.135516714331563e-06,
      "loss": 0.4845,
      "step": 7083
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4793749217638605,
      "learning_rate": 5.134358731889122e-06,
      "loss": 0.2084,
      "step": 7084
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3366911326784332,
      "learning_rate": 5.133200742234837e-06,
      "loss": 0.1832,
      "step": 7085
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3293063402347323,
      "learning_rate": 5.1320427454308676e-06,
      "loss": 0.1967,
      "step": 7086
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3780726807443109,
      "learning_rate": 5.130884741539367e-06,
      "loss": 0.2029,
      "step": 7087
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3542279544229225,
      "learning_rate": 5.1297267306224925e-06,
      "loss": 0.1783,
      "step": 7088
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3952434370688405,
      "learning_rate": 5.1285687127424035e-06,
      "loss": 0.17,
      "step": 7089
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2430523442778054,
      "learning_rate": 5.127410687961256e-06,
      "loss": 0.1224,
      "step": 7090
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3472701139484442,
      "learning_rate": 5.126252656341211e-06,
      "loss": 0.1638,
      "step": 7091
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4235821670685538,
      "learning_rate": 5.125094617944424e-06,
      "loss": 0.1955,
      "step": 7092
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4102468753273756,
      "learning_rate": 5.123936572833053e-06,
      "loss": 0.2421,
      "step": 7093
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3748186155844737,
      "learning_rate": 5.122778521069261e-06,
      "loss": 0.2139,
      "step": 7094
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2313035212758103,
      "learning_rate": 5.121620462715205e-06,
      "loss": 0.1587,
      "step": 7095
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.526335250573845,
      "learning_rate": 5.120462397833047e-06,
      "loss": 0.5863,
      "step": 7096
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5514627619664212,
      "learning_rate": 5.119304326484944e-06,
      "loss": 0.2621,
      "step": 7097
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4356120022720127,
      "learning_rate": 5.11814624873306e-06,
      "loss": 0.1902,
      "step": 7098
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3353845638519553,
      "learning_rate": 5.116988164639556e-06,
      "loss": 0.1931,
      "step": 7099
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2916993189894708,
      "learning_rate": 5.115830074266592e-06,
      "loss": 0.1845,
      "step": 7100
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5881740908852107,
      "learning_rate": 5.114671977676328e-06,
      "loss": 0.2238,
      "step": 7101
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3445272740206737,
      "learning_rate": 5.113513874930928e-06,
      "loss": 0.2097,
      "step": 7102
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2974752678916706,
      "learning_rate": 5.112355766092554e-06,
      "loss": 0.1679,
      "step": 7103
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5682372204030228,
      "learning_rate": 5.11119765122337e-06,
      "loss": 0.1951,
      "step": 7104
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3558619730624946,
      "learning_rate": 5.1100395303855374e-06,
      "loss": 0.1661,
      "step": 7105
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.313818720959302,
      "learning_rate": 5.108881403641218e-06,
      "loss": 0.1479,
      "step": 7106
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2969136904250302,
      "learning_rate": 5.10772327105258e-06,
      "loss": 0.1995,
      "step": 7107
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4036773331390153,
      "learning_rate": 5.106565132681784e-06,
      "loss": 0.1822,
      "step": 7108
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.424345880653183,
      "learning_rate": 5.105406988590996e-06,
      "loss": 0.2049,
      "step": 7109
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4638574021270017,
      "learning_rate": 5.104248838842377e-06,
      "loss": 0.1925,
      "step": 7110
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3579398012646053,
      "learning_rate": 5.1030906834980965e-06,
      "loss": 0.1757,
      "step": 7111
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4209363850392511,
      "learning_rate": 5.101932522620317e-06,
      "loss": 0.2374,
      "step": 7112
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2636129932211884,
      "learning_rate": 5.1007743562712065e-06,
      "loss": 0.2344,
      "step": 7113
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3922972458480978,
      "learning_rate": 5.099616184512926e-06,
      "loss": 0.1788,
      "step": 7114
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2923905677860181,
      "learning_rate": 5.098458007407646e-06,
      "loss": 0.1716,
      "step": 7115
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4472064003350025,
      "learning_rate": 5.097299825017532e-06,
      "loss": 0.1873,
      "step": 7116
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.197994884601184,
      "learning_rate": 5.096141637404749e-06,
      "loss": 0.2006,
      "step": 7117
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3430428059975343,
      "learning_rate": 5.094983444631464e-06,
      "loss": 0.1684,
      "step": 7118
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4511034025138099,
      "learning_rate": 5.093825246759846e-06,
      "loss": 0.2206,
      "step": 7119
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.347489751604239,
      "learning_rate": 5.092667043852062e-06,
      "loss": 0.198,
      "step": 7120
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3615646558004235,
      "learning_rate": 5.091508835970278e-06,
      "loss": 0.2063,
      "step": 7121
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.354780935240967,
      "learning_rate": 5.0903506231766645e-06,
      "loss": 0.1891,
      "step": 7122
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2621190988596216,
      "learning_rate": 5.089192405533387e-06,
      "loss": 0.1956,
      "step": 7123
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5609273907338883,
      "learning_rate": 5.088034183102617e-06,
      "loss": 0.2045,
      "step": 7124
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3846848133681071,
      "learning_rate": 5.086875955946521e-06,
      "loss": 0.2378,
      "step": 7125
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2634008727311616,
      "learning_rate": 5.085717724127269e-06,
      "loss": 0.1843,
      "step": 7126
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.501295664398636,
      "learning_rate": 5.084559487707029e-06,
      "loss": 0.2247,
      "step": 7127
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.232901951792052,
      "learning_rate": 5.083401246747973e-06,
      "loss": 0.2268,
      "step": 7128
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3292853066079058,
      "learning_rate": 5.08224300131227e-06,
      "loss": 0.1728,
      "step": 7129
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.31585657939233,
      "learning_rate": 5.0810847514620884e-06,
      "loss": 0.1953,
      "step": 7130
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.482503384818558,
      "learning_rate": 5.0799264972596e-06,
      "loss": 0.2119,
      "step": 7131
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4076306674495631,
      "learning_rate": 5.0787682387669745e-06,
      "loss": 0.1974,
      "step": 7132
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4125891884076005,
      "learning_rate": 5.077609976046383e-06,
      "loss": 0.1998,
      "step": 7133
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.567041027090421,
      "learning_rate": 5.0764517091599965e-06,
      "loss": 0.2124,
      "step": 7134
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4465693417770462,
      "learning_rate": 5.075293438169985e-06,
      "loss": 0.2062,
      "step": 7135
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3230522438494927,
      "learning_rate": 5.074135163138521e-06,
      "loss": 0.1689,
      "step": 7136
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5847703677739586,
      "learning_rate": 5.0729768841277774e-06,
      "loss": 0.2245,
      "step": 7137
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4372068170293133,
      "learning_rate": 5.071818601199924e-06,
      "loss": 0.2214,
      "step": 7138
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.888439022591472,
      "learning_rate": 5.070660314417132e-06,
      "loss": 0.577,
      "step": 7139
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2882563938288163,
      "learning_rate": 5.069502023841576e-06,
      "loss": 0.202,
      "step": 7140
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.725048368243982,
      "learning_rate": 5.068343729535427e-06,
      "loss": 0.6147,
      "step": 7141
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2285588875399995,
      "learning_rate": 5.06718543156086e-06,
      "loss": 0.1554,
      "step": 7142
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3146404492714245,
      "learning_rate": 5.066027129980043e-06,
      "loss": 0.1832,
      "step": 7143
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.093121969165497,
      "learning_rate": 5.064868824855155e-06,
      "loss": 0.6943,
      "step": 7144
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.328526264128496,
      "learning_rate": 5.063710516248364e-06,
      "loss": 0.201,
      "step": 7145
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.311193202854565,
      "learning_rate": 5.0625522042218475e-06,
      "loss": 0.1916,
      "step": 7146
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3329668612325984,
      "learning_rate": 5.061393888837776e-06,
      "loss": 0.1718,
      "step": 7147
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2082439717078732,
      "learning_rate": 5.0602355701583235e-06,
      "loss": 0.1706,
      "step": 7148
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1234271194178975,
      "learning_rate": 5.059077248245666e-06,
      "loss": 0.1452,
      "step": 7149
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1654082305902267,
      "learning_rate": 5.057918923161978e-06,
      "loss": 0.1275,
      "step": 7150
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4855142455069617,
      "learning_rate": 5.056760594969431e-06,
      "loss": 0.2153,
      "step": 7151
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2988362945869183,
      "learning_rate": 5.0556022637302e-06,
      "loss": 0.1868,
      "step": 7152
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3063296163157463,
      "learning_rate": 5.0544439295064595e-06,
      "loss": 0.1704,
      "step": 7153
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4118721325138428,
      "learning_rate": 5.053285592360386e-06,
      "loss": 0.199,
      "step": 7154
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.603240108008233,
      "learning_rate": 5.052127252354155e-06,
      "loss": 0.5054,
      "step": 7155
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.16939967707264,
      "learning_rate": 5.050968909549939e-06,
      "loss": 0.1772,
      "step": 7156
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.268540087210407,
      "learning_rate": 5.049810564009915e-06,
      "loss": 0.1781,
      "step": 7157
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3227016862269099,
      "learning_rate": 5.048652215796257e-06,
      "loss": 0.1825,
      "step": 7158
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4365322812761614,
      "learning_rate": 5.047493864971142e-06,
      "loss": 0.2385,
      "step": 7159
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.19419135517295,
      "learning_rate": 5.046335511596746e-06,
      "loss": 0.1495,
      "step": 7160
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1207966817976098,
      "learning_rate": 5.045177155735241e-06,
      "loss": 0.1871,
      "step": 7161
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.443122081696332,
      "learning_rate": 5.044018797448809e-06,
      "loss": 0.238,
      "step": 7162
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.753371015677805,
      "learning_rate": 5.04286043679962e-06,
      "loss": 0.6796,
      "step": 7163
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.976693357369042,
      "learning_rate": 5.041702073849856e-06,
      "loss": 0.6184,
      "step": 7164
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3731540395904565,
      "learning_rate": 5.040543708661687e-06,
      "loss": 0.1974,
      "step": 7165
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3299295373038282,
      "learning_rate": 5.039385341297295e-06,
      "loss": 0.2264,
      "step": 7166
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.46403157349107,
      "learning_rate": 5.038226971818854e-06,
      "loss": 0.2155,
      "step": 7167
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2648680629874816,
      "learning_rate": 5.037068600288542e-06,
      "loss": 0.1817,
      "step": 7168
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0977994613283544,
      "learning_rate": 5.0359102267685324e-06,
      "loss": 0.1653,
      "step": 7169
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.365319919825025,
      "learning_rate": 5.0347518513210056e-06,
      "loss": 0.2083,
      "step": 7170
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.268268619293284,
      "learning_rate": 5.0335934740081385e-06,
      "loss": 0.1704,
      "step": 7171
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2645990387115449,
      "learning_rate": 5.032435094892106e-06,
      "loss": 0.1677,
      "step": 7172
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.289443906840907,
      "learning_rate": 5.031276714035086e-06,
      "loss": 0.1899,
      "step": 7173
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.287302540760077,
      "learning_rate": 5.030118331499257e-06,
      "loss": 0.1671,
      "step": 7174
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.527273755572574,
      "learning_rate": 5.028959947346797e-06,
      "loss": 0.654,
      "step": 7175
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3105406569884996,
      "learning_rate": 5.0278015616398805e-06,
      "loss": 0.2069,
      "step": 7176
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4233773858241843,
      "learning_rate": 5.026643174440686e-06,
      "loss": 0.2172,
      "step": 7177
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.0603740676183655,
      "learning_rate": 5.025484785811393e-06,
      "loss": 0.4947,
      "step": 7178
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.200089658853904,
      "learning_rate": 5.024326395814178e-06,
      "loss": 0.1493,
      "step": 7179
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5707208950544234,
      "learning_rate": 5.0231680045112174e-06,
      "loss": 0.2106,
      "step": 7180
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.275835884655264,
      "learning_rate": 5.022009611964691e-06,
      "loss": 0.1904,
      "step": 7181
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3321657911195346,
      "learning_rate": 5.0208512182367766e-06,
      "loss": 0.199,
      "step": 7182
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.318506769226695,
      "learning_rate": 5.0196928233896516e-06,
      "loss": 0.2072,
      "step": 7183
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.487178949015922,
      "learning_rate": 5.018534427485494e-06,
      "loss": 0.2013,
      "step": 7184
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4446371097399218,
      "learning_rate": 5.017376030586481e-06,
      "loss": 0.2523,
      "step": 7185
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.39385896214302,
      "learning_rate": 5.016217632754792e-06,
      "loss": 0.164,
      "step": 7186
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.227147006811772,
      "learning_rate": 5.015059234052605e-06,
      "loss": 0.5688,
      "step": 7187
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4184924267552947,
      "learning_rate": 5.013900834542099e-06,
      "loss": 0.201,
      "step": 7188
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2737997741165885,
      "learning_rate": 5.012742434285452e-06,
      "loss": 0.1964,
      "step": 7189
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2762622751733355,
      "learning_rate": 5.0115840333448386e-06,
      "loss": 0.1876,
      "step": 7190
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4243683003054242,
      "learning_rate": 5.010425631782443e-06,
      "loss": 0.1955,
      "step": 7191
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3048485420763423,
      "learning_rate": 5.00926722966044e-06,
      "loss": 0.1689,
      "step": 7192
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3571129157436526,
      "learning_rate": 5.00810882704101e-06,
      "loss": 0.1641,
      "step": 7193
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3497116202850905,
      "learning_rate": 5.006950423986328e-06,
      "loss": 0.1854,
      "step": 7194
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5095096263051992,
      "learning_rate": 5.005792020558577e-06,
      "loss": 0.239,
      "step": 7195
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3080272985335348,
      "learning_rate": 5.004633616819934e-06,
      "loss": 0.1866,
      "step": 7196
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6997148461561171,
      "learning_rate": 5.003475212832576e-06,
      "loss": 0.1781,
      "step": 7197
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.172788469801866,
      "learning_rate": 5.00231680865868e-06,
      "loss": 0.1648,
      "step": 7198
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2767903494474397,
      "learning_rate": 5.00115840436043e-06,
      "loss": 0.1986,
      "step": 7199
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.041474753827072,
      "learning_rate": 5e-06,
      "loss": 0.4518,
      "step": 7200
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2536997121175588,
      "learning_rate": 4.9988415956395706e-06,
      "loss": 0.1821,
      "step": 7201
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3672113597070907,
      "learning_rate": 4.9976831913413205e-06,
      "loss": 0.1916,
      "step": 7202
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.582090708222644,
      "learning_rate": 4.996524787167425e-06,
      "loss": 0.2201,
      "step": 7203
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3620995852998068,
      "learning_rate": 4.995366383180068e-06,
      "loss": 0.228,
      "step": 7204
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5879134785385052,
      "learning_rate": 4.9942079794414226e-06,
      "loss": 0.213,
      "step": 7205
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2148935314774743,
      "learning_rate": 4.9930495760136725e-06,
      "loss": 0.2066,
      "step": 7206
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3081760021480335,
      "learning_rate": 4.991891172958991e-06,
      "loss": 0.183,
      "step": 7207
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2331301385897968,
      "learning_rate": 4.990732770339561e-06,
      "loss": 0.1481,
      "step": 7208
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.279907423070733,
      "learning_rate": 4.989574368217558e-06,
      "loss": 0.1944,
      "step": 7209
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3597232937832466,
      "learning_rate": 4.988415966655162e-06,
      "loss": 0.1914,
      "step": 7210
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5107577266807566,
      "learning_rate": 4.987257565714551e-06,
      "loss": 0.2282,
      "step": 7211
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.445690571469777,
      "learning_rate": 4.9860991654579025e-06,
      "loss": 0.2215,
      "step": 7212
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.17942560246018,
      "learning_rate": 4.984940765947395e-06,
      "loss": 0.1598,
      "step": 7213
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3621631167789856,
      "learning_rate": 4.98378236724521e-06,
      "loss": 0.1944,
      "step": 7214
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4129475082639673,
      "learning_rate": 4.982623969413521e-06,
      "loss": 0.2072,
      "step": 7215
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2199817422306318,
      "learning_rate": 4.981465572514509e-06,
      "loss": 0.2154,
      "step": 7216
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.267152765163097,
      "learning_rate": 4.98030717661035e-06,
      "loss": 0.1766,
      "step": 7217
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3291759934935892,
      "learning_rate": 4.979148781763226e-06,
      "loss": 0.1622,
      "step": 7218
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3953027974581143,
      "learning_rate": 4.97799038803531e-06,
      "loss": 0.2287,
      "step": 7219
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4787157786553355,
      "learning_rate": 4.976831995488784e-06,
      "loss": 0.2537,
      "step": 7220
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.452085840695506,
      "learning_rate": 4.975673604185824e-06,
      "loss": 0.195,
      "step": 7221
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3237866763462198,
      "learning_rate": 4.9745152141886096e-06,
      "loss": 0.2108,
      "step": 7222
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.289557695544683,
      "learning_rate": 4.973356825559315e-06,
      "loss": 0.6423,
      "step": 7223
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3837522135104725,
      "learning_rate": 4.972198438360122e-06,
      "loss": 0.1959,
      "step": 7224
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4240604515708708,
      "learning_rate": 4.971040052653205e-06,
      "loss": 0.2155,
      "step": 7225
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0929029495203233,
      "learning_rate": 4.969881668500745e-06,
      "loss": 0.141,
      "step": 7226
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2183924010946645,
      "learning_rate": 4.968723285964915e-06,
      "loss": 0.1635,
      "step": 7227
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2381034566037676,
      "learning_rate": 4.9675649051078965e-06,
      "loss": 0.177,
      "step": 7228
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2324970197661276,
      "learning_rate": 4.966406525991863e-06,
      "loss": 0.1812,
      "step": 7229
    },
    {
      "epoch": 0.52,
      "grad_norm": 12.67712826366447,
      "learning_rate": 4.965248148678997e-06,
      "loss": 0.7243,
      "step": 7230
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3704204329014857,
      "learning_rate": 4.964089773231469e-06,
      "loss": 0.2183,
      "step": 7231
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1366786738316605,
      "learning_rate": 4.962931399711461e-06,
      "loss": 0.1471,
      "step": 7232
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.358856610583483,
      "learning_rate": 4.961773028181147e-06,
      "loss": 0.1786,
      "step": 7233
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2722444780062026,
      "learning_rate": 4.960614658702705e-06,
      "loss": 0.2235,
      "step": 7234
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2684994142282349,
      "learning_rate": 4.959456291338314e-06,
      "loss": 0.2169,
      "step": 7235
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4585472899932999,
      "learning_rate": 4.958297926150146e-06,
      "loss": 0.1974,
      "step": 7236
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2536169354652384,
      "learning_rate": 4.957139563200381e-06,
      "loss": 0.1816,
      "step": 7237
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2679151195075582,
      "learning_rate": 4.955981202551192e-06,
      "loss": 0.1848,
      "step": 7238
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2961624324536725,
      "learning_rate": 4.95482284426476e-06,
      "loss": 0.1885,
      "step": 7239
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5136766882175596,
      "learning_rate": 4.953664488403256e-06,
      "loss": 0.1699,
      "step": 7240
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4258857279984767,
      "learning_rate": 4.952506135028859e-06,
      "loss": 0.184,
      "step": 7241
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3284345979884045,
      "learning_rate": 4.951347784203744e-06,
      "loss": 0.197,
      "step": 7242
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.021895950883319,
      "learning_rate": 4.950189435990087e-06,
      "loss": 0.5727,
      "step": 7243
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.383968723727223,
      "learning_rate": 4.949031090450062e-06,
      "loss": 0.2434,
      "step": 7244
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5153094862349283,
      "learning_rate": 4.9478727476458465e-06,
      "loss": 0.247,
      "step": 7245
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.446523160530515,
      "learning_rate": 4.946714407639614e-06,
      "loss": 0.2202,
      "step": 7246
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2291567626918671,
      "learning_rate": 4.945556070493542e-06,
      "loss": 0.1501,
      "step": 7247
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3395754633905068,
      "learning_rate": 4.944397736269803e-06,
      "loss": 0.1936,
      "step": 7248
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7101356585680272,
      "learning_rate": 4.943239405030572e-06,
      "loss": 0.1979,
      "step": 7249
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.251054021132169,
      "learning_rate": 4.942081076838024e-06,
      "loss": 0.1662,
      "step": 7250
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2864146345263114,
      "learning_rate": 4.940922751754336e-06,
      "loss": 0.1824,
      "step": 7251
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4474859665105393,
      "learning_rate": 4.939764429841677e-06,
      "loss": 0.2224,
      "step": 7252
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.806272501150447,
      "learning_rate": 4.9386061111622266e-06,
      "loss": 0.5632,
      "step": 7253
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1174045090648597,
      "learning_rate": 4.937447795778154e-06,
      "loss": 0.1268,
      "step": 7254
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3445036050985604,
      "learning_rate": 4.936289483751638e-06,
      "loss": 0.185,
      "step": 7255
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3899334386654179,
      "learning_rate": 4.9351311751448465e-06,
      "loss": 0.1586,
      "step": 7256
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3582388646262364,
      "learning_rate": 4.9339728700199575e-06,
      "loss": 0.1742,
      "step": 7257
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4797610692058374,
      "learning_rate": 4.932814568439142e-06,
      "loss": 0.2222,
      "step": 7258
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6336743400438396,
      "learning_rate": 4.931656270464574e-06,
      "loss": 0.1898,
      "step": 7259
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4547344512481002,
      "learning_rate": 4.9304979761584256e-06,
      "loss": 0.2186,
      "step": 7260
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1713156718117477,
      "learning_rate": 4.92933968558287e-06,
      "loss": 0.1807,
      "step": 7261
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4692113224133134,
      "learning_rate": 4.928181398800078e-06,
      "loss": 0.1757,
      "step": 7262
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1235768391195224,
      "learning_rate": 4.927023115872225e-06,
      "loss": 0.1663,
      "step": 7263
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5541117011835093,
      "learning_rate": 4.9258648368614796e-06,
      "loss": 0.2246,
      "step": 7264
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2904602747690472,
      "learning_rate": 4.924706561830015e-06,
      "loss": 0.2148,
      "step": 7265
    },
    {
      "epoch": 0.52,
      "grad_norm": 13.952410729401024,
      "learning_rate": 4.923548290840005e-06,
      "loss": 0.4815,
      "step": 7266
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.936815840751015,
      "learning_rate": 4.922390023953618e-06,
      "loss": 0.5943,
      "step": 7267
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.32528501384882,
      "learning_rate": 4.921231761233027e-06,
      "loss": 0.1896,
      "step": 7268
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4407976403259906,
      "learning_rate": 4.9200735027404e-06,
      "loss": 0.2312,
      "step": 7269
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4380365246466567,
      "learning_rate": 4.918915248537912e-06,
      "loss": 0.2366,
      "step": 7270
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0975541839221759,
      "learning_rate": 4.91775699868773e-06,
      "loss": 0.1797,
      "step": 7271
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2486216966963446,
      "learning_rate": 4.9165987532520276e-06,
      "loss": 0.1758,
      "step": 7272
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.501037422263749,
      "learning_rate": 4.915440512292971e-06,
      "loss": 0.2041,
      "step": 7273
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2208790924061084,
      "learning_rate": 4.914282275872732e-06,
      "loss": 0.1662,
      "step": 7274
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.9215426657282593,
      "learning_rate": 4.913124044053479e-06,
      "loss": 0.5136,
      "step": 7275
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3753466016961633,
      "learning_rate": 4.911965816897385e-06,
      "loss": 0.161,
      "step": 7276
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2745614531755465,
      "learning_rate": 4.910807594466613e-06,
      "loss": 0.1475,
      "step": 7277
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.19092527556746,
      "learning_rate": 4.909649376823337e-06,
      "loss": 0.155,
      "step": 7278
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4481344305296573,
      "learning_rate": 4.908491164029723e-06,
      "loss": 0.2063,
      "step": 7279
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.329560080587342,
      "learning_rate": 4.90733295614794e-06,
      "loss": 0.2525,
      "step": 7280
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2523304727835112,
      "learning_rate": 4.906174753240155e-06,
      "loss": 0.19,
      "step": 7281
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4429945406171445,
      "learning_rate": 4.905016555368538e-06,
      "loss": 0.2218,
      "step": 7282
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3177743814041725,
      "learning_rate": 4.9038583625952525e-06,
      "loss": 0.204,
      "step": 7283
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.276489494284318,
      "learning_rate": 4.902700174982471e-06,
      "loss": 0.1768,
      "step": 7284
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.37507967873485,
      "learning_rate": 4.901541992592355e-06,
      "loss": 0.1652,
      "step": 7285
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3298418072268594,
      "learning_rate": 4.900383815487075e-06,
      "loss": 0.1957,
      "step": 7286
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2971847884593566,
      "learning_rate": 4.899225643728795e-06,
      "loss": 0.1912,
      "step": 7287
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1235143231379767,
      "learning_rate": 4.8980674773796845e-06,
      "loss": 0.1456,
      "step": 7288
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5511653697656087,
      "learning_rate": 4.896909316501904e-06,
      "loss": 0.2052,
      "step": 7289
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2497891335025184,
      "learning_rate": 4.895751161157624e-06,
      "loss": 0.2156,
      "step": 7290
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2982925137909171,
      "learning_rate": 4.894593011409006e-06,
      "loss": 0.1841,
      "step": 7291
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.609937657003099,
      "learning_rate": 4.893434867318218e-06,
      "loss": 0.6731,
      "step": 7292
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.337492621747176,
      "learning_rate": 4.892276728947421e-06,
      "loss": 0.2329,
      "step": 7293
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2141356776287322,
      "learning_rate": 4.8911185963587834e-06,
      "loss": 0.1669,
      "step": 7294
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5056853718097494,
      "learning_rate": 4.889960469614464e-06,
      "loss": 0.243,
      "step": 7295
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3035038475868024,
      "learning_rate": 4.88880234877663e-06,
      "loss": 0.1248,
      "step": 7296
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.390247196683685,
      "learning_rate": 4.887644233907448e-06,
      "loss": 0.2051,
      "step": 7297
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3209611805997716,
      "learning_rate": 4.886486125069073e-06,
      "loss": 0.177,
      "step": 7298
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2989256119094814,
      "learning_rate": 4.885328022323674e-06,
      "loss": 0.1899,
      "step": 7299
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2629127564680402,
      "learning_rate": 4.884169925733409e-06,
      "loss": 0.2017,
      "step": 7300
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.141562905397782,
      "learning_rate": 4.883011835360446e-06,
      "loss": 0.1262,
      "step": 7301
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2695280126944313,
      "learning_rate": 4.881853751266939e-06,
      "loss": 0.1835,
      "step": 7302
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2108592159030154,
      "learning_rate": 4.880695673515057e-06,
      "loss": 0.164,
      "step": 7303
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2205562621302926,
      "learning_rate": 4.879537602166953e-06,
      "loss": 0.1844,
      "step": 7304
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3908186249477168,
      "learning_rate": 4.878379537284796e-06,
      "loss": 0.1812,
      "step": 7305
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3569030537246227,
      "learning_rate": 4.877221478930739e-06,
      "loss": 0.1674,
      "step": 7306
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.280329524705964,
      "learning_rate": 4.876063427166948e-06,
      "loss": 0.2007,
      "step": 7307
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3008768496869145,
      "learning_rate": 4.874905382055578e-06,
      "loss": 0.1788,
      "step": 7308
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3407500633903797,
      "learning_rate": 4.873747343658791e-06,
      "loss": 0.2049,
      "step": 7309
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.390948046598325,
      "learning_rate": 4.872589312038744e-06,
      "loss": 0.1898,
      "step": 7310
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2537742403987546,
      "learning_rate": 4.871431287257599e-06,
      "loss": 0.1903,
      "step": 7311
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4346444093672632,
      "learning_rate": 4.870273269377508e-06,
      "loss": 0.1695,
      "step": 7312
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.354917485961351,
      "learning_rate": 4.869115258460636e-06,
      "loss": 0.1927,
      "step": 7313
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2857880213701396,
      "learning_rate": 4.867957254569134e-06,
      "loss": 0.1861,
      "step": 7314
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.15796483859027,
      "learning_rate": 4.8667992577651646e-06,
      "loss": 0.1625,
      "step": 7315
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4327851727478282,
      "learning_rate": 4.86564126811088e-06,
      "loss": 0.1738,
      "step": 7316
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2862982121576751,
      "learning_rate": 4.8644832856684394e-06,
      "loss": 0.2144,
      "step": 7317
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6023695815945183,
      "learning_rate": 4.863325310499996e-06,
      "loss": 0.1993,
      "step": 7318
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5543733946376874,
      "learning_rate": 4.862167342667709e-06,
      "loss": 0.2116,
      "step": 7319
    },
    {
      "epoch": 0.52,
      "grad_norm": 9.339510360605741,
      "learning_rate": 4.86100938223373e-06,
      "loss": 0.586,
      "step": 7320
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3631627320827617,
      "learning_rate": 4.859851429260217e-06,
      "loss": 0.1683,
      "step": 7321
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2872997094989862,
      "learning_rate": 4.8586934838093204e-06,
      "loss": 0.187,
      "step": 7322
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3033670413965044,
      "learning_rate": 4.8575355459431985e-06,
      "loss": 0.1505,
      "step": 7323
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3982163016026572,
      "learning_rate": 4.856377615724001e-06,
      "loss": 0.2444,
      "step": 7324
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.541083080057253,
      "learning_rate": 4.8552196932138825e-06,
      "loss": 0.3898,
      "step": 7325
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4586587922520873,
      "learning_rate": 4.854061778474995e-06,
      "loss": 0.2184,
      "step": 7326
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2563978131545368,
      "learning_rate": 4.852903871569494e-06,
      "loss": 0.1896,
      "step": 7327
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3299224526319926,
      "learning_rate": 4.851745972559526e-06,
      "loss": 0.2022,
      "step": 7328
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.206617621403913,
      "learning_rate": 4.850588081507246e-06,
      "loss": 0.1611,
      "step": 7329
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.224242849284024,
      "learning_rate": 4.8494301984748045e-06,
      "loss": 0.2086,
      "step": 7330
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.269434723831549,
      "learning_rate": 4.848272323524351e-06,
      "loss": 0.1818,
      "step": 7331
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3987682151947491,
      "learning_rate": 4.847114456718039e-06,
      "loss": 0.2254,
      "step": 7332
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0772069578775725,
      "learning_rate": 4.845956598118012e-06,
      "loss": 0.1523,
      "step": 7333
    },
    {
      "epoch": 0.52,
      "grad_norm": 11.949436907086621,
      "learning_rate": 4.844798747786425e-06,
      "loss": 0.5902,
      "step": 7334
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3276774537208853,
      "learning_rate": 4.843640905785423e-06,
      "loss": 0.1978,
      "step": 7335
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6061625362681544,
      "learning_rate": 4.84248307217716e-06,
      "loss": 0.2129,
      "step": 7336
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4096761993167843,
      "learning_rate": 4.841325247023776e-06,
      "loss": 0.2032,
      "step": 7337
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0976466420717723,
      "learning_rate": 4.840167430387423e-06,
      "loss": 0.1471,
      "step": 7338
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.966879966219873,
      "learning_rate": 4.839009622330247e-06,
      "loss": 0.6256,
      "step": 7339
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1759283384509396,
      "learning_rate": 4.837851822914397e-06,
      "loss": 0.2016,
      "step": 7340
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0655314852737376,
      "learning_rate": 4.836694032202015e-06,
      "loss": 0.1386,
      "step": 7341
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.285416307320226,
      "learning_rate": 4.83553625025525e-06,
      "loss": 0.1709,
      "step": 7342
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3097383374586917,
      "learning_rate": 4.834378477136244e-06,
      "loss": 0.1742,
      "step": 7343
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2942769505797476,
      "learning_rate": 4.8332207129071455e-06,
      "loss": 0.1936,
      "step": 7344
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2984396213558214,
      "learning_rate": 4.832062957630095e-06,
      "loss": 0.2043,
      "step": 7345
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3171272321483678,
      "learning_rate": 4.830905211367238e-06,
      "loss": 0.2066,
      "step": 7346
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2074312636259634,
      "learning_rate": 4.829747474180717e-06,
      "loss": 0.176,
      "step": 7347
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.025901396791245,
      "learning_rate": 4.828589746132677e-06,
      "loss": 0.6538,
      "step": 7348
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.036414668709515,
      "learning_rate": 4.827432027285257e-06,
      "loss": 0.1872,
      "step": 7349
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5767917376482379,
      "learning_rate": 4.8262743177006e-06,
      "loss": 0.2192,
      "step": 7350
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5482291273528677,
      "learning_rate": 4.825116617440848e-06,
      "loss": 0.2522,
      "step": 7351
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4126738962314416,
      "learning_rate": 4.823958926568142e-06,
      "loss": 0.2297,
      "step": 7352
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3682289194408515,
      "learning_rate": 4.82280124514462e-06,
      "loss": 0.1847,
      "step": 7353
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2850062988418043,
      "learning_rate": 4.821643573232424e-06,
      "loss": 0.1714,
      "step": 7354
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6083594227127784,
      "learning_rate": 4.8204859108936915e-06,
      "loss": 0.2067,
      "step": 7355
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5074176420267948,
      "learning_rate": 4.819328258190564e-06,
      "loss": 0.2392,
      "step": 7356
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4540648071303213,
      "learning_rate": 4.818170615185176e-06,
      "loss": 0.219,
      "step": 7357
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.422685529173656,
      "learning_rate": 4.817012981939668e-06,
      "loss": 0.2197,
      "step": 7358
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5068611539789891,
      "learning_rate": 4.815855358516175e-06,
      "loss": 0.1801,
      "step": 7359
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.44486765944861,
      "learning_rate": 4.814697744976835e-06,
      "loss": 0.6532,
      "step": 7360
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3196680337249584,
      "learning_rate": 4.813540141383786e-06,
      "loss": 0.1932,
      "step": 7361
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3848005487820005,
      "learning_rate": 4.812382547799159e-06,
      "loss": 0.1772,
      "step": 7362
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.403622958238679,
      "learning_rate": 4.811224964285092e-06,
      "loss": 0.2088,
      "step": 7363
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3377788197977927,
      "learning_rate": 4.8100673909037185e-06,
      "loss": 0.1793,
      "step": 7364
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4195117443644716,
      "learning_rate": 4.808909827717175e-06,
      "loss": 0.1896,
      "step": 7365
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2542331270645244,
      "learning_rate": 4.807752274787592e-06,
      "loss": 0.1919,
      "step": 7366
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2640004220954202,
      "learning_rate": 4.8065947321771025e-06,
      "loss": 0.1916,
      "step": 7367
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.994655482737985,
      "learning_rate": 4.805437199947838e-06,
      "loss": 0.5996,
      "step": 7368
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5101926574229056,
      "learning_rate": 4.804279678161935e-06,
      "loss": 0.1927,
      "step": 7369
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.307332046894558,
      "learning_rate": 4.803122166881519e-06,
      "loss": 0.5202,
      "step": 7370
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.445615989769837,
      "learning_rate": 4.801964666168723e-06,
      "loss": 0.2136,
      "step": 7371
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.072966814033737,
      "learning_rate": 4.800807176085677e-06,
      "loss": 0.3656,
      "step": 7372
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.355198698443477,
      "learning_rate": 4.799649696694512e-06,
      "loss": 0.1735,
      "step": 7373
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3406970666226732,
      "learning_rate": 4.7984922280573535e-06,
      "loss": 0.2022,
      "step": 7374
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4860760816275518,
      "learning_rate": 4.7973347702363336e-06,
      "loss": 0.2404,
      "step": 7375
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.305830879975686,
      "learning_rate": 4.796177323293576e-06,
      "loss": 0.1722,
      "step": 7376
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4354173743669323,
      "learning_rate": 4.795019887291212e-06,
      "loss": 0.2449,
      "step": 7377
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2934662271546125,
      "learning_rate": 4.793862462291364e-06,
      "loss": 0.1762,
      "step": 7378
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.216034262255652,
      "learning_rate": 4.792705048356161e-06,
      "loss": 0.1899,
      "step": 7379
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3691484553183713,
      "learning_rate": 4.791547645547727e-06,
      "loss": 0.2153,
      "step": 7380
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2679985149349193,
      "learning_rate": 4.790390253928189e-06,
      "loss": 0.1903,
      "step": 7381
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2434538104662756,
      "learning_rate": 4.789232873559667e-06,
      "loss": 0.1468,
      "step": 7382
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.373274718606192,
      "learning_rate": 4.788075504504289e-06,
      "loss": 0.1627,
      "step": 7383
    },
    {
      "epoch": 0.53,
      "grad_norm": 10.26386636235655,
      "learning_rate": 4.7869181468241735e-06,
      "loss": 0.5167,
      "step": 7384
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5483747842385602,
      "learning_rate": 4.785760800581448e-06,
      "loss": 0.2335,
      "step": 7385
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.50393022660587,
      "learning_rate": 4.784603465838231e-06,
      "loss": 0.25,
      "step": 7386
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3695507054958678,
      "learning_rate": 4.783446142656643e-06,
      "loss": 0.2185,
      "step": 7387
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3315442409394023,
      "learning_rate": 4.782288831098806e-06,
      "loss": 0.1834,
      "step": 7388
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3363698480693038,
      "learning_rate": 4.781131531226842e-06,
      "loss": 0.1948,
      "step": 7389
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.400715007782383,
      "learning_rate": 4.779974243102865e-06,
      "loss": 0.2007,
      "step": 7390
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3040043148159026,
      "learning_rate": 4.778816966788998e-06,
      "loss": 0.1652,
      "step": 7391
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3135387225702022,
      "learning_rate": 4.777659702347356e-06,
      "loss": 0.1936,
      "step": 7392
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4537390399218433,
      "learning_rate": 4.776502449840058e-06,
      "loss": 0.168,
      "step": 7393
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3480107666932997,
      "learning_rate": 4.775345209329224e-06,
      "loss": 0.1763,
      "step": 7394
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2992857960915318,
      "learning_rate": 4.774187980876963e-06,
      "loss": 0.2042,
      "step": 7395
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4740279415729771,
      "learning_rate": 4.773030764545395e-06,
      "loss": 0.2113,
      "step": 7396
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3182544174688373,
      "learning_rate": 4.771873560396634e-06,
      "loss": 0.2,
      "step": 7397
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3711383785423088,
      "learning_rate": 4.770716368492795e-06,
      "loss": 0.2283,
      "step": 7398
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3289072097781582,
      "learning_rate": 4.769559188895989e-06,
      "loss": 0.1949,
      "step": 7399
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3983638592501968,
      "learning_rate": 4.768402021668332e-06,
      "loss": 0.174,
      "step": 7400
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2628474933458138,
      "learning_rate": 4.767244866871931e-06,
      "loss": 0.1843,
      "step": 7401
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.380312232625276,
      "learning_rate": 4.7660877245689045e-06,
      "loss": 0.5109,
      "step": 7402
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.235837644737068,
      "learning_rate": 4.764930594821357e-06,
      "loss": 0.1482,
      "step": 7403
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.0942579645975785,
      "learning_rate": 4.7637734776914045e-06,
      "loss": 0.7299,
      "step": 7404
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.258313857088213,
      "learning_rate": 4.762616373241151e-06,
      "loss": 0.1932,
      "step": 7405
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3604786415206325,
      "learning_rate": 4.7614592815327076e-06,
      "loss": 0.161,
      "step": 7406
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.323515850043093,
      "learning_rate": 4.760302202628182e-06,
      "loss": 0.1998,
      "step": 7407
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2450979967927167,
      "learning_rate": 4.759145136589683e-06,
      "loss": 0.179,
      "step": 7408
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3952837437189276,
      "learning_rate": 4.757988083479315e-06,
      "loss": 0.1935,
      "step": 7409
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1107711498467083,
      "learning_rate": 4.756831043359186e-06,
      "loss": 0.1306,
      "step": 7410
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.283819317095004,
      "learning_rate": 4.7556740162913995e-06,
      "loss": 0.1669,
      "step": 7411
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4586722670483339,
      "learning_rate": 4.754517002338064e-06,
      "loss": 0.1994,
      "step": 7412
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3961467076645901,
      "learning_rate": 4.753360001561277e-06,
      "loss": 0.2379,
      "step": 7413
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3582956062301503,
      "learning_rate": 4.752203014023147e-06,
      "loss": 0.2069,
      "step": 7414
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1938817043450436,
      "learning_rate": 4.7510460397857736e-06,
      "loss": 0.1645,
      "step": 7415
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3700288286196967,
      "learning_rate": 4.749889078911262e-06,
      "loss": 0.196,
      "step": 7416
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3045945122636486,
      "learning_rate": 4.748732131461709e-06,
      "loss": 0.2054,
      "step": 7417
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2976636552735104,
      "learning_rate": 4.747575197499218e-06,
      "loss": 0.188,
      "step": 7418
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4570696624844335,
      "learning_rate": 4.746418277085887e-06,
      "loss": 0.229,
      "step": 7419
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.17924883019989,
      "learning_rate": 4.7452613702838166e-06,
      "loss": 0.6187,
      "step": 7420
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3237423818177605,
      "learning_rate": 4.744104477155102e-06,
      "loss": 0.256,
      "step": 7421
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3302607452389341,
      "learning_rate": 4.742947597761844e-06,
      "loss": 0.2334,
      "step": 7422
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3270104082507541,
      "learning_rate": 4.741790732166136e-06,
      "loss": 0.1957,
      "step": 7423
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2403488693908353,
      "learning_rate": 4.740633880430076e-06,
      "loss": 0.1545,
      "step": 7424
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3007745766263163,
      "learning_rate": 4.739477042615761e-06,
      "loss": 0.2103,
      "step": 7425
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4512589295588125,
      "learning_rate": 4.738320218785281e-06,
      "loss": 0.2166,
      "step": 7426
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.264973534188588,
      "learning_rate": 4.737163409000733e-06,
      "loss": 0.1884,
      "step": 7427
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.288248777721767,
      "learning_rate": 4.736006613324209e-06,
      "loss": 0.1572,
      "step": 7428
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5302413702590902,
      "learning_rate": 4.734849831817803e-06,
      "loss": 0.2239,
      "step": 7429
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3954710446929588,
      "learning_rate": 4.7336930645436015e-06,
      "loss": 0.2184,
      "step": 7430
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.415124432577513,
      "learning_rate": 4.7325363115637005e-06,
      "loss": 0.621,
      "step": 7431
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.123139573331286,
      "learning_rate": 4.7313795729401855e-06,
      "loss": 0.5965,
      "step": 7432
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4131847385151435,
      "learning_rate": 4.730222848735151e-06,
      "loss": 0.2261,
      "step": 7433
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.376843979404407,
      "learning_rate": 4.72906613901068e-06,
      "loss": 0.1794,
      "step": 7434
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5843620467800101,
      "learning_rate": 4.727909443828862e-06,
      "loss": 0.1976,
      "step": 7435
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3093756956624991,
      "learning_rate": 4.726752763251785e-06,
      "loss": 0.1942,
      "step": 7436
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5223379067965448,
      "learning_rate": 4.7255960973415345e-06,
      "loss": 0.2039,
      "step": 7437
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4044720436427227,
      "learning_rate": 4.7244394461601944e-06,
      "loss": 0.2192,
      "step": 7438
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3230537537598956,
      "learning_rate": 4.723282809769851e-06,
      "loss": 0.1693,
      "step": 7439
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.352622555753637,
      "learning_rate": 4.722126188232586e-06,
      "loss": 0.1913,
      "step": 7440
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4375877015157081,
      "learning_rate": 4.720969581610485e-06,
      "loss": 0.1867,
      "step": 7441
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.147654026772895,
      "learning_rate": 4.7198129899656266e-06,
      "loss": 0.173,
      "step": 7442
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1345156910842074,
      "learning_rate": 4.7186564133600945e-06,
      "loss": 0.1657,
      "step": 7443
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3109128805295256,
      "learning_rate": 4.717499851855967e-06,
      "loss": 0.2205,
      "step": 7444
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.211770489295749,
      "learning_rate": 4.716343305515329e-06,
      "loss": 0.2163,
      "step": 7445
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4094523703247277,
      "learning_rate": 4.715186774400251e-06,
      "loss": 0.1874,
      "step": 7446
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3395440935286418,
      "learning_rate": 4.714030258572818e-06,
      "loss": 0.1841,
      "step": 7447
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.121677750773963,
      "learning_rate": 4.7128737580951024e-06,
      "loss": 0.1791,
      "step": 7448
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.407790279044471,
      "learning_rate": 4.711717273029187e-06,
      "loss": 0.2071,
      "step": 7449
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2893210635600285,
      "learning_rate": 4.710560803437139e-06,
      "loss": 0.1854,
      "step": 7450
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3638126877601455,
      "learning_rate": 4.709404349381038e-06,
      "loss": 0.1856,
      "step": 7451
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.307991108455758,
      "learning_rate": 4.708247910922958e-06,
      "loss": 0.1723,
      "step": 7452
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.142350626110311,
      "learning_rate": 4.707091488124972e-06,
      "loss": 0.465,
      "step": 7453
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1069841203677042,
      "learning_rate": 4.705935081049149e-06,
      "loss": 0.1912,
      "step": 7454
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.498612101800132,
      "learning_rate": 4.704778689757564e-06,
      "loss": 0.208,
      "step": 7455
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.280886250263835,
      "learning_rate": 4.7036223143122845e-06,
      "loss": 0.2169,
      "step": 7456
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1795375075721113,
      "learning_rate": 4.702465954775382e-06,
      "loss": 0.1969,
      "step": 7457
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.142871764078737,
      "learning_rate": 4.701309611208927e-06,
      "loss": 0.7533,
      "step": 7458
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.948718341608626,
      "learning_rate": 4.700153283674984e-06,
      "loss": 0.6144,
      "step": 7459
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4117807233681705,
      "learning_rate": 4.698996972235622e-06,
      "loss": 0.1671,
      "step": 7460
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3606036414712264,
      "learning_rate": 4.697840676952905e-06,
      "loss": 0.2165,
      "step": 7461
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2644966230161345,
      "learning_rate": 4.696684397888904e-06,
      "loss": 0.1832,
      "step": 7462
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3744329374141788,
      "learning_rate": 4.695528135105676e-06,
      "loss": 0.1898,
      "step": 7463
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6040346584296856,
      "learning_rate": 4.694371888665289e-06,
      "loss": 0.1852,
      "step": 7464
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1906228712447857,
      "learning_rate": 4.693215658629805e-06,
      "loss": 0.1744,
      "step": 7465
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3656986106250484,
      "learning_rate": 4.692059445061287e-06,
      "loss": 0.2176,
      "step": 7466
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3854840961978976,
      "learning_rate": 4.690903248021792e-06,
      "loss": 0.2009,
      "step": 7467
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.318983167426717,
      "learning_rate": 4.6897470675733845e-06,
      "loss": 0.1753,
      "step": 7468
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.421237380734001,
      "learning_rate": 4.68859090377812e-06,
      "loss": 0.2127,
      "step": 7469
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4008717756530908,
      "learning_rate": 4.6874347566980615e-06,
      "loss": 0.2463,
      "step": 7470
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.343637960279099,
      "learning_rate": 4.686278626395262e-06,
      "loss": 0.1773,
      "step": 7471
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3895064173217797,
      "learning_rate": 4.685122512931779e-06,
      "loss": 0.187,
      "step": 7472
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3697850465019976,
      "learning_rate": 4.683966416369668e-06,
      "loss": 0.2068,
      "step": 7473
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3948972938503301,
      "learning_rate": 4.682810336770987e-06,
      "loss": 0.2518,
      "step": 7474
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5889777664355842,
      "learning_rate": 4.681654274197784e-06,
      "loss": 0.2258,
      "step": 7475
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.311569480262528,
      "learning_rate": 4.680498228712116e-06,
      "loss": 0.2146,
      "step": 7476
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3034001980187484,
      "learning_rate": 4.6793422003760335e-06,
      "loss": 0.1844,
      "step": 7477
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.232291608204105,
      "learning_rate": 4.678186189251589e-06,
      "loss": 0.1741,
      "step": 7478
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.377896008808318,
      "learning_rate": 4.67703019540083e-06,
      "loss": 0.2015,
      "step": 7479
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2794902669522432,
      "learning_rate": 4.6758742188858074e-06,
      "loss": 0.1934,
      "step": 7480
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3800895729808818,
      "learning_rate": 4.674718259768569e-06,
      "loss": 0.2113,
      "step": 7481
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3367197074514325,
      "learning_rate": 4.673562318111163e-06,
      "loss": 0.2051,
      "step": 7482
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5832329453571012,
      "learning_rate": 4.672406393975634e-06,
      "loss": 0.2527,
      "step": 7483
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3684726914153058,
      "learning_rate": 4.671250487424028e-06,
      "loss": 0.1914,
      "step": 7484
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3536269044892348,
      "learning_rate": 4.67009459851839e-06,
      "loss": 0.1867,
      "step": 7485
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4262834924989731,
      "learning_rate": 4.6689387273207645e-06,
      "loss": 0.1721,
      "step": 7486
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2255674605852958,
      "learning_rate": 4.667782873893192e-06,
      "loss": 0.1861,
      "step": 7487
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4113078698314363,
      "learning_rate": 4.666627038297713e-06,
      "loss": 0.1686,
      "step": 7488
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1850875320488745,
      "learning_rate": 4.665471220596373e-06,
      "loss": 0.1526,
      "step": 7489
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3635934995601118,
      "learning_rate": 4.664315420851207e-06,
      "loss": 0.1939,
      "step": 7490
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2169037576271426,
      "learning_rate": 4.6631596391242575e-06,
      "loss": 0.1639,
      "step": 7491
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3870638583559538,
      "learning_rate": 4.662003875477559e-06,
      "loss": 0.1849,
      "step": 7492
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3929222477791734,
      "learning_rate": 4.66084812997315e-06,
      "loss": 0.1716,
      "step": 7493
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5004389583288535,
      "learning_rate": 4.659692402673066e-06,
      "loss": 0.2015,
      "step": 7494
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2789358029101445,
      "learning_rate": 4.6585366936393426e-06,
      "loss": 0.1674,
      "step": 7495
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.251974506077591,
      "learning_rate": 4.657381002934011e-06,
      "loss": 0.1939,
      "step": 7496
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2939403466858708,
      "learning_rate": 4.656225330619109e-06,
      "loss": 0.1747,
      "step": 7497
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3076814191177963,
      "learning_rate": 4.655069676756662e-06,
      "loss": 0.1888,
      "step": 7498
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4023440936505025,
      "learning_rate": 4.653914041408708e-06,
      "loss": 0.1917,
      "step": 7499
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4275234864810196,
      "learning_rate": 4.652758424637271e-06,
      "loss": 0.2178,
      "step": 7500
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3259454583573873,
      "learning_rate": 4.651602826504384e-06,
      "loss": 0.2261,
      "step": 7501
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4680337762354,
      "learning_rate": 4.65044724707207e-06,
      "loss": 0.2325,
      "step": 7502
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4337010140557995,
      "learning_rate": 4.649291686402362e-06,
      "loss": 0.1967,
      "step": 7503
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3486810938849187,
      "learning_rate": 4.648136144557282e-06,
      "loss": 0.2117,
      "step": 7504
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.109743908957312,
      "learning_rate": 4.646980621598857e-06,
      "loss": 0.1537,
      "step": 7505
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4549351604558525,
      "learning_rate": 4.645825117589108e-06,
      "loss": 0.2045,
      "step": 7506
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2968438773634885,
      "learning_rate": 4.644669632590061e-06,
      "loss": 0.1714,
      "step": 7507
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.686711346071377,
      "learning_rate": 4.643514166663735e-06,
      "loss": 0.5402,
      "step": 7508
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2988118235537465,
      "learning_rate": 4.642358719872154e-06,
      "loss": 0.1872,
      "step": 7509
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1427461215374288,
      "learning_rate": 4.641203292277334e-06,
      "loss": 0.1632,
      "step": 7510
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2831724595970118,
      "learning_rate": 4.640047883941297e-06,
      "loss": 0.172,
      "step": 7511
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.865323521539073,
      "learning_rate": 4.638892494926059e-06,
      "loss": 0.6696,
      "step": 7512
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4393047060717468,
      "learning_rate": 4.63773712529364e-06,
      "loss": 0.1904,
      "step": 7513
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3759243712168225,
      "learning_rate": 4.63658177510605e-06,
      "loss": 0.1896,
      "step": 7514
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3481478225622276,
      "learning_rate": 4.6354264444253074e-06,
      "loss": 0.1842,
      "step": 7515
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3884466004113205,
      "learning_rate": 4.634271133313423e-06,
      "loss": 0.1802,
      "step": 7516
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2918102157741091,
      "learning_rate": 4.633115841832415e-06,
      "loss": 0.2138,
      "step": 7517
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3310691111340116,
      "learning_rate": 4.631960570044289e-06,
      "loss": 0.1508,
      "step": 7518
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.264819766548765,
      "learning_rate": 4.630805318011057e-06,
      "loss": 0.5573,
      "step": 7519
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1631998281480738,
      "learning_rate": 4.629650085794728e-06,
      "loss": 0.1952,
      "step": 7520
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3439238083948173,
      "learning_rate": 4.628494873457312e-06,
      "loss": 0.1699,
      "step": 7521
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.842345628272662,
      "learning_rate": 4.627339681060817e-06,
      "loss": 0.5393,
      "step": 7522
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0914234065820243,
      "learning_rate": 4.626184508667245e-06,
      "loss": 0.16,
      "step": 7523
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3198296337502753,
      "learning_rate": 4.625029356338605e-06,
      "loss": 0.1945,
      "step": 7524
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.240577267266923,
      "learning_rate": 4.623874224136899e-06,
      "loss": 0.1719,
      "step": 7525
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.3020901308010515,
      "learning_rate": 4.622719112124131e-06,
      "loss": 0.4178,
      "step": 7526
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.019397175830867,
      "learning_rate": 4.621564020362301e-06,
      "loss": 0.5541,
      "step": 7527
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.101645737169648,
      "learning_rate": 4.620408948913411e-06,
      "loss": 0.1543,
      "step": 7528
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4374458817863476,
      "learning_rate": 4.61925389783946e-06,
      "loss": 0.1734,
      "step": 7529
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3755195847278285,
      "learning_rate": 4.61809886720245e-06,
      "loss": 0.1906,
      "step": 7530
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.30963836133564,
      "learning_rate": 4.616943857064374e-06,
      "loss": 0.2104,
      "step": 7531
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2075327034494507,
      "learning_rate": 4.61578886748723e-06,
      "loss": 0.1746,
      "step": 7532
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.779955664580308,
      "learning_rate": 4.614633898533012e-06,
      "loss": 0.6518,
      "step": 7533
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2242157545446113,
      "learning_rate": 4.613478950263718e-06,
      "loss": 0.2192,
      "step": 7534
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3555237232100728,
      "learning_rate": 4.612324022741336e-06,
      "loss": 0.1843,
      "step": 7535
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4840891814848796,
      "learning_rate": 4.611169116027861e-06,
      "loss": 0.2418,
      "step": 7536
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.306316589654052,
      "learning_rate": 4.610014230185283e-06,
      "loss": 0.2046,
      "step": 7537
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.249244277568884,
      "learning_rate": 4.608859365275592e-06,
      "loss": 0.2019,
      "step": 7538
    },
    {
      "epoch": 0.54,
      "grad_norm": 3.4920018052575506,
      "learning_rate": 4.6077045213607765e-06,
      "loss": 0.6069,
      "step": 7539
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4129609254472701,
      "learning_rate": 4.606549698502824e-06,
      "loss": 0.1677,
      "step": 7540
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2423768338269527,
      "learning_rate": 4.605394896763719e-06,
      "loss": 0.1848,
      "step": 7541
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2695356553963164,
      "learning_rate": 4.6042401162054505e-06,
      "loss": 0.1698,
      "step": 7542
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.096200604788515,
      "learning_rate": 4.6030853568899985e-06,
      "loss": 0.6798,
      "step": 7543
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3630152253698309,
      "learning_rate": 4.601930618879349e-06,
      "loss": 0.1722,
      "step": 7544
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2593746356112543,
      "learning_rate": 4.600775902235481e-06,
      "loss": 0.1632,
      "step": 7545
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.027903368709246,
      "learning_rate": 4.599621207020378e-06,
      "loss": 0.4485,
      "step": 7546
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.374625255429694,
      "learning_rate": 4.598466533296017e-06,
      "loss": 0.2033,
      "step": 7547
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.332674177182077,
      "learning_rate": 4.597311881124378e-06,
      "loss": 0.1971,
      "step": 7548
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.34482185464182,
      "learning_rate": 4.5961572505674366e-06,
      "loss": 0.2033,
      "step": 7549
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.300902800842386,
      "learning_rate": 4.595002641687171e-06,
      "loss": 0.1764,
      "step": 7550
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2058640786208512,
      "learning_rate": 4.593848054545554e-06,
      "loss": 0.1452,
      "step": 7551
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2849828629062505,
      "learning_rate": 4.592693489204558e-06,
      "loss": 0.1982,
      "step": 7552
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4008869405139501,
      "learning_rate": 4.591538945726159e-06,
      "loss": 0.1973,
      "step": 7553
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.304243118961976,
      "learning_rate": 4.590384424172326e-06,
      "loss": 0.2125,
      "step": 7554
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.342644413558855,
      "learning_rate": 4.589229924605031e-06,
      "loss": 0.1614,
      "step": 7555
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.308308718935779,
      "learning_rate": 4.588075447086241e-06,
      "loss": 0.1505,
      "step": 7556
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.265498558405239,
      "learning_rate": 4.586920991677925e-06,
      "loss": 0.1833,
      "step": 7557
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2584525625248717,
      "learning_rate": 4.585766558442048e-06,
      "loss": 0.1877,
      "step": 7558
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.309145710531472,
      "learning_rate": 4.5846121474405785e-06,
      "loss": 0.1564,
      "step": 7559
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3969622179243657,
      "learning_rate": 4.583457758735477e-06,
      "loss": 0.1888,
      "step": 7560
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2793479149031906,
      "learning_rate": 4.582303392388708e-06,
      "loss": 0.1566,
      "step": 7561
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.823089829673408,
      "learning_rate": 4.581149048462234e-06,
      "loss": 0.4043,
      "step": 7562
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2684776190768705,
      "learning_rate": 4.579994727018016e-06,
      "loss": 0.1395,
      "step": 7563
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2576126783819088,
      "learning_rate": 4.578840428118011e-06,
      "loss": 0.183,
      "step": 7564
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2702637318743508,
      "learning_rate": 4.57768615182418e-06,
      "loss": 0.1948,
      "step": 7565
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4805358862960174,
      "learning_rate": 4.576531898198476e-06,
      "loss": 0.1965,
      "step": 7566
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.26378113181666,
      "learning_rate": 4.575377667302861e-06,
      "loss": 0.2153,
      "step": 7567
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6633196945167759,
      "learning_rate": 4.574223459199284e-06,
      "loss": 0.2223,
      "step": 7568
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.376264941339721,
      "learning_rate": 4.573069273949699e-06,
      "loss": 0.194,
      "step": 7569
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2850630917590276,
      "learning_rate": 4.57191511161606e-06,
      "loss": 0.1705,
      "step": 7570
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.511314043309628,
      "learning_rate": 4.570760972260319e-06,
      "loss": 0.1983,
      "step": 7571
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5023694424470495,
      "learning_rate": 4.5696068559444225e-06,
      "loss": 0.1911,
      "step": 7572
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.254463503366863,
      "learning_rate": 4.56845276273032e-06,
      "loss": 0.2016,
      "step": 7573
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6821706405477255,
      "learning_rate": 4.567298692679958e-06,
      "loss": 0.2431,
      "step": 7574
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.198669834798029,
      "learning_rate": 4.566144645855285e-06,
      "loss": 0.1715,
      "step": 7575
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4474491698567924,
      "learning_rate": 4.5649906223182425e-06,
      "loss": 0.1791,
      "step": 7576
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2769803923089442,
      "learning_rate": 4.5638366221307754e-06,
      "loss": 0.1705,
      "step": 7577
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3908191385630573,
      "learning_rate": 4.562682645354825e-06,
      "loss": 0.2149,
      "step": 7578
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2633180333909195,
      "learning_rate": 4.5615286920523345e-06,
      "loss": 0.189,
      "step": 7579
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2960606295975712,
      "learning_rate": 4.56037476228524e-06,
      "loss": 0.1673,
      "step": 7580
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2738513537287657,
      "learning_rate": 4.5592208561154825e-06,
      "loss": 0.2171,
      "step": 7581
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2724456045668626,
      "learning_rate": 4.5580669736049975e-06,
      "loss": 0.2003,
      "step": 7582
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1839303717154426,
      "learning_rate": 4.556913114815724e-06,
      "loss": 0.1692,
      "step": 7583
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4657497201831253,
      "learning_rate": 4.555759279809591e-06,
      "loss": 0.1638,
      "step": 7584
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.563836419321114,
      "learning_rate": 4.554605468648537e-06,
      "loss": 0.5793,
      "step": 7585
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2695155758587455,
      "learning_rate": 4.553451681394492e-06,
      "loss": 0.2016,
      "step": 7586
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.44041059713771,
      "learning_rate": 4.552297918109385e-06,
      "loss": 0.1619,
      "step": 7587
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5426582928806007,
      "learning_rate": 4.55114417885515e-06,
      "loss": 0.2542,
      "step": 7588
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2512214925917666,
      "learning_rate": 4.54999046369371e-06,
      "loss": 0.1885,
      "step": 7589
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.278180505609115,
      "learning_rate": 4.548836772686995e-06,
      "loss": 0.1848,
      "step": 7590
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.485658284115507,
      "learning_rate": 4.54768310589693e-06,
      "loss": 0.1983,
      "step": 7591
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2676608549655977,
      "learning_rate": 4.546529463385439e-06,
      "loss": 0.1561,
      "step": 7592
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2340633501414175,
      "learning_rate": 4.545375845214443e-06,
      "loss": 0.1747,
      "step": 7593
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2386382547596009,
      "learning_rate": 4.544222251445868e-06,
      "loss": 0.1839,
      "step": 7594
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6249778921223967,
      "learning_rate": 4.54306868214163e-06,
      "loss": 0.1797,
      "step": 7595
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.436637589173101,
      "learning_rate": 4.541915137363651e-06,
      "loss": 0.1863,
      "step": 7596
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3826570737009853,
      "learning_rate": 4.5407616171738465e-06,
      "loss": 0.2105,
      "step": 7597
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.392578278075935,
      "learning_rate": 4.539608121634136e-06,
      "loss": 0.1852,
      "step": 7598
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.266900368224982,
      "learning_rate": 4.538454650806429e-06,
      "loss": 0.1965,
      "step": 7599
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.253535870652874,
      "learning_rate": 4.537301204752647e-06,
      "loss": 0.2074,
      "step": 7600
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.294760603910289,
      "learning_rate": 4.5361477835346945e-06,
      "loss": 0.1667,
      "step": 7601
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2931409383464647,
      "learning_rate": 4.534994387214489e-06,
      "loss": 0.2008,
      "step": 7602
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.361316797668219,
      "learning_rate": 4.5338410158539345e-06,
      "loss": 0.1645,
      "step": 7603
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2760861034435151,
      "learning_rate": 4.532687669514944e-06,
      "loss": 0.1299,
      "step": 7604
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2325380237400623,
      "learning_rate": 4.531534348259422e-06,
      "loss": 0.1746,
      "step": 7605
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5145411503839208,
      "learning_rate": 4.530381052149278e-06,
      "loss": 0.1769,
      "step": 7606
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2809818861273423,
      "learning_rate": 4.52922778124641e-06,
      "loss": 0.1699,
      "step": 7607
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3025998291888405,
      "learning_rate": 4.528074535612725e-06,
      "loss": 0.2047,
      "step": 7608
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.7381682788340695,
      "learning_rate": 4.526921315310124e-06,
      "loss": 0.7048,
      "step": 7609
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5126814182320447,
      "learning_rate": 4.52576812040051e-06,
      "loss": 0.1934,
      "step": 7610
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2100335473942259,
      "learning_rate": 4.524614950945776e-06,
      "loss": 0.1834,
      "step": 7611
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.4140016368185035,
      "learning_rate": 4.523461807007825e-06,
      "loss": 0.7099,
      "step": 7612
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1806095883348924,
      "learning_rate": 4.522308688648549e-06,
      "loss": 0.1752,
      "step": 7613
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2539706363369176,
      "learning_rate": 4.521155595929847e-06,
      "loss": 0.2141,
      "step": 7614
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4794814984763651,
      "learning_rate": 4.520002528913609e-06,
      "loss": 0.2121,
      "step": 7615
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2805312934785509,
      "learning_rate": 4.518849487661727e-06,
      "loss": 0.1746,
      "step": 7616
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3741800419092545,
      "learning_rate": 4.517696472236095e-06,
      "loss": 0.1752,
      "step": 7617
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.432854104065085,
      "learning_rate": 4.516543482698599e-06,
      "loss": 0.1679,
      "step": 7618
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.257439467499808,
      "learning_rate": 4.51539051911113e-06,
      "loss": 0.2049,
      "step": 7619
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3339940107768395,
      "learning_rate": 4.514237581535571e-06,
      "loss": 0.2032,
      "step": 7620
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.852518096986699,
      "learning_rate": 4.51308467003381e-06,
      "loss": 0.6243,
      "step": 7621
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3656007815350302,
      "learning_rate": 4.511931784667728e-06,
      "loss": 0.2076,
      "step": 7622
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5777214263900718,
      "learning_rate": 4.510778925499211e-06,
      "loss": 0.2184,
      "step": 7623
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2257423908898823,
      "learning_rate": 4.509626092590136e-06,
      "loss": 0.1719,
      "step": 7624
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.385941713122498,
      "learning_rate": 4.508473286002385e-06,
      "loss": 0.1949,
      "step": 7625
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3782898211041559,
      "learning_rate": 4.507320505797833e-06,
      "loss": 0.21,
      "step": 7626
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3109291431526804,
      "learning_rate": 4.506167752038363e-06,
      "loss": 0.159,
      "step": 7627
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3410471853579888,
      "learning_rate": 4.505015024785843e-06,
      "loss": 0.1748,
      "step": 7628
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3847826119841864,
      "learning_rate": 4.503862324102152e-06,
      "loss": 0.1889,
      "step": 7629
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3148862569750293,
      "learning_rate": 4.502709650049158e-06,
      "loss": 0.1607,
      "step": 7630
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.540855081073755,
      "learning_rate": 4.501557002688738e-06,
      "loss": 0.5838,
      "step": 7631
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.570811409071723,
      "learning_rate": 4.5004043820827544e-06,
      "loss": 0.2174,
      "step": 7632
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.276373187646043,
      "learning_rate": 4.4992517882930805e-06,
      "loss": 0.1573,
      "step": 7633
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3324442199690818,
      "learning_rate": 4.49809922138158e-06,
      "loss": 0.1836,
      "step": 7634
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2190857705188312,
      "learning_rate": 4.496946681410121e-06,
      "loss": 0.1587,
      "step": 7635
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2294148997243335,
      "learning_rate": 4.495794168440564e-06,
      "loss": 0.1709,
      "step": 7636
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2031429042361435,
      "learning_rate": 4.494641682534774e-06,
      "loss": 0.1809,
      "step": 7637
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2322880093262887,
      "learning_rate": 4.49348922375461e-06,
      "loss": 0.1682,
      "step": 7638
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2378188690572838,
      "learning_rate": 4.492336792161934e-06,
      "loss": 0.1555,
      "step": 7639
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.178347627996018,
      "learning_rate": 4.4911843878186e-06,
      "loss": 0.6494,
      "step": 7640
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.618804740303724,
      "learning_rate": 4.490032010786468e-06,
      "loss": 0.4912,
      "step": 7641
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3047088388363703,
      "learning_rate": 4.48887966112739e-06,
      "loss": 0.2243,
      "step": 7642
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.332952914112738,
      "learning_rate": 4.4877273389032235e-06,
      "loss": 0.5274,
      "step": 7643
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.396317172745299,
      "learning_rate": 4.486575044175817e-06,
      "loss": 0.2087,
      "step": 7644
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.83831864704029,
      "learning_rate": 4.485422777007022e-06,
      "loss": 0.6053,
      "step": 7645
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2944023438719365,
      "learning_rate": 4.484270537458688e-06,
      "loss": 0.2014,
      "step": 7646
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3810259300977725,
      "learning_rate": 4.4831183255926645e-06,
      "loss": 0.1871,
      "step": 7647
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4064722014692068,
      "learning_rate": 4.481966141470794e-06,
      "loss": 0.2015,
      "step": 7648
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.528260472698978,
      "learning_rate": 4.480813985154923e-06,
      "loss": 0.6004,
      "step": 7649
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.194047236025288,
      "learning_rate": 4.479661856706896e-06,
      "loss": 0.1484,
      "step": 7650
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3495871399599806,
      "learning_rate": 4.478509756188551e-06,
      "loss": 0.2414,
      "step": 7651
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2853243161128571,
      "learning_rate": 4.477357683661734e-06,
      "loss": 0.1789,
      "step": 7652
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3553009704959187,
      "learning_rate": 4.476205639188277e-06,
      "loss": 0.1771,
      "step": 7653
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.385990867737515,
      "learning_rate": 4.475053622830022e-06,
      "loss": 0.6294,
      "step": 7654
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1536496408226689,
      "learning_rate": 4.473901634648802e-06,
      "loss": 0.1575,
      "step": 7655
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3129452913116322,
      "learning_rate": 4.472749674706454e-06,
      "loss": 0.1753,
      "step": 7656
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7466696746898132,
      "learning_rate": 4.471597743064807e-06,
      "loss": 0.1916,
      "step": 7657
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2692794703412886,
      "learning_rate": 4.470445839785695e-06,
      "loss": 0.1549,
      "step": 7658
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2739099082518146,
      "learning_rate": 4.469293964930945e-06,
      "loss": 0.1983,
      "step": 7659
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1100022327362709,
      "learning_rate": 4.468142118562389e-06,
      "loss": 0.1641,
      "step": 7660
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5158421991195783,
      "learning_rate": 4.466990300741849e-06,
      "loss": 0.2573,
      "step": 7661
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3749292982859778,
      "learning_rate": 4.465838511531153e-06,
      "loss": 0.2253,
      "step": 7662
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3289800539419694,
      "learning_rate": 4.464686750992123e-06,
      "loss": 0.1992,
      "step": 7663
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2374468961601377,
      "learning_rate": 4.463535019186583e-06,
      "loss": 0.1941,
      "step": 7664
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2065443527052542,
      "learning_rate": 4.4623833161763504e-06,
      "loss": 0.1776,
      "step": 7665
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3738863431450612,
      "learning_rate": 4.461231642023247e-06,
      "loss": 0.1824,
      "step": 7666
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2259418504058657,
      "learning_rate": 4.460079996789086e-06,
      "loss": 0.1918,
      "step": 7667
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.130615184541869,
      "learning_rate": 4.458928380535689e-06,
      "loss": 0.5615,
      "step": 7668
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2043430761350364,
      "learning_rate": 4.457776793324865e-06,
      "loss": 0.1846,
      "step": 7669
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3500739793228955,
      "learning_rate": 4.4566252352184295e-06,
      "loss": 0.2588,
      "step": 7670
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.375942822650095,
      "learning_rate": 4.455473706278191e-06,
      "loss": 0.2346,
      "step": 7671
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4918395470846977,
      "learning_rate": 4.454322206565964e-06,
      "loss": 0.2179,
      "step": 7672
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4866011387046485,
      "learning_rate": 4.45317073614355e-06,
      "loss": 0.2003,
      "step": 7673
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.775925502580284,
      "learning_rate": 4.452019295072759e-06,
      "loss": 0.5148,
      "step": 7674
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1478124023425795,
      "learning_rate": 4.450867883415395e-06,
      "loss": 0.1391,
      "step": 7675
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3199512792412464,
      "learning_rate": 4.449716501233264e-06,
      "loss": 0.2183,
      "step": 7676
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5675912938937444,
      "learning_rate": 4.448565148588162e-06,
      "loss": 0.2414,
      "step": 7677
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2442251995241251,
      "learning_rate": 4.447413825541893e-06,
      "loss": 0.1486,
      "step": 7678
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3126986408180328,
      "learning_rate": 4.446262532156255e-06,
      "loss": 0.2054,
      "step": 7679
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1866863748913334,
      "learning_rate": 4.4451112684930424e-06,
      "loss": 0.1577,
      "step": 7680
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2554538463734985,
      "learning_rate": 4.443960034614056e-06,
      "loss": 0.1682,
      "step": 7681
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.091769917561093,
      "learning_rate": 4.442808830581083e-06,
      "loss": 0.1736,
      "step": 7682
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3551510520020444,
      "learning_rate": 4.44165765645592e-06,
      "loss": 0.2204,
      "step": 7683
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3215884841792749,
      "learning_rate": 4.4405065123003545e-06,
      "loss": 0.1812,
      "step": 7684
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.486951679518656,
      "learning_rate": 4.439355398176179e-06,
      "loss": 0.1765,
      "step": 7685
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.357060061908719,
      "learning_rate": 4.438204314145176e-06,
      "loss": 0.1813,
      "step": 7686
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.35465376411796,
      "learning_rate": 4.437053260269134e-06,
      "loss": 0.1777,
      "step": 7687
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4573037479351156,
      "learning_rate": 4.435902236609837e-06,
      "loss": 0.2284,
      "step": 7688
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4884091768505359,
      "learning_rate": 4.434751243229068e-06,
      "loss": 0.2153,
      "step": 7689
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2969874536365351,
      "learning_rate": 4.433600280188605e-06,
      "loss": 0.1948,
      "step": 7690
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1516612674513016,
      "learning_rate": 4.432449347550229e-06,
      "loss": 0.1574,
      "step": 7691
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3876325812048094,
      "learning_rate": 4.431298445375717e-06,
      "loss": 0.225,
      "step": 7692
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.283482116884178,
      "learning_rate": 4.430147573726846e-06,
      "loss": 0.1547,
      "step": 7693
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5369157680070666,
      "learning_rate": 4.428996732665389e-06,
      "loss": 0.2003,
      "step": 7694
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3606996292738782,
      "learning_rate": 4.4278459222531186e-06,
      "loss": 0.1775,
      "step": 7695
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3572762170629935,
      "learning_rate": 4.426695142551806e-06,
      "loss": 0.2054,
      "step": 7696
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.199715928021808,
      "learning_rate": 4.425544393623222e-06,
      "loss": 0.1577,
      "step": 7697
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3356654721674448,
      "learning_rate": 4.424393675529131e-06,
      "loss": 0.1517,
      "step": 7698
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.212667352098118,
      "learning_rate": 4.4232429883313025e-06,
      "loss": 0.1446,
      "step": 7699
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5512404322431161,
      "learning_rate": 4.422092332091497e-06,
      "loss": 0.2168,
      "step": 7700
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3176614496514911,
      "learning_rate": 4.4209417068714815e-06,
      "loss": 0.2102,
      "step": 7701
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3230262363448326,
      "learning_rate": 4.419791112733012e-06,
      "loss": 0.2017,
      "step": 7702
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3877718872339417,
      "learning_rate": 4.418640549737853e-06,
      "loss": 0.1845,
      "step": 7703
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1812623000476177,
      "learning_rate": 4.417490017947757e-06,
      "loss": 0.1626,
      "step": 7704
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.339011679846205,
      "learning_rate": 4.416339517424485e-06,
      "loss": 0.1712,
      "step": 7705
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.456604196950525,
      "learning_rate": 4.415189048229788e-06,
      "loss": 0.1765,
      "step": 7706
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5838390171455192,
      "learning_rate": 4.414038610425421e-06,
      "loss": 0.2243,
      "step": 7707
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.648551389807109,
      "learning_rate": 4.412888204073131e-06,
      "loss": 0.6188,
      "step": 7708
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3215850028136427,
      "learning_rate": 4.411737829234671e-06,
      "loss": 0.193,
      "step": 7709
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.398305488628446,
      "learning_rate": 4.4105874859717865e-06,
      "loss": 0.2211,
      "step": 7710
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2868866505526957,
      "learning_rate": 4.4094371743462265e-06,
      "loss": 0.1759,
      "step": 7711
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3518860568322544,
      "learning_rate": 4.40828689441973e-06,
      "loss": 0.2202,
      "step": 7712
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3471251694251827,
      "learning_rate": 4.407136646254042e-06,
      "loss": 0.2406,
      "step": 7713
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2948724788396258,
      "learning_rate": 4.405986429910906e-06,
      "loss": 0.1811,
      "step": 7714
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2456302034219648,
      "learning_rate": 4.404836245452056e-06,
      "loss": 0.1546,
      "step": 7715
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1099816376976996,
      "learning_rate": 4.403686092939235e-06,
      "loss": 0.1794,
      "step": 7716
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.378698276222049,
      "learning_rate": 4.402535972434172e-06,
      "loss": 0.2076,
      "step": 7717
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3467082509390003,
      "learning_rate": 4.4013858839986074e-06,
      "loss": 0.1444,
      "step": 7718
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4001862795204687,
      "learning_rate": 4.400235827694268e-06,
      "loss": 0.1816,
      "step": 7719
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.204061581459019,
      "learning_rate": 4.399085803582889e-06,
      "loss": 0.5724,
      "step": 7720
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.479328599941078,
      "learning_rate": 4.397935811726195e-06,
      "loss": 0.2071,
      "step": 7721
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.290825754864189,
      "learning_rate": 4.396785852185916e-06,
      "loss": 0.1622,
      "step": 7722
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2936411470425098,
      "learning_rate": 4.395635925023774e-06,
      "loss": 0.1625,
      "step": 7723
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.3736115250499425,
      "learning_rate": 4.3944860303014975e-06,
      "loss": 0.6355,
      "step": 7724
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.42249702890177,
      "learning_rate": 4.3933361680808025e-06,
      "loss": 0.1811,
      "step": 7725
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3360099910398058,
      "learning_rate": 4.392186338423413e-06,
      "loss": 0.2064,
      "step": 7726
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2890383273613972,
      "learning_rate": 4.391036541391046e-06,
      "loss": 0.1537,
      "step": 7727
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3937289595335949,
      "learning_rate": 4.38988677704542e-06,
      "loss": 0.1767,
      "step": 7728
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3668161056545798,
      "learning_rate": 4.388737045448245e-06,
      "loss": 0.1605,
      "step": 7729
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3630058465619108,
      "learning_rate": 4.387587346661239e-06,
      "loss": 0.1607,
      "step": 7730
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1323630667342475,
      "learning_rate": 4.386437680746109e-06,
      "loss": 0.169,
      "step": 7731
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.347872470377837,
      "learning_rate": 4.385288047764571e-06,
      "loss": 0.2148,
      "step": 7732
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3327594204550057,
      "learning_rate": 4.3841384477783245e-06,
      "loss": 0.1572,
      "step": 7733
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3378062657827072,
      "learning_rate": 4.382988880849081e-06,
      "loss": 0.2009,
      "step": 7734
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1823208737770983,
      "learning_rate": 4.3818393470385415e-06,
      "loss": 0.1612,
      "step": 7735
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4422132982690001,
      "learning_rate": 4.380689846408413e-06,
      "loss": 0.1764,
      "step": 7736
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1332765231931248,
      "learning_rate": 4.379540379020391e-06,
      "loss": 0.1631,
      "step": 7737
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1555485927762972,
      "learning_rate": 4.378390944936177e-06,
      "loss": 0.1563,
      "step": 7738
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.337241472251638,
      "learning_rate": 4.3772415442174665e-06,
      "loss": 0.182,
      "step": 7739
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1483837410748234,
      "learning_rate": 4.3760921769259585e-06,
      "loss": 0.1672,
      "step": 7740
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4549152001977281,
      "learning_rate": 4.374942843123341e-06,
      "loss": 0.2017,
      "step": 7741
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3961642380538595,
      "learning_rate": 4.37379354287131e-06,
      "loss": 0.2127,
      "step": 7742
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3451628456972229,
      "learning_rate": 4.372644276231553e-06,
      "loss": 0.1704,
      "step": 7743
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.415441103104087,
      "learning_rate": 4.371495043265759e-06,
      "loss": 0.2084,
      "step": 7744
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.330154176818541,
      "learning_rate": 4.370345844035615e-06,
      "loss": 0.257,
      "step": 7745
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.254171297260292,
      "learning_rate": 4.369196678602803e-06,
      "loss": 0.2006,
      "step": 7746
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2749663368075461,
      "learning_rate": 4.368047547029009e-06,
      "loss": 0.1961,
      "step": 7747
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.105323084636687,
      "learning_rate": 4.366898449375911e-06,
      "loss": 0.1929,
      "step": 7748
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2488997881719994,
      "learning_rate": 4.365749385705191e-06,
      "loss": 0.1454,
      "step": 7749
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2795606981307899,
      "learning_rate": 4.364600356078521e-06,
      "loss": 0.1682,
      "step": 7750
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2876196983727848,
      "learning_rate": 4.363451360557583e-06,
      "loss": 0.1636,
      "step": 7751
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1734999732280171,
      "learning_rate": 4.362302399204044e-06,
      "loss": 0.1787,
      "step": 7752
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.326313846416316,
      "learning_rate": 4.361153472079581e-06,
      "loss": 0.1969,
      "step": 7753
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.8639253908071165,
      "learning_rate": 4.36000457924586e-06,
      "loss": 0.5312,
      "step": 7754
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3786000634223605,
      "learning_rate": 4.3588557207645525e-06,
      "loss": 0.164,
      "step": 7755
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2963185734271767,
      "learning_rate": 4.35770689669732e-06,
      "loss": 0.225,
      "step": 7756
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2929559467648828,
      "learning_rate": 4.356558107105833e-06,
      "loss": 0.1952,
      "step": 7757
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.215835579170114,
      "learning_rate": 4.355409352051748e-06,
      "loss": 0.1226,
      "step": 7758
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2669085324837415,
      "learning_rate": 4.354260631596729e-06,
      "loss": 0.1759,
      "step": 7759
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2025736754959084,
      "learning_rate": 4.353111945802433e-06,
      "loss": 0.2027,
      "step": 7760
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3456872739614776,
      "learning_rate": 4.35196329473052e-06,
      "loss": 0.2108,
      "step": 7761
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3532575574840335,
      "learning_rate": 4.350814678442641e-06,
      "loss": 0.2267,
      "step": 7762
    },
    {
      "epoch": 0.56,
      "grad_norm": 23.413922799173346,
      "learning_rate": 4.349666097000452e-06,
      "loss": 0.6456,
      "step": 7763
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4268280648441312,
      "learning_rate": 4.348517550465602e-06,
      "loss": 0.1927,
      "step": 7764
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5365798544856561,
      "learning_rate": 4.347369038899744e-06,
      "loss": 0.1907,
      "step": 7765
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.327622353099601,
      "learning_rate": 4.3462205623645205e-06,
      "loss": 0.1696,
      "step": 7766
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.402951663062515,
      "learning_rate": 4.345072120921581e-06,
      "loss": 0.192,
      "step": 7767
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2576826856790329,
      "learning_rate": 4.343923714632567e-06,
      "loss": 0.1675,
      "step": 7768
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5436845533994012,
      "learning_rate": 4.342775343559125e-06,
      "loss": 0.1949,
      "step": 7769
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2834247045929448,
      "learning_rate": 4.341627007762888e-06,
      "loss": 0.2001,
      "step": 7770
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3750610398448708,
      "learning_rate": 4.3404787073054995e-06,
      "loss": 0.1624,
      "step": 7771
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.309295656554821,
      "learning_rate": 4.3393304422485925e-06,
      "loss": 0.1844,
      "step": 7772
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.186387836871521,
      "learning_rate": 4.338182212653805e-06,
      "loss": 0.1632,
      "step": 7773
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.150770753829921,
      "learning_rate": 4.337034018582764e-06,
      "loss": 0.6821,
      "step": 7774
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2610148343072487,
      "learning_rate": 4.335885860097106e-06,
      "loss": 0.2073,
      "step": 7775
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2447074238968894,
      "learning_rate": 4.334737737258454e-06,
      "loss": 0.1487,
      "step": 7776
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5440712397083656,
      "learning_rate": 4.333589650128437e-06,
      "loss": 0.1988,
      "step": 7777
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1871687601520655,
      "learning_rate": 4.332441598768684e-06,
      "loss": 0.1262,
      "step": 7778
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.1565367206625226,
      "learning_rate": 4.33129358324081e-06,
      "loss": 0.7563,
      "step": 7779
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2738855503154205,
      "learning_rate": 4.3301456036064415e-06,
      "loss": 0.1503,
      "step": 7780
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4075016927803767,
      "learning_rate": 4.328997659927196e-06,
      "loss": 0.1526,
      "step": 7781
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4648039679916371,
      "learning_rate": 4.3278497522646905e-06,
      "loss": 0.2169,
      "step": 7782
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2573430201257563,
      "learning_rate": 4.326701880680539e-06,
      "loss": 0.2135,
      "step": 7783
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3898256534259048,
      "learning_rate": 4.325554045236357e-06,
      "loss": 0.2042,
      "step": 7784
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3543516767008708,
      "learning_rate": 4.324406245993752e-06,
      "loss": 0.2101,
      "step": 7785
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.226943921383533,
      "learning_rate": 4.323258483014339e-06,
      "loss": 0.575,
      "step": 7786
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4314763589505322,
      "learning_rate": 4.322110756359719e-06,
      "loss": 0.1576,
      "step": 7787
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.409908566610597,
      "learning_rate": 4.320963066091503e-06,
      "loss": 0.2192,
      "step": 7788
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2814594217472992,
      "learning_rate": 4.3198154122712895e-06,
      "loss": 0.185,
      "step": 7789
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.122784697364961,
      "learning_rate": 4.318667794960685e-06,
      "loss": 0.173,
      "step": 7790
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3435530736413335,
      "learning_rate": 4.317520214221284e-06,
      "loss": 0.1928,
      "step": 7791
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.393642304912983,
      "learning_rate": 4.316372670114689e-06,
      "loss": 0.5051,
      "step": 7792
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1967346924798075,
      "learning_rate": 4.315225162702491e-06,
      "loss": 0.1717,
      "step": 7793
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2881015550384576,
      "learning_rate": 4.314077692046288e-06,
      "loss": 0.194,
      "step": 7794
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5559467502100368,
      "learning_rate": 4.312930258207668e-06,
      "loss": 0.2058,
      "step": 7795
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.694538469782157,
      "learning_rate": 4.311782861248223e-06,
      "loss": 0.6859,
      "step": 7796
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2535198148762579,
      "learning_rate": 4.310635501229537e-06,
      "loss": 0.1765,
      "step": 7797
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.367019054527165,
      "learning_rate": 4.3094881782132025e-06,
      "loss": 0.1704,
      "step": 7798
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3147584899923683,
      "learning_rate": 4.308340892260798e-06,
      "loss": 0.2129,
      "step": 7799
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3463262510839447,
      "learning_rate": 4.307193643433907e-06,
      "loss": 0.1665,
      "step": 7800
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3796274317058435,
      "learning_rate": 4.306046431794107e-06,
      "loss": 0.1914,
      "step": 7801
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1781154971752665,
      "learning_rate": 4.304899257402979e-06,
      "loss": 0.1393,
      "step": 7802
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.170996813416576,
      "learning_rate": 4.303752120322096e-06,
      "loss": 0.155,
      "step": 7803
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3239728894785163,
      "learning_rate": 4.302605020613036e-06,
      "loss": 0.1902,
      "step": 7804
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5258568606904732,
      "learning_rate": 4.301457958337365e-06,
      "loss": 0.2113,
      "step": 7805
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2684662439207885,
      "learning_rate": 4.300310933556657e-06,
      "loss": 0.1863,
      "step": 7806
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1479658599919356,
      "learning_rate": 4.299163946332477e-06,
      "loss": 0.1695,
      "step": 7807
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3493721063311346,
      "learning_rate": 4.298016996726393e-06,
      "loss": 0.172,
      "step": 7808
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1644324805896755,
      "learning_rate": 4.29687008479997e-06,
      "loss": 0.1933,
      "step": 7809
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3667846801454973,
      "learning_rate": 4.295723210614764e-06,
      "loss": 0.1986,
      "step": 7810
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.388402057871557,
      "learning_rate": 4.29457637423234e-06,
      "loss": 0.2285,
      "step": 7811
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1189473036902835,
      "learning_rate": 4.2934295757142526e-06,
      "loss": 0.1391,
      "step": 7812
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4363010927709214,
      "learning_rate": 4.29228281512206e-06,
      "loss": 0.226,
      "step": 7813
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.405228917955243,
      "learning_rate": 4.291136092517313e-06,
      "loss": 0.1985,
      "step": 7814
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3095548882209582,
      "learning_rate": 4.289989407961566e-06,
      "loss": 0.1808,
      "step": 7815
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2926219544885713,
      "learning_rate": 4.288842761516364e-06,
      "loss": 0.1778,
      "step": 7816
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.449507286740925,
      "learning_rate": 4.287696153243261e-06,
      "loss": 0.1971,
      "step": 7817
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3323738032018198,
      "learning_rate": 4.286549583203796e-06,
      "loss": 0.1796,
      "step": 7818
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3987228609516833,
      "learning_rate": 4.285403051459516e-06,
      "loss": 0.2065,
      "step": 7819
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4951066627375353,
      "learning_rate": 4.28425655807196e-06,
      "loss": 0.2353,
      "step": 7820
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3420235913424998,
      "learning_rate": 4.283110103102671e-06,
      "loss": 0.188,
      "step": 7821
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6448780035576838,
      "learning_rate": 4.281963686613181e-06,
      "loss": 0.2538,
      "step": 7822
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5207459318515872,
      "learning_rate": 4.280817308665029e-06,
      "loss": 0.2122,
      "step": 7823
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3138608424173452,
      "learning_rate": 4.2796709693197455e-06,
      "loss": 0.1841,
      "step": 7824
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.306407149009988,
      "learning_rate": 4.278524668638865e-06,
      "loss": 0.2036,
      "step": 7825
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.318060315295124,
      "learning_rate": 4.277378406683912e-06,
      "loss": 0.1834,
      "step": 7826
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3468972692614767,
      "learning_rate": 4.276232183516416e-06,
      "loss": 0.1377,
      "step": 7827
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.544812529316365,
      "learning_rate": 4.2750859991979e-06,
      "loss": 0.5618,
      "step": 7828
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2590485770516662,
      "learning_rate": 4.27393985378989e-06,
      "loss": 0.1929,
      "step": 7829
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.169790051404965,
      "learning_rate": 4.272793747353902e-06,
      "loss": 0.6183,
      "step": 7830
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7547877431539443,
      "learning_rate": 4.2716476799514574e-06,
      "loss": 0.2129,
      "step": 7831
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4132170230477592,
      "learning_rate": 4.2705016516440705e-06,
      "loss": 0.1819,
      "step": 7832
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.450838731543012,
      "learning_rate": 4.26935566249326e-06,
      "loss": 0.2044,
      "step": 7833
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.24003940348491,
      "learning_rate": 4.268209712560531e-06,
      "loss": 0.1764,
      "step": 7834
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4179993261516857,
      "learning_rate": 4.2670638019074e-06,
      "loss": 0.2479,
      "step": 7835
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4821622761917188,
      "learning_rate": 4.265917930595371e-06,
      "loss": 0.2044,
      "step": 7836
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2765371592707864,
      "learning_rate": 4.264772098685952e-06,
      "loss": 0.1788,
      "step": 7837
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3862953257444357,
      "learning_rate": 4.263626306240645e-06,
      "loss": 0.2,
      "step": 7838
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1633404260035345,
      "learning_rate": 4.262480553320954e-06,
      "loss": 0.1488,
      "step": 7839
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.241820596330925,
      "learning_rate": 4.261334839988375e-06,
      "loss": 0.1776,
      "step": 7840
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3396847449718594,
      "learning_rate": 4.260189166304407e-06,
      "loss": 0.199,
      "step": 7841
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2939209144666226,
      "learning_rate": 4.259043532330547e-06,
      "loss": 0.2097,
      "step": 7842
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4467526616411694,
      "learning_rate": 4.257897938128286e-06,
      "loss": 0.1672,
      "step": 7843
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1315376499932335,
      "learning_rate": 4.256752383759115e-06,
      "loss": 0.13,
      "step": 7844
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2743685374316147,
      "learning_rate": 4.255606869284523e-06,
      "loss": 0.1548,
      "step": 7845
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5428863183943617,
      "learning_rate": 4.254461394765999e-06,
      "loss": 0.1879,
      "step": 7846
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1800055995918612,
      "learning_rate": 4.253315960265024e-06,
      "loss": 0.176,
      "step": 7847
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5319742339025262,
      "learning_rate": 4.252170565843082e-06,
      "loss": 0.1715,
      "step": 7848
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2124224970404862,
      "learning_rate": 4.251025211561652e-06,
      "loss": 0.1654,
      "step": 7849
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3708294611253513,
      "learning_rate": 4.249879897482215e-06,
      "loss": 0.1508,
      "step": 7850
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3850634602537883,
      "learning_rate": 4.2487346236662435e-06,
      "loss": 0.1941,
      "step": 7851
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3717528097252316,
      "learning_rate": 4.247589390175214e-06,
      "loss": 0.1897,
      "step": 7852
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3860924625725142,
      "learning_rate": 4.246444197070596e-06,
      "loss": 0.1916,
      "step": 7853
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4362901853748111,
      "learning_rate": 4.2452990444138605e-06,
      "loss": 0.192,
      "step": 7854
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3173784577725747,
      "learning_rate": 4.244153932266473e-06,
      "loss": 0.1908,
      "step": 7855
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.337733009455502,
      "learning_rate": 4.243008860689901e-06,
      "loss": 0.1968,
      "step": 7856
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8255030076008663,
      "learning_rate": 4.241863829745604e-06,
      "loss": 0.2001,
      "step": 7857
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3257029755817045,
      "learning_rate": 4.240718839495047e-06,
      "loss": 0.1626,
      "step": 7858
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4366838151419672,
      "learning_rate": 4.239573889999685e-06,
      "loss": 0.2299,
      "step": 7859
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3306153270602634,
      "learning_rate": 4.2384289813209754e-06,
      "loss": 0.1845,
      "step": 7860
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.393184694831203,
      "learning_rate": 4.2372841135203725e-06,
      "loss": 0.193,
      "step": 7861
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2091667154866548,
      "learning_rate": 4.23613928665933e-06,
      "loss": 0.153,
      "step": 7862
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5236115475384011,
      "learning_rate": 4.234994500799293e-06,
      "loss": 0.2106,
      "step": 7863
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1456291973797892,
      "learning_rate": 4.233849756001714e-06,
      "loss": 0.1351,
      "step": 7864
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4553273025869284,
      "learning_rate": 4.232705052328036e-06,
      "loss": 0.191,
      "step": 7865
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.184378848796924,
      "learning_rate": 4.231560389839703e-06,
      "loss": 0.1749,
      "step": 7866
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.19820089275979,
      "learning_rate": 4.230415768598154e-06,
      "loss": 0.1729,
      "step": 7867
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2446325652360506,
      "learning_rate": 4.229271188664831e-06,
      "loss": 0.1654,
      "step": 7868
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2761419371451115,
      "learning_rate": 4.228126650101166e-06,
      "loss": 0.1694,
      "step": 7869
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.407126955601497,
      "learning_rate": 4.226982152968599e-06,
      "loss": 0.5106,
      "step": 7870
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.41692687117347,
      "learning_rate": 4.2258376973285566e-06,
      "loss": 0.1767,
      "step": 7871
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3069620435077771,
      "learning_rate": 4.2246932832424715e-06,
      "loss": 0.1804,
      "step": 7872
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2661105537629833,
      "learning_rate": 4.223548910771772e-06,
      "loss": 0.1603,
      "step": 7873
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3133761922692821,
      "learning_rate": 4.22240457997788e-06,
      "loss": 0.1699,
      "step": 7874
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3405997983644282,
      "learning_rate": 4.2212602909222235e-06,
      "loss": 0.2206,
      "step": 7875
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3197282998639068,
      "learning_rate": 4.220116043666219e-06,
      "loss": 0.1965,
      "step": 7876
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1809412372634818,
      "learning_rate": 4.218971838271289e-06,
      "loss": 0.1452,
      "step": 7877
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5855816284552255,
      "learning_rate": 4.217827674798845e-06,
      "loss": 0.2504,
      "step": 7878
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3674772086218194,
      "learning_rate": 4.216683553310308e-06,
      "loss": 0.192,
      "step": 7879
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2389672126651905,
      "learning_rate": 4.2155394738670814e-06,
      "loss": 0.175,
      "step": 7880
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3784447823987964,
      "learning_rate": 4.214395436530582e-06,
      "loss": 0.1678,
      "step": 7881
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.769303510929099,
      "learning_rate": 4.213251441362214e-06,
      "loss": 0.5595,
      "step": 7882
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5267078788007482,
      "learning_rate": 4.212107488423383e-06,
      "loss": 0.2294,
      "step": 7883
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3939647565274438,
      "learning_rate": 4.210963577775492e-06,
      "loss": 0.1863,
      "step": 7884
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2255321542013293,
      "learning_rate": 4.209819709479942e-06,
      "loss": 0.1586,
      "step": 7885
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3430518485877314,
      "learning_rate": 4.208675883598129e-06,
      "loss": 0.2046,
      "step": 7886
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3353008012062375,
      "learning_rate": 4.207532100191454e-06,
      "loss": 0.1894,
      "step": 7887
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0338987165288436,
      "learning_rate": 4.206388359321304e-06,
      "loss": 0.0915,
      "step": 7888
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.364735435305999,
      "learning_rate": 4.205244661049076e-06,
      "loss": 0.5883,
      "step": 7889
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3327251070751325,
      "learning_rate": 4.204101005436156e-06,
      "loss": 0.1891,
      "step": 7890
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3704925212680386,
      "learning_rate": 4.202957392543934e-06,
      "loss": 0.1904,
      "step": 7891
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1906147218188272,
      "learning_rate": 4.2018138224337915e-06,
      "loss": 0.1583,
      "step": 7892
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.277297639707543,
      "learning_rate": 4.200670295167112e-06,
      "loss": 0.1768,
      "step": 7893
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3005763464178688,
      "learning_rate": 4.199526810805274e-06,
      "loss": 0.2023,
      "step": 7894
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3231971534917217,
      "learning_rate": 4.1983833694096595e-06,
      "loss": 0.1778,
      "step": 7895
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3513145261692916,
      "learning_rate": 4.19723997104164e-06,
      "loss": 0.1629,
      "step": 7896
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1766817696503347,
      "learning_rate": 4.19609661576259e-06,
      "loss": 0.1388,
      "step": 7897
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3214014213663674,
      "learning_rate": 4.194953303633877e-06,
      "loss": 0.1861,
      "step": 7898
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5039734449338713,
      "learning_rate": 4.193810034716877e-06,
      "loss": 0.2223,
      "step": 7899
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1293393454396783,
      "learning_rate": 4.192666809072948e-06,
      "loss": 0.156,
      "step": 7900
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2753459212792047,
      "learning_rate": 4.19152362676346e-06,
      "loss": 0.1708,
      "step": 7901
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4984188778770722,
      "learning_rate": 4.19038048784977e-06,
      "loss": 0.1941,
      "step": 7902
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4545772199429274,
      "learning_rate": 4.18923739239324e-06,
      "loss": 0.2005,
      "step": 7903
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.4479870673565096,
      "learning_rate": 4.1880943404552234e-06,
      "loss": 0.6561,
      "step": 7904
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.280202969892627,
      "learning_rate": 4.186951332097078e-06,
      "loss": 0.1854,
      "step": 7905
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4854053724482144,
      "learning_rate": 4.185808367380158e-06,
      "loss": 0.2417,
      "step": 7906
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3396469893932181,
      "learning_rate": 4.184665446365807e-06,
      "loss": 0.1784,
      "step": 7907
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3376596206176443,
      "learning_rate": 4.183522569115377e-06,
      "loss": 0.2057,
      "step": 7908
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2703552540015433,
      "learning_rate": 4.1823797356902106e-06,
      "loss": 0.197,
      "step": 7909
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4920973790879564,
      "learning_rate": 4.181236946151654e-06,
      "loss": 0.1911,
      "step": 7910
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2324281286720502,
      "learning_rate": 4.1800942005610434e-06,
      "loss": 0.1991,
      "step": 7911
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2509754125826382,
      "learning_rate": 4.17895149897972e-06,
      "loss": 0.1887,
      "step": 7912
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.52545610270568,
      "learning_rate": 4.177808841469017e-06,
      "loss": 0.1904,
      "step": 7913
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.241717214721998,
      "learning_rate": 4.176666228090271e-06,
      "loss": 0.166,
      "step": 7914
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.433130260215562,
      "learning_rate": 4.175523658904808e-06,
      "loss": 0.2296,
      "step": 7915
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4693810354116887,
      "learning_rate": 4.174381133973962e-06,
      "loss": 0.1594,
      "step": 7916
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2478683249709333,
      "learning_rate": 4.173238653359055e-06,
      "loss": 0.198,
      "step": 7917
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7727414218985758,
      "learning_rate": 4.172096217121415e-06,
      "loss": 0.2117,
      "step": 7918
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2256931107201692,
      "learning_rate": 4.170953825322359e-06,
      "loss": 0.1873,
      "step": 7919
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2817622715348236,
      "learning_rate": 4.1698114780232085e-06,
      "loss": 0.1991,
      "step": 7920
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4205988934077478,
      "learning_rate": 4.168669175285279e-06,
      "loss": 0.2008,
      "step": 7921
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.817551560208983,
      "learning_rate": 4.167526917169887e-06,
      "loss": 0.5361,
      "step": 7922
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.6471396835925685,
      "learning_rate": 4.166384703738341e-06,
      "loss": 0.6133,
      "step": 7923
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4071471051372368,
      "learning_rate": 4.165242535051955e-06,
      "loss": 0.204,
      "step": 7924
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4788873020805953,
      "learning_rate": 4.164100411172029e-06,
      "loss": 0.2295,
      "step": 7925
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.5855535231841245,
      "learning_rate": 4.162958332159877e-06,
      "loss": 0.4392,
      "step": 7926
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.404526040877423,
      "learning_rate": 4.161816298076794e-06,
      "loss": 0.1915,
      "step": 7927
    },
    {
      "epoch": 0.57,
      "grad_norm": 12.796283605702229,
      "learning_rate": 4.160674308984081e-06,
      "loss": 0.502,
      "step": 7928
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3115609994851167,
      "learning_rate": 4.159532364943038e-06,
      "loss": 0.2264,
      "step": 7929
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3253801997827823,
      "learning_rate": 4.15839046601496e-06,
      "loss": 0.2208,
      "step": 7930
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1734463179433283,
      "learning_rate": 4.157248612261136e-06,
      "loss": 0.1474,
      "step": 7931
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.483046417977614,
      "learning_rate": 4.15610680374286e-06,
      "loss": 0.1966,
      "step": 7932
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.313854712878686,
      "learning_rate": 4.154965040521416e-06,
      "loss": 0.203,
      "step": 7933
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2791878050048069,
      "learning_rate": 4.153823322658094e-06,
      "loss": 0.183,
      "step": 7934
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2983990141004578,
      "learning_rate": 4.152681650214173e-06,
      "loss": 0.1822,
      "step": 7935
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.406445226434695,
      "learning_rate": 4.151540023250934e-06,
      "loss": 0.1849,
      "step": 7936
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.251403758514166,
      "learning_rate": 4.150398441829658e-06,
      "loss": 0.1879,
      "step": 7937
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.434084048647608,
      "learning_rate": 4.149256906011616e-06,
      "loss": 0.213,
      "step": 7938
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.54561678851188,
      "learning_rate": 4.148115415858086e-06,
      "loss": 0.1986,
      "step": 7939
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3178779597857702,
      "learning_rate": 4.146973971430333e-06,
      "loss": 0.1786,
      "step": 7940
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5637452029816388,
      "learning_rate": 4.1458325727896305e-06,
      "loss": 0.2245,
      "step": 7941
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5246863484283584,
      "learning_rate": 4.1446912199972414e-06,
      "loss": 0.2243,
      "step": 7942
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2428416470249841,
      "learning_rate": 4.143549913114431e-06,
      "loss": 0.1641,
      "step": 7943
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5906078498177274,
      "learning_rate": 4.142408652202458e-06,
      "loss": 0.2085,
      "step": 7944
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.268357267014983,
      "learning_rate": 4.141267437322581e-06,
      "loss": 0.5253,
      "step": 7945
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3334862992255274,
      "learning_rate": 4.140126268536055e-06,
      "loss": 0.1982,
      "step": 7946
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2054809623702682,
      "learning_rate": 4.1389851459041385e-06,
      "loss": 0.1556,
      "step": 7947
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3285097239945738,
      "learning_rate": 4.137844069488076e-06,
      "loss": 0.2055,
      "step": 7948
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3500727721884291,
      "learning_rate": 4.136703039349119e-06,
      "loss": 0.1687,
      "step": 7949
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3049991646383885,
      "learning_rate": 4.135562055548513e-06,
      "loss": 0.1816,
      "step": 7950
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4075096808638043,
      "learning_rate": 4.1344211181475035e-06,
      "loss": 0.1868,
      "step": 7951
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3382121044835877,
      "learning_rate": 4.133280227207328e-06,
      "loss": 0.1757,
      "step": 7952
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.69681926087116,
      "learning_rate": 4.132139382789226e-06,
      "loss": 0.3713,
      "step": 7953
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2100555266042954,
      "learning_rate": 4.130998584954435e-06,
      "loss": 0.1843,
      "step": 7954
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.700673045415729,
      "learning_rate": 4.129857833764189e-06,
      "loss": 0.4342,
      "step": 7955
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5162515140724764,
      "learning_rate": 4.128717129279715e-06,
      "loss": 0.2077,
      "step": 7956
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.321707210297209,
      "learning_rate": 4.127576471562246e-06,
      "loss": 0.1821,
      "step": 7957
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.494633770501327,
      "learning_rate": 4.126435860673004e-06,
      "loss": 0.2127,
      "step": 7958
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.392136285695771,
      "learning_rate": 4.1252952966732176e-06,
      "loss": 0.1852,
      "step": 7959
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9601008314685433,
      "learning_rate": 4.124154779624101e-06,
      "loss": 0.1874,
      "step": 7960
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1478828898684157,
      "learning_rate": 4.123014309586879e-06,
      "loss": 0.1701,
      "step": 7961
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.175464910557783,
      "learning_rate": 4.121873886622763e-06,
      "loss": 0.1606,
      "step": 7962
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.523393907111342,
      "learning_rate": 4.120733510792971e-06,
      "loss": 0.234,
      "step": 7963
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2525333325543262,
      "learning_rate": 4.119593182158707e-06,
      "loss": 0.1563,
      "step": 7964
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6693110910218578,
      "learning_rate": 4.118452900781187e-06,
      "loss": 0.1715,
      "step": 7965
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4175473564526528,
      "learning_rate": 4.11731266672161e-06,
      "loss": 0.1551,
      "step": 7966
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.0092382877351485,
      "learning_rate": 4.116172480041185e-06,
      "loss": 0.5857,
      "step": 7967
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.87387150018845,
      "learning_rate": 4.115032340801108e-06,
      "loss": 0.6907,
      "step": 7968
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3489692214501614,
      "learning_rate": 4.1138922490625795e-06,
      "loss": 0.1971,
      "step": 7969
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2338561327277546,
      "learning_rate": 4.112752204886796e-06,
      "loss": 0.1704,
      "step": 7970
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0519680459001135,
      "learning_rate": 4.111612208334947e-06,
      "loss": 0.1503,
      "step": 7971
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2340214520677595,
      "learning_rate": 4.110472259468228e-06,
      "loss": 0.1712,
      "step": 7972
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3488628829857716,
      "learning_rate": 4.109332358347823e-06,
      "loss": 0.158,
      "step": 7973
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2202628139469027,
      "learning_rate": 4.108192505034918e-06,
      "loss": 0.1622,
      "step": 7974
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4482880537063456,
      "learning_rate": 4.107052699590697e-06,
      "loss": 0.2027,
      "step": 7975
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2679028821731093,
      "learning_rate": 4.10591294207634e-06,
      "loss": 0.1843,
      "step": 7976
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1605989004383006,
      "learning_rate": 4.104773232553023e-06,
      "loss": 0.1754,
      "step": 7977
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2354085821261411,
      "learning_rate": 4.103633571081923e-06,
      "loss": 0.2066,
      "step": 7978
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3146859083101374,
      "learning_rate": 4.102493957724212e-06,
      "loss": 0.1649,
      "step": 7979
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3926103391711135,
      "learning_rate": 4.101354392541061e-06,
      "loss": 0.1752,
      "step": 7980
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4399301264220867,
      "learning_rate": 4.100214875593634e-06,
      "loss": 0.1985,
      "step": 7981
    },
    {
      "epoch": 0.57,
      "grad_norm": 10.434037916203813,
      "learning_rate": 4.0990754069430986e-06,
      "loss": 0.4859,
      "step": 7982
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.7090365227332995,
      "learning_rate": 4.097935986650615e-06,
      "loss": 0.5872,
      "step": 7983
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2298491572833825,
      "learning_rate": 4.096796614777347e-06,
      "loss": 0.1535,
      "step": 7984
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2930728204229172,
      "learning_rate": 4.095657291384445e-06,
      "loss": 0.1666,
      "step": 7985
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3141402802806195,
      "learning_rate": 4.094518016533069e-06,
      "loss": 0.1949,
      "step": 7986
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2227180620732314,
      "learning_rate": 4.093378790284368e-06,
      "loss": 0.22,
      "step": 7987
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1940926963020462,
      "learning_rate": 4.092239612699492e-06,
      "loss": 0.1886,
      "step": 7988
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2099501678305673,
      "learning_rate": 4.0911004838395865e-06,
      "loss": 0.2041,
      "step": 7989
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5610292457570594,
      "learning_rate": 4.089961403765797e-06,
      "loss": 0.1967,
      "step": 7990
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4452804821195513,
      "learning_rate": 4.088822372539263e-06,
      "loss": 0.1708,
      "step": 7991
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4017633775219112,
      "learning_rate": 4.0876833902211255e-06,
      "loss": 0.1946,
      "step": 7992
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3602119672908588,
      "learning_rate": 4.086544456872518e-06,
      "loss": 0.2035,
      "step": 7993
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2365785098726378,
      "learning_rate": 4.085405572554576e-06,
      "loss": 0.1783,
      "step": 7994
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3170550956466496,
      "learning_rate": 4.084266737328428e-06,
      "loss": 0.2118,
      "step": 7995
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2162572244941738,
      "learning_rate": 4.083127951255205e-06,
      "loss": 0.151,
      "step": 7996
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3777326752473358,
      "learning_rate": 4.0819892143960295e-06,
      "loss": 0.2027,
      "step": 7997
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.389027794930302,
      "learning_rate": 4.080850526812028e-06,
      "loss": 0.5152,
      "step": 7998
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5275908080749763,
      "learning_rate": 4.079711888564317e-06,
      "loss": 0.2105,
      "step": 7999
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4858716107749628,
      "learning_rate": 4.078573299714014e-06,
      "loss": 0.2131,
      "step": 8000
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3348155744430208,
      "learning_rate": 4.07743476032224e-06,
      "loss": 0.1812,
      "step": 8001
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3260179942010921,
      "learning_rate": 4.076296270450099e-06,
      "loss": 0.1723,
      "step": 8002
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3173780937817325,
      "learning_rate": 4.075157830158708e-06,
      "loss": 0.192,
      "step": 8003
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2736999158418605,
      "learning_rate": 4.074019439509168e-06,
      "loss": 0.1812,
      "step": 8004
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5648938691796492,
      "learning_rate": 4.072881098562587e-06,
      "loss": 0.2136,
      "step": 8005
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.375060297619061,
      "learning_rate": 4.071742807380064e-06,
      "loss": 0.7304,
      "step": 8006
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2253732040821348,
      "learning_rate": 4.070604566022701e-06,
      "loss": 0.1421,
      "step": 8007
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2496259566027623,
      "learning_rate": 4.06946637455159e-06,
      "loss": 0.2016,
      "step": 8008
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5253626822798474,
      "learning_rate": 4.06832823302783e-06,
      "loss": 0.1992,
      "step": 8009
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.349783521995507,
      "learning_rate": 4.067190141512507e-06,
      "loss": 0.1908,
      "step": 8010
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.585683000418182,
      "learning_rate": 4.066052100066714e-06,
      "loss": 0.7002,
      "step": 8011
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.343799989317747,
      "learning_rate": 4.0649141087515314e-06,
      "loss": 0.2479,
      "step": 8012
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4227749973611183,
      "learning_rate": 4.0637761676280466e-06,
      "loss": 0.1886,
      "step": 8013
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.353880241541804,
      "learning_rate": 4.062638276757335e-06,
      "loss": 0.2051,
      "step": 8014
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4603702696167313,
      "learning_rate": 4.0615004362004794e-06,
      "loss": 0.1788,
      "step": 8015
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2036066407916044,
      "learning_rate": 4.06036264601855e-06,
      "loss": 0.1668,
      "step": 8016
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3950374878702316,
      "learning_rate": 4.059224906272622e-06,
      "loss": 0.1853,
      "step": 8017
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4212078004498507,
      "learning_rate": 4.0580872170237625e-06,
      "loss": 0.1665,
      "step": 8018
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4003850997340643,
      "learning_rate": 4.056949578333042e-06,
      "loss": 0.2202,
      "step": 8019
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4151876329871242,
      "learning_rate": 4.055811990261518e-06,
      "loss": 0.1626,
      "step": 8020
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2139987561570011,
      "learning_rate": 4.054674452870257e-06,
      "loss": 0.1488,
      "step": 8021
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1906911161957223,
      "learning_rate": 4.053536966220315e-06,
      "loss": 0.1832,
      "step": 8022
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3537434961775903,
      "learning_rate": 4.05239953037275e-06,
      "loss": 0.2104,
      "step": 8023
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2421183430747667,
      "learning_rate": 4.051262145388612e-06,
      "loss": 0.184,
      "step": 8024
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.085963051612594,
      "learning_rate": 4.050124811328954e-06,
      "loss": 0.4591,
      "step": 8025
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3012486911818981,
      "learning_rate": 4.048987528254821e-06,
      "loss": 0.1789,
      "step": 8026
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1895873176699898,
      "learning_rate": 4.047850296227261e-06,
      "loss": 0.1828,
      "step": 8027
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1096143505769158,
      "learning_rate": 4.046713115307314e-06,
      "loss": 0.1436,
      "step": 8028
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2968522325621459,
      "learning_rate": 4.04557598555602e-06,
      "loss": 0.1793,
      "step": 8029
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4138172044396728,
      "learning_rate": 4.044438907034414e-06,
      "loss": 0.1963,
      "step": 8030
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.897422582296905,
      "learning_rate": 4.043301879803533e-06,
      "loss": 0.6714,
      "step": 8031
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1567264509301127,
      "learning_rate": 4.042164903924405e-06,
      "loss": 0.1605,
      "step": 8032
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2698959271277395,
      "learning_rate": 4.041027979458058e-06,
      "loss": 0.1757,
      "step": 8033
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3540541364026093,
      "learning_rate": 4.039891106465522e-06,
      "loss": 0.1885,
      "step": 8034
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.363769317103242,
      "learning_rate": 4.038754285007816e-06,
      "loss": 0.1741,
      "step": 8035
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2921362979153503,
      "learning_rate": 4.037617515145963e-06,
      "loss": 0.2098,
      "step": 8036
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0576040317485342,
      "learning_rate": 4.036480796940976e-06,
      "loss": 0.152,
      "step": 8037
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2185878074217285,
      "learning_rate": 4.035344130453874e-06,
      "loss": 0.2024,
      "step": 8038
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.281999057338327,
      "learning_rate": 4.034207515745665e-06,
      "loss": 0.2045,
      "step": 8039
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.653684805842123,
      "learning_rate": 4.033070952877362e-06,
      "loss": 0.6067,
      "step": 8040
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2625739425973685,
      "learning_rate": 4.031934441909966e-06,
      "loss": 0.1998,
      "step": 8041
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1989897813654178,
      "learning_rate": 4.030797982904485e-06,
      "loss": 0.158,
      "step": 8042
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2853595408482956,
      "learning_rate": 4.029661575921916e-06,
      "loss": 0.159,
      "step": 8043
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.269686840951969,
      "learning_rate": 4.028525221023262e-06,
      "loss": 0.1937,
      "step": 8044
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3806153845902618,
      "learning_rate": 4.027388918269511e-06,
      "loss": 0.1678,
      "step": 8045
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.862354304764796,
      "learning_rate": 4.026252667721661e-06,
      "loss": 0.6053,
      "step": 8046
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1862540278605633,
      "learning_rate": 4.025116469440698e-06,
      "loss": 0.186,
      "step": 8047
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2753619649190167,
      "learning_rate": 4.023980323487612e-06,
      "loss": 0.1542,
      "step": 8048
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0770854449796035,
      "learning_rate": 4.022844229923383e-06,
      "loss": 0.1578,
      "step": 8049
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.169596180769709,
      "learning_rate": 4.021708188808994e-06,
      "loss": 0.7305,
      "step": 8050
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3479432669443594,
      "learning_rate": 4.0205722002054225e-06,
      "loss": 0.179,
      "step": 8051
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4178520896072546,
      "learning_rate": 4.019436264173646e-06,
      "loss": 0.202,
      "step": 8052
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3796429959079035,
      "learning_rate": 4.018300380774633e-06,
      "loss": 0.1599,
      "step": 8053
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.027281622390009,
      "learning_rate": 4.017164550069357e-06,
      "loss": 0.6118,
      "step": 8054
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2433805319592304,
      "learning_rate": 4.016028772118783e-06,
      "loss": 0.1677,
      "step": 8055
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.293486956367564,
      "learning_rate": 4.014893046983876e-06,
      "loss": 0.171,
      "step": 8056
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3338623027846672,
      "learning_rate": 4.013757374725596e-06,
      "loss": 0.186,
      "step": 8057
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4656626020410528,
      "learning_rate": 4.012621755404901e-06,
      "loss": 0.1975,
      "step": 8058
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2284020301944203,
      "learning_rate": 4.011486189082749e-06,
      "loss": 0.1739,
      "step": 8059
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3087009547084283,
      "learning_rate": 4.010350675820091e-06,
      "loss": 0.1886,
      "step": 8060
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.200274072139302,
      "learning_rate": 4.009215215677876e-06,
      "loss": 0.1684,
      "step": 8061
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3350982159954299,
      "learning_rate": 4.0080798087170534e-06,
      "loss": 0.209,
      "step": 8062
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2955833773718224,
      "learning_rate": 4.006944454998564e-06,
      "loss": 0.179,
      "step": 8063
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.252919418797698,
      "learning_rate": 4.005809154583351e-06,
      "loss": 0.1873,
      "step": 8064
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.296132965140545,
      "learning_rate": 4.004673907532356e-06,
      "loss": 0.1716,
      "step": 8065
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3375059727991874,
      "learning_rate": 4.003538713906509e-06,
      "loss": 0.1872,
      "step": 8066
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4604908562601489,
      "learning_rate": 4.002403573766746e-06,
      "loss": 0.1937,
      "step": 8067
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.225260264180561,
      "learning_rate": 4.001268487173995e-06,
      "loss": 0.1508,
      "step": 8068
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.523383920068594,
      "learning_rate": 4.000133454189185e-06,
      "loss": 0.2159,
      "step": 8069
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3131023549424223,
      "learning_rate": 3.998998474873239e-06,
      "loss": 0.1708,
      "step": 8070
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.182638511348339,
      "learning_rate": 3.997863549287077e-06,
      "loss": 0.1372,
      "step": 8071
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4261854727241108,
      "learning_rate": 3.996728677491619e-06,
      "loss": 0.1828,
      "step": 8072
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4427880523211525,
      "learning_rate": 3.995593859547781e-06,
      "loss": 0.2306,
      "step": 8073
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.727590997230231,
      "learning_rate": 3.9944590955164716e-06,
      "loss": 0.5859,
      "step": 8074
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5250499473045989,
      "learning_rate": 3.993324385458605e-06,
      "loss": 0.1999,
      "step": 8075
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3766032782187008,
      "learning_rate": 3.992189729435085e-06,
      "loss": 0.1819,
      "step": 8076
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3010079284543583,
      "learning_rate": 3.991055127506819e-06,
      "loss": 0.1575,
      "step": 8077
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2302796495627735,
      "learning_rate": 3.989920579734703e-06,
      "loss": 0.1766,
      "step": 8078
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2718770805178725,
      "learning_rate": 3.988786086179638e-06,
      "loss": 0.189,
      "step": 8079
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1623334072052143,
      "learning_rate": 3.987651646902518e-06,
      "loss": 0.1333,
      "step": 8080
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.783985448173852,
      "learning_rate": 3.9865172619642374e-06,
      "loss": 0.6099,
      "step": 8081
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2905100679672514,
      "learning_rate": 3.985382931425682e-06,
      "loss": 0.188,
      "step": 8082
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4167772549320934,
      "learning_rate": 3.984248655347741e-06,
      "loss": 0.2053,
      "step": 8083
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.625763945399266,
      "learning_rate": 3.983114433791297e-06,
      "loss": 0.6303,
      "step": 8084
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5901993326178454,
      "learning_rate": 3.9819802668172305e-06,
      "loss": 0.2685,
      "step": 8085
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4726484418808994,
      "learning_rate": 3.9808461544864176e-06,
      "loss": 0.2064,
      "step": 8086
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4604875050312616,
      "learning_rate": 3.979712096859735e-06,
      "loss": 0.1383,
      "step": 8087
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1755201492600569,
      "learning_rate": 3.9785780939980525e-06,
      "loss": 0.1239,
      "step": 8088
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0476284442546069,
      "learning_rate": 3.977444145962243e-06,
      "loss": 0.1468,
      "step": 8089
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.536942637421289,
      "learning_rate": 3.976310252813167e-06,
      "loss": 0.6333,
      "step": 8090
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3793193177642775,
      "learning_rate": 3.97517641461169e-06,
      "loss": 0.2024,
      "step": 8091
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4296887192598908,
      "learning_rate": 3.974042631418671e-06,
      "loss": 0.1646,
      "step": 8092
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4151675425742891,
      "learning_rate": 3.97290890329497e-06,
      "loss": 0.2279,
      "step": 8093
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.47985618961368,
      "learning_rate": 3.971775230301437e-06,
      "loss": 0.4685,
      "step": 8094
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.3574708188182463,
      "learning_rate": 3.970641612498926e-06,
      "loss": 0.4977,
      "step": 8095
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2936997172404536,
      "learning_rate": 3.969508049948281e-06,
      "loss": 0.1601,
      "step": 8096
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2838259096078106,
      "learning_rate": 3.968374542710351e-06,
      "loss": 0.1886,
      "step": 8097
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4768638704511474,
      "learning_rate": 3.967241090845979e-06,
      "loss": 0.1918,
      "step": 8098
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3538735998815188,
      "learning_rate": 3.966107694416e-06,
      "loss": 0.1917,
      "step": 8099
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2488067948452406,
      "learning_rate": 3.964974353481254e-06,
      "loss": 0.2131,
      "step": 8100
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.353988965058894,
      "learning_rate": 3.963841068102571e-06,
      "loss": 0.2121,
      "step": 8101
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.653108048282838,
      "learning_rate": 3.962707838340786e-06,
      "loss": 0.5144,
      "step": 8102
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.117135606476216,
      "learning_rate": 3.961574664256721e-06,
      "loss": 0.5411,
      "step": 8103
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4892678489001696,
      "learning_rate": 3.960441545911205e-06,
      "loss": 0.2044,
      "step": 8104
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0843689378064763,
      "learning_rate": 3.959308483365054e-06,
      "loss": 0.146,
      "step": 8105
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2234665936952536,
      "learning_rate": 3.95817547667909e-06,
      "loss": 0.1346,
      "step": 8106
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5504714741171117,
      "learning_rate": 3.957042525914128e-06,
      "loss": 0.2416,
      "step": 8107
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3269004982761294,
      "learning_rate": 3.9559096311309805e-06,
      "loss": 0.1574,
      "step": 8108
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.542983227846797,
      "learning_rate": 3.9547767923904546e-06,
      "loss": 0.7287,
      "step": 8109
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2982295122789667,
      "learning_rate": 3.953644009753358e-06,
      "loss": 0.2008,
      "step": 8110
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3836076696045754,
      "learning_rate": 3.952511283280494e-06,
      "loss": 0.1621,
      "step": 8111
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3206727494925798,
      "learning_rate": 3.951378613032665e-06,
      "loss": 0.199,
      "step": 8112
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.33267583373924,
      "learning_rate": 3.950245999070664e-06,
      "loss": 0.2091,
      "step": 8113
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3856570162995032,
      "learning_rate": 3.949113441455288e-06,
      "loss": 0.1954,
      "step": 8114
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2798782490966174,
      "learning_rate": 3.947980940247326e-06,
      "loss": 0.1779,
      "step": 8115
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2752247404567465,
      "learning_rate": 3.94684849550757e-06,
      "loss": 0.1585,
      "step": 8116
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3078870380164824,
      "learning_rate": 3.9457161072968015e-06,
      "loss": 0.1768,
      "step": 8117
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.431809603015343,
      "learning_rate": 3.944583775675804e-06,
      "loss": 0.1829,
      "step": 8118
    },
    {
      "epoch": 0.58,
      "grad_norm": 12.636393874353214,
      "learning_rate": 3.943451500705357e-06,
      "loss": 0.6949,
      "step": 8119
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3824909787490212,
      "learning_rate": 3.942319282446236e-06,
      "loss": 0.2046,
      "step": 8120
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3500255504885168,
      "learning_rate": 3.9411871209592135e-06,
      "loss": 0.1837,
      "step": 8121
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5083475443041952,
      "learning_rate": 3.94005501630506e-06,
      "loss": 0.2156,
      "step": 8122
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2872850366754804,
      "learning_rate": 3.938922968544542e-06,
      "loss": 0.1778,
      "step": 8123
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.255649672017287,
      "learning_rate": 3.937790977738425e-06,
      "loss": 0.1632,
      "step": 8124
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.412161209826929,
      "learning_rate": 3.936659043947467e-06,
      "loss": 0.5399,
      "step": 8125
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.145448645996042,
      "learning_rate": 3.935527167232427e-06,
      "loss": 0.1592,
      "step": 8126
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4601783110998678,
      "learning_rate": 3.9343953476540595e-06,
      "loss": 0.2123,
      "step": 8127
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2931321748575118,
      "learning_rate": 3.933263585273117e-06,
      "loss": 0.1938,
      "step": 8128
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.3386220986459625,
      "learning_rate": 3.932131880150348e-06,
      "loss": 0.4618,
      "step": 8129
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3137978154310215,
      "learning_rate": 3.931000232346496e-06,
      "loss": 0.1815,
      "step": 8130
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1551659926038138,
      "learning_rate": 3.929868641922306e-06,
      "loss": 0.1598,
      "step": 8131
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2788923538545707,
      "learning_rate": 3.928737108938514e-06,
      "loss": 0.1423,
      "step": 8132
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1302259426939052,
      "learning_rate": 3.92760563345586e-06,
      "loss": 0.1649,
      "step": 8133
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1969739945177365,
      "learning_rate": 3.926474215535074e-06,
      "loss": 0.1768,
      "step": 8134
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3167420710310629,
      "learning_rate": 3.925342855236888e-06,
      "loss": 0.1901,
      "step": 8135
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3594382073605924,
      "learning_rate": 3.924211552622026e-06,
      "loss": 0.2088,
      "step": 8136
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3043423941741843,
      "learning_rate": 3.923080307751218e-06,
      "loss": 0.1939,
      "step": 8137
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3022019456897431,
      "learning_rate": 3.921949120685178e-06,
      "loss": 0.1767,
      "step": 8138
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.171416154249515,
      "learning_rate": 3.920817991484627e-06,
      "loss": 0.513,
      "step": 8139
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.315203169807279,
      "learning_rate": 3.9196869202102775e-06,
      "loss": 0.1853,
      "step": 8140
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2123076747104098,
      "learning_rate": 3.918555906922845e-06,
      "loss": 0.1472,
      "step": 8141
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4227486995343832,
      "learning_rate": 3.917424951683032e-06,
      "loss": 0.1741,
      "step": 8142
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2909050526561214,
      "learning_rate": 3.916294054551548e-06,
      "loss": 0.1845,
      "step": 8143
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3135556641506279,
      "learning_rate": 3.915163215589094e-06,
      "loss": 0.1476,
      "step": 8144
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1899592895624993,
      "learning_rate": 3.91403243485637e-06,
      "loss": 0.1551,
      "step": 8145
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3856053710386913,
      "learning_rate": 3.912901712414068e-06,
      "loss": 0.1994,
      "step": 8146
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4223187317432422,
      "learning_rate": 3.911771048322885e-06,
      "loss": 0.2103,
      "step": 8147
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3563099429245546,
      "learning_rate": 3.910640442643508e-06,
      "loss": 0.1977,
      "step": 8148
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2706505057831987,
      "learning_rate": 3.9095098954366264e-06,
      "loss": 0.2196,
      "step": 8149
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3622788213792312,
      "learning_rate": 3.908379406762918e-06,
      "loss": 0.1818,
      "step": 8150
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3292331972810614,
      "learning_rate": 3.907248976683069e-06,
      "loss": 0.1989,
      "step": 8151
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2430086877384752,
      "learning_rate": 3.9061186052577515e-06,
      "loss": 0.2086,
      "step": 8152
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4004852508407137,
      "learning_rate": 3.904988292547643e-06,
      "loss": 0.2334,
      "step": 8153
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.183495840610173,
      "learning_rate": 3.903858038613412e-06,
      "loss": 0.2382,
      "step": 8154
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4325619158460001,
      "learning_rate": 3.902727843515727e-06,
      "loss": 0.241,
      "step": 8155
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3760137453176498,
      "learning_rate": 3.9015977073152516e-06,
      "loss": 0.2231,
      "step": 8156
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.341312526403342,
      "learning_rate": 3.90046763007265e-06,
      "loss": 0.1946,
      "step": 8157
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.407829326775763,
      "learning_rate": 3.899337611848575e-06,
      "loss": 0.1913,
      "step": 8158
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.326656399944762,
      "learning_rate": 3.898207652703685e-06,
      "loss": 0.2296,
      "step": 8159
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5429949108067782,
      "learning_rate": 3.89707775269863e-06,
      "loss": 0.2112,
      "step": 8160
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3854076500714492,
      "learning_rate": 3.89594791189406e-06,
      "loss": 0.1646,
      "step": 8161
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3322226829916661,
      "learning_rate": 3.894818130350622e-06,
      "loss": 0.1451,
      "step": 8162
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5979229346416453,
      "learning_rate": 3.893688408128955e-06,
      "loss": 0.2058,
      "step": 8163
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.435527348301019,
      "learning_rate": 3.892558745289698e-06,
      "loss": 0.1887,
      "step": 8164
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4488540473470253,
      "learning_rate": 3.891429141893489e-06,
      "loss": 0.2043,
      "step": 8165
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3039421615118558,
      "learning_rate": 3.890299598000961e-06,
      "loss": 0.1732,
      "step": 8166
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2043196690173117,
      "learning_rate": 3.889170113672739e-06,
      "loss": 0.1574,
      "step": 8167
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4651840568526107,
      "learning_rate": 3.888040688969454e-06,
      "loss": 0.1603,
      "step": 8168
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3565189895583047,
      "learning_rate": 3.886911323951725e-06,
      "loss": 0.1856,
      "step": 8169
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2905263478886697,
      "learning_rate": 3.8857820186801775e-06,
      "loss": 0.2003,
      "step": 8170
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2994892590047724,
      "learning_rate": 3.8846527732154224e-06,
      "loss": 0.1421,
      "step": 8171
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.244924802065644,
      "learning_rate": 3.883523587618077e-06,
      "loss": 0.1607,
      "step": 8172
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.669071457381142,
      "learning_rate": 3.882394461948748e-06,
      "loss": 0.439,
      "step": 8173
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3135493277674954,
      "learning_rate": 3.881265396268046e-06,
      "loss": 0.1908,
      "step": 8174
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2933195709072633,
      "learning_rate": 3.880136390636571e-06,
      "loss": 0.1873,
      "step": 8175
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6118211658697195,
      "learning_rate": 3.879007445114927e-06,
      "loss": 0.2261,
      "step": 8176
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3922521239012664,
      "learning_rate": 3.877878559763709e-06,
      "loss": 0.227,
      "step": 8177
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3678209399132972,
      "learning_rate": 3.8767497346435145e-06,
      "loss": 0.1847,
      "step": 8178
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3268655274217813,
      "learning_rate": 3.875620969814929e-06,
      "loss": 0.1658,
      "step": 8179
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3334428915797054,
      "learning_rate": 3.874492265338544e-06,
      "loss": 0.189,
      "step": 8180
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3608532815525396,
      "learning_rate": 3.873363621274942e-06,
      "loss": 0.1405,
      "step": 8181
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4123860902121381,
      "learning_rate": 3.872235037684707e-06,
      "loss": 0.1899,
      "step": 8182
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2087985097031808,
      "learning_rate": 3.871106514628412e-06,
      "loss": 0.1778,
      "step": 8183
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3389011035323852,
      "learning_rate": 3.869978052166637e-06,
      "loss": 0.1583,
      "step": 8184
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2086628760258693,
      "learning_rate": 3.868849650359948e-06,
      "loss": 0.1756,
      "step": 8185
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5768497302293405,
      "learning_rate": 3.867721309268919e-06,
      "loss": 0.2028,
      "step": 8186
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.648931620027212,
      "learning_rate": 3.86659302895411e-06,
      "loss": 0.6985,
      "step": 8187
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5278640646907544,
      "learning_rate": 3.865464809476085e-06,
      "loss": 0.1961,
      "step": 8188
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1234800719551359,
      "learning_rate": 3.8643366508954e-06,
      "loss": 0.1751,
      "step": 8189
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.335651297310427,
      "learning_rate": 3.8632085532726156e-06,
      "loss": 0.1703,
      "step": 8190
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.601863663293131,
      "learning_rate": 3.862080516668277e-06,
      "loss": 0.2074,
      "step": 8191
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1625932326851998,
      "learning_rate": 3.860952541142935e-06,
      "loss": 0.1591,
      "step": 8192
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2244072588594317,
      "learning_rate": 3.859824626757137e-06,
      "loss": 0.1819,
      "step": 8193
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4139756552210125,
      "learning_rate": 3.858696773571422e-06,
      "loss": 0.217,
      "step": 8194
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.378263873819433,
      "learning_rate": 3.857568981646332e-06,
      "loss": 0.1936,
      "step": 8195
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4455239076067516,
      "learning_rate": 3.856441251042399e-06,
      "loss": 0.2101,
      "step": 8196
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3058299826787094,
      "learning_rate": 3.8553135818201585e-06,
      "loss": 0.193,
      "step": 8197
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3418131299208313,
      "learning_rate": 3.854185974040133e-06,
      "loss": 0.1978,
      "step": 8198
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3306553579379938,
      "learning_rate": 3.853058427762857e-06,
      "loss": 0.1874,
      "step": 8199
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.468056985017301,
      "learning_rate": 3.851930943048845e-06,
      "loss": 0.2002,
      "step": 8200
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2723241790375805,
      "learning_rate": 3.850803519958621e-06,
      "loss": 0.1964,
      "step": 8201
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4376556542227754,
      "learning_rate": 3.849676158552697e-06,
      "loss": 0.2231,
      "step": 8202
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2356535534939614,
      "learning_rate": 3.848548858891588e-06,
      "loss": 0.1682,
      "step": 8203
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4308626690880544,
      "learning_rate": 3.847421621035801e-06,
      "loss": 0.1854,
      "step": 8204
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.384868146765296,
      "learning_rate": 3.846294445045843e-06,
      "loss": 0.233,
      "step": 8205
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5551709903763036,
      "learning_rate": 3.845167330982215e-06,
      "loss": 0.2407,
      "step": 8206
    },
    {
      "epoch": 0.59,
      "grad_norm": 14.730020667441869,
      "learning_rate": 3.844040278905418e-06,
      "loss": 0.7133,
      "step": 8207
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4934493691673087,
      "learning_rate": 3.842913288875944e-06,
      "loss": 0.2118,
      "step": 8208
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2922259669969103,
      "learning_rate": 3.841786360954291e-06,
      "loss": 0.1812,
      "step": 8209
    },
    {
      "epoch": 0.59,
      "grad_norm": 11.701881204586385,
      "learning_rate": 3.840659495200942e-06,
      "loss": 0.5567,
      "step": 8210
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4461140283267908,
      "learning_rate": 3.8395326916763875e-06,
      "loss": 0.1885,
      "step": 8211
    },
    {
      "epoch": 0.59,
      "grad_norm": 10.267828768474322,
      "learning_rate": 3.838405950441106e-06,
      "loss": 0.7477,
      "step": 8212
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.229406198385201,
      "learning_rate": 3.83727927155558e-06,
      "loss": 0.1641,
      "step": 8213
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4619759754412416,
      "learning_rate": 3.836152655080282e-06,
      "loss": 0.2127,
      "step": 8214
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.99694058346617,
      "learning_rate": 3.835026101075687e-06,
      "loss": 0.5915,
      "step": 8215
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2396661704370746,
      "learning_rate": 3.833899609602262e-06,
      "loss": 0.1874,
      "step": 8216
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.080170500603185,
      "learning_rate": 3.832773180720475e-06,
      "loss": 0.3854,
      "step": 8217
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3525940250399073,
      "learning_rate": 3.831646814490784e-06,
      "loss": 0.1734,
      "step": 8218
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2768447025790666,
      "learning_rate": 3.830520510973652e-06,
      "loss": 0.1735,
      "step": 8219
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2940805380524523,
      "learning_rate": 3.829394270229531e-06,
      "loss": 0.1896,
      "step": 8220
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.253648582268682,
      "learning_rate": 3.8282680923188775e-06,
      "loss": 0.1625,
      "step": 8221
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0743483650338528,
      "learning_rate": 3.827141977302135e-06,
      "loss": 0.148,
      "step": 8222
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3210407251528782,
      "learning_rate": 3.826015925239755e-06,
      "loss": 0.1851,
      "step": 8223
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.279716320114616,
      "learning_rate": 3.824889936192173e-06,
      "loss": 0.1926,
      "step": 8224
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3006319973903484,
      "learning_rate": 3.823764010219831e-06,
      "loss": 0.1757,
      "step": 8225
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.427344672363482,
      "learning_rate": 3.822638147383166e-06,
      "loss": 0.2296,
      "step": 8226
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.277319151260448,
      "learning_rate": 3.821512347742606e-06,
      "loss": 0.1778,
      "step": 8227
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.847510883770891,
      "learning_rate": 3.820386611358583e-06,
      "loss": 0.6468,
      "step": 8228
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.399461133991848,
      "learning_rate": 3.8192609382915195e-06,
      "loss": 0.2005,
      "step": 8229
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2967583961834626,
      "learning_rate": 3.818135328601838e-06,
      "loss": 0.1913,
      "step": 8230
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.413063792608432,
      "learning_rate": 3.817009782349957e-06,
      "loss": 0.1684,
      "step": 8231
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.482659523859138,
      "learning_rate": 3.815884299596291e-06,
      "loss": 0.23,
      "step": 8232
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4470971548160143,
      "learning_rate": 3.81475888040125e-06,
      "loss": 0.207,
      "step": 8233
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3435603059391588,
      "learning_rate": 3.813633524825247e-06,
      "loss": 0.1617,
      "step": 8234
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3268370068650563,
      "learning_rate": 3.8125082329286807e-06,
      "loss": 0.1749,
      "step": 8235
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3797434450555608,
      "learning_rate": 3.8113830047719556e-06,
      "loss": 0.1712,
      "step": 8236
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4655184210881325,
      "learning_rate": 3.8102578404154677e-06,
      "loss": 0.1528,
      "step": 8237
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2641358308136652,
      "learning_rate": 3.8091327399196143e-06,
      "loss": 0.1854,
      "step": 8238
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.399825429243039,
      "learning_rate": 3.808007703344782e-06,
      "loss": 0.1862,
      "step": 8239
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2113498066344248,
      "learning_rate": 3.8068827307513624e-06,
      "loss": 0.1424,
      "step": 8240
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2298689145993762,
      "learning_rate": 3.805757822199736e-06,
      "loss": 0.1749,
      "step": 8241
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3571604770707844,
      "learning_rate": 3.8046329777502883e-06,
      "loss": 0.2002,
      "step": 8242
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3819176978148626,
      "learning_rate": 3.80350819746339e-06,
      "loss": 0.2058,
      "step": 8243
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2866728311936917,
      "learning_rate": 3.802383481399421e-06,
      "loss": 0.1507,
      "step": 8244
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.386009820351826,
      "learning_rate": 3.8012588296187465e-06,
      "loss": 0.199,
      "step": 8245
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.227706633242937,
      "learning_rate": 3.800134242181738e-06,
      "loss": 0.2003,
      "step": 8246
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3693890357411833,
      "learning_rate": 3.7990097191487544e-06,
      "loss": 0.1971,
      "step": 8247
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3765619997111092,
      "learning_rate": 3.797885260580159e-06,
      "loss": 0.2002,
      "step": 8248
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3113474115542139,
      "learning_rate": 3.7967608665363065e-06,
      "loss": 0.1698,
      "step": 8249
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4765323183380141,
      "learning_rate": 3.7956365370775522e-06,
      "loss": 0.2076,
      "step": 8250
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2571400634408958,
      "learning_rate": 3.7945122722642416e-06,
      "loss": 0.1974,
      "step": 8251
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2273369888422752,
      "learning_rate": 3.793388072156724e-06,
      "loss": 0.1649,
      "step": 8252
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4006924757587365,
      "learning_rate": 3.79226393681534e-06,
      "loss": 0.1835,
      "step": 8253
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3007943997750702,
      "learning_rate": 3.791139866300433e-06,
      "loss": 0.1567,
      "step": 8254
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.344181333223399,
      "learning_rate": 3.790015860672332e-06,
      "loss": 0.3917,
      "step": 8255
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5594539304519044,
      "learning_rate": 3.788891919991373e-06,
      "loss": 0.2237,
      "step": 8256
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3925451739585688,
      "learning_rate": 3.7877680443178856e-06,
      "loss": 0.1659,
      "step": 8257
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2943944826471174,
      "learning_rate": 3.7866442337121922e-06,
      "loss": 0.2113,
      "step": 8258
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3931469883195335,
      "learning_rate": 3.7855204882346186e-06,
      "loss": 0.1529,
      "step": 8259
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4331546500990469,
      "learning_rate": 3.7843968079454773e-06,
      "loss": 0.2091,
      "step": 8260
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1256385703064187,
      "learning_rate": 3.783273192905088e-06,
      "loss": 0.1394,
      "step": 8261
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3498903273146463,
      "learning_rate": 3.782149643173758e-06,
      "loss": 0.1967,
      "step": 8262
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2764352541584139,
      "learning_rate": 3.7810261588118e-06,
      "loss": 0.1746,
      "step": 8263
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.350285025648923,
      "learning_rate": 3.7799027398795123e-06,
      "loss": 0.1859,
      "step": 8264
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3502985092897761,
      "learning_rate": 3.778779386437199e-06,
      "loss": 0.1892,
      "step": 8265
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.406906182444263,
      "learning_rate": 3.7776560985451564e-06,
      "loss": 0.1865,
      "step": 8266
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.130643488150585,
      "learning_rate": 3.77653287626368e-06,
      "loss": 0.1516,
      "step": 8267
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.363674506252413,
      "learning_rate": 3.7754097196530566e-06,
      "loss": 0.2292,
      "step": 8268
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.342491485070118,
      "learning_rate": 3.7742866287735747e-06,
      "loss": 0.2078,
      "step": 8269
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.343696181476595,
      "learning_rate": 3.7731636036855164e-06,
      "loss": 0.2149,
      "step": 8270
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3334790869628896,
      "learning_rate": 3.7720406444491643e-06,
      "loss": 0.1951,
      "step": 8271
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2842693707122037,
      "learning_rate": 3.770917751124789e-06,
      "loss": 0.1767,
      "step": 8272
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3212106337797853,
      "learning_rate": 3.769794923772668e-06,
      "loss": 0.1851,
      "step": 8273
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2031926842590743,
      "learning_rate": 3.7686721624530665e-06,
      "loss": 0.1156,
      "step": 8274
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4597403771905013,
      "learning_rate": 3.7675494672262534e-06,
      "loss": 0.1718,
      "step": 8275
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.51198559035681,
      "learning_rate": 3.766426838152487e-06,
      "loss": 0.2135,
      "step": 8276
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3487929063786075,
      "learning_rate": 3.765304275292028e-06,
      "loss": 0.1811,
      "step": 8277
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4632126581909304,
      "learning_rate": 3.764181778705129e-06,
      "loss": 0.1727,
      "step": 8278
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2671296547199031,
      "learning_rate": 3.763059348452045e-06,
      "loss": 0.1739,
      "step": 8279
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3622540923506359,
      "learning_rate": 3.7619369845930195e-06,
      "loss": 0.1927,
      "step": 8280
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3850389467070323,
      "learning_rate": 3.7608146871882985e-06,
      "loss": 0.2098,
      "step": 8281
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.419122698963883,
      "learning_rate": 3.759692456298121e-06,
      "loss": 0.1995,
      "step": 8282
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1817374957157123,
      "learning_rate": 3.758570291982727e-06,
      "loss": 0.183,
      "step": 8283
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.2418702738125456,
      "learning_rate": 3.7574481943023454e-06,
      "loss": 0.7239,
      "step": 8284
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3127347361271424,
      "learning_rate": 3.7563261633172105e-06,
      "loss": 0.2035,
      "step": 8285
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.23905080016415,
      "learning_rate": 3.7552041990875444e-06,
      "loss": 0.1781,
      "step": 8286
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2627608692663168,
      "learning_rate": 3.754082301673574e-06,
      "loss": 0.1993,
      "step": 8287
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.294428821791146,
      "learning_rate": 3.752960471135513e-06,
      "loss": 0.181,
      "step": 8288
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3239653083357747,
      "learning_rate": 3.7518387075335805e-06,
      "loss": 0.2374,
      "step": 8289
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2991920327128375,
      "learning_rate": 3.750717010927989e-06,
      "loss": 0.1855,
      "step": 8290
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1021347738004053,
      "learning_rate": 3.749595381378943e-06,
      "loss": 0.1408,
      "step": 8291
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.31935867882752,
      "learning_rate": 3.748473818946652e-06,
      "loss": 0.1656,
      "step": 8292
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0977527496575792,
      "learning_rate": 3.747352323691312e-06,
      "loss": 0.1511,
      "step": 8293
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4393089833922683,
      "learning_rate": 3.746230895673124e-06,
      "loss": 0.198,
      "step": 8294
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.7235182633734585,
      "learning_rate": 3.7451095349522798e-06,
      "loss": 0.5684,
      "step": 8295
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3175745563104093,
      "learning_rate": 3.743988241588972e-06,
      "loss": 0.183,
      "step": 8296
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.323830107527644,
      "learning_rate": 3.742867015643384e-06,
      "loss": 0.1791,
      "step": 8297
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3924303324238536,
      "learning_rate": 3.7417458571757016e-06,
      "loss": 0.1893,
      "step": 8298
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.70187732558381,
      "learning_rate": 3.7406247662461004e-06,
      "loss": 0.6204,
      "step": 8299
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3370869566911594,
      "learning_rate": 3.7395037429147615e-06,
      "loss": 0.1813,
      "step": 8300
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3304972038375855,
      "learning_rate": 3.738382787241851e-06,
      "loss": 0.1958,
      "step": 8301
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1666797289930977,
      "learning_rate": 3.7372618992875433e-06,
      "loss": 0.1303,
      "step": 8302
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5048103632722376,
      "learning_rate": 3.736141079111998e-06,
      "loss": 0.2026,
      "step": 8303
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.244507243592505,
      "learning_rate": 3.73502032677538e-06,
      "loss": 0.1374,
      "step": 8304
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.415973427973317,
      "learning_rate": 3.733899642337844e-06,
      "loss": 0.1637,
      "step": 8305
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.466779997603522,
      "learning_rate": 3.732779025859547e-06,
      "loss": 0.5694,
      "step": 8306
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4688324912054063,
      "learning_rate": 3.7316584774006354e-06,
      "loss": 0.2409,
      "step": 8307
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5237183546471513,
      "learning_rate": 3.7305379970212595e-06,
      "loss": 0.1996,
      "step": 8308
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.242250545782007,
      "learning_rate": 3.729417584781559e-06,
      "loss": 0.1798,
      "step": 8309
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.282090258760859,
      "learning_rate": 3.7282972407416774e-06,
      "loss": 0.187,
      "step": 8310
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3401399295235474,
      "learning_rate": 3.727176964961745e-06,
      "loss": 0.1722,
      "step": 8311
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1523268483665714,
      "learning_rate": 3.726056757501898e-06,
      "loss": 0.1352,
      "step": 8312
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3578040352652172,
      "learning_rate": 3.7249366184222613e-06,
      "loss": 0.1695,
      "step": 8313
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.123123848221532,
      "learning_rate": 3.7238165477829645e-06,
      "loss": 0.1876,
      "step": 8314
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.469065363856272,
      "learning_rate": 3.7226965456441233e-06,
      "loss": 0.2102,
      "step": 8315
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.322311888156187,
      "learning_rate": 3.7215766120658568e-06,
      "loss": 0.1589,
      "step": 8316
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.521963998706819,
      "learning_rate": 3.7204567471082793e-06,
      "loss": 0.2071,
      "step": 8317
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2645379441833726,
      "learning_rate": 3.7193369508315014e-06,
      "loss": 0.1754,
      "step": 8318
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.375301815340607,
      "learning_rate": 3.718217223295627e-06,
      "loss": 0.205,
      "step": 8319
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4260287784795684,
      "learning_rate": 3.7170975645607587e-06,
      "loss": 0.187,
      "step": 8320
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3546069926375115,
      "learning_rate": 3.7159779746869984e-06,
      "loss": 0.2156,
      "step": 8321
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3559037643080034,
      "learning_rate": 3.714858453734438e-06,
      "loss": 0.2266,
      "step": 8322
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3393984147838809,
      "learning_rate": 3.7137390017631725e-06,
      "loss": 0.2072,
      "step": 8323
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5607051803207834,
      "learning_rate": 3.712619618833284e-06,
      "loss": 0.2208,
      "step": 8324
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.182263001545025,
      "learning_rate": 3.7115003050048625e-06,
      "loss": 0.7286,
      "step": 8325
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3822935717263325,
      "learning_rate": 3.7103810603379843e-06,
      "loss": 0.1631,
      "step": 8326
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3903897878007845,
      "learning_rate": 3.7092618848927287e-06,
      "loss": 0.1907,
      "step": 8327
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4111078928387744,
      "learning_rate": 3.708142778729166e-06,
      "loss": 0.1957,
      "step": 8328
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.300768296528793,
      "learning_rate": 3.7070237419073674e-06,
      "loss": 0.1786,
      "step": 8329
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4548948432457116,
      "learning_rate": 3.705904774487396e-06,
      "loss": 0.2106,
      "step": 8330
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4597409202330278,
      "learning_rate": 3.7047858765293178e-06,
      "loss": 0.2023,
      "step": 8331
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2621307227237388,
      "learning_rate": 3.7036670480931856e-06,
      "loss": 0.1591,
      "step": 8332
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2036017489465787,
      "learning_rate": 3.7025482892390584e-06,
      "loss": 0.1593,
      "step": 8333
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3054145603719276,
      "learning_rate": 3.7014296000269823e-06,
      "loss": 0.1675,
      "step": 8334
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3319224901742184,
      "learning_rate": 3.700310980517009e-06,
      "loss": 0.2204,
      "step": 8335
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2939423546900832,
      "learning_rate": 3.6991924307691766e-06,
      "loss": 0.1688,
      "step": 8336
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3024415426511768,
      "learning_rate": 3.6980739508435283e-06,
      "loss": 0.1621,
      "step": 8337
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3918656068426778,
      "learning_rate": 3.6969555408000955e-06,
      "loss": 0.1564,
      "step": 8338
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3700234727487073,
      "learning_rate": 3.695837200698916e-06,
      "loss": 0.1896,
      "step": 8339
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.518091278615905,
      "learning_rate": 3.694718930600012e-06,
      "loss": 0.1954,
      "step": 8340
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2516303827194313,
      "learning_rate": 3.6936007305634116e-06,
      "loss": 0.1792,
      "step": 8341
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.342132593428376,
      "learning_rate": 3.692482600649132e-06,
      "loss": 0.168,
      "step": 8342
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4798619422975465,
      "learning_rate": 3.691364540917195e-06,
      "loss": 0.2215,
      "step": 8343
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3107462316622562,
      "learning_rate": 3.6902465514276075e-06,
      "loss": 0.1929,
      "step": 8344
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.736391622133606,
      "learning_rate": 3.689128632240383e-06,
      "loss": 0.5382,
      "step": 8345
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4084280297576446,
      "learning_rate": 3.688010783415525e-06,
      "loss": 0.19,
      "step": 8346
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4844754195241587,
      "learning_rate": 3.686893005013038e-06,
      "loss": 0.2245,
      "step": 8347
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2736474432936669,
      "learning_rate": 3.685775297092915e-06,
      "loss": 0.1631,
      "step": 8348
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.584847713461996,
      "learning_rate": 3.6846576597151545e-06,
      "loss": 0.6822,
      "step": 8349
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.283399436266776,
      "learning_rate": 3.6835400929397447e-06,
      "loss": 0.1773,
      "step": 8350
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.261610272042946,
      "learning_rate": 3.6824225968266736e-06,
      "loss": 0.1605,
      "step": 8351
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2391470324038933,
      "learning_rate": 3.6813051714359217e-06,
      "loss": 0.1729,
      "step": 8352
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5542715073205327,
      "learning_rate": 3.6801878168274697e-06,
      "loss": 0.2074,
      "step": 8353
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.183661611213176,
      "learning_rate": 3.6790705330612924e-06,
      "loss": 0.1635,
      "step": 8354
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.40952360089313,
      "learning_rate": 3.6779533201973606e-06,
      "loss": 0.1667,
      "step": 8355
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.843695765403123,
      "learning_rate": 3.676836178295645e-06,
      "loss": 0.4557,
      "step": 8356
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.6487129470323385,
      "learning_rate": 3.6757191074161034e-06,
      "loss": 0.6148,
      "step": 8357
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2997819865333031,
      "learning_rate": 3.674602107618701e-06,
      "loss": 0.1742,
      "step": 8358
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2680921027929253,
      "learning_rate": 3.6734851789633906e-06,
      "loss": 0.1452,
      "step": 8359
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4087280760901224,
      "learning_rate": 3.672368321510128e-06,
      "loss": 0.1761,
      "step": 8360
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.345190441285059,
      "learning_rate": 3.671251535318857e-06,
      "loss": 0.1971,
      "step": 8361
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.393749018793386,
      "learning_rate": 3.6701348204495262e-06,
      "loss": 0.1936,
      "step": 8362
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5118403122003532,
      "learning_rate": 3.669018176962074e-06,
      "loss": 0.1826,
      "step": 8363
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3187411805744231,
      "learning_rate": 3.66790160491644e-06,
      "loss": 0.1787,
      "step": 8364
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.8516701564271285,
      "learning_rate": 3.6667851043725543e-06,
      "loss": 0.5359,
      "step": 8365
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.233558914855478,
      "learning_rate": 3.6656686753903482e-06,
      "loss": 0.1353,
      "step": 8366
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3787787973060996,
      "learning_rate": 3.6645523180297453e-06,
      "loss": 0.2266,
      "step": 8367
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1780168016353185,
      "learning_rate": 3.6634360323506714e-06,
      "loss": 0.18,
      "step": 8368
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3633533956493566,
      "learning_rate": 3.662319818413038e-06,
      "loss": 0.2268,
      "step": 8369
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2731965549745274,
      "learning_rate": 3.6612036762767646e-06,
      "loss": 0.1537,
      "step": 8370
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3693526121301356,
      "learning_rate": 3.6600876060017576e-06,
      "loss": 0.1484,
      "step": 8371
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.235004671785468,
      "learning_rate": 3.658971607647927e-06,
      "loss": 0.4584,
      "step": 8372
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3027019794553836,
      "learning_rate": 3.6578556812751707e-06,
      "loss": 0.15,
      "step": 8373
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4334972852586056,
      "learning_rate": 3.656739826943391e-06,
      "loss": 0.1887,
      "step": 8374
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2783856344129527,
      "learning_rate": 3.655624044712479e-06,
      "loss": 0.1966,
      "step": 8375
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3145390929201037,
      "learning_rate": 3.6545083346423295e-06,
      "loss": 0.1891,
      "step": 8376
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.193447655064492,
      "learning_rate": 3.6533926967928256e-06,
      "loss": 0.1442,
      "step": 8377
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2793217052561896,
      "learning_rate": 3.6522771312238525e-06,
      "loss": 0.1954,
      "step": 8378
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4348332898916734,
      "learning_rate": 3.6511616379952886e-06,
      "loss": 0.1977,
      "step": 8379
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.050732560496598,
      "learning_rate": 3.6500462171670104e-06,
      "loss": 0.6795,
      "step": 8380
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4192554461856675,
      "learning_rate": 3.648930868798887e-06,
      "loss": 0.1922,
      "step": 8381
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2903795974408132,
      "learning_rate": 3.6478155929507876e-06,
      "loss": 0.1997,
      "step": 8382
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3919959264618358,
      "learning_rate": 3.6467003896825737e-06,
      "loss": 0.2132,
      "step": 8383
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2785309645998517,
      "learning_rate": 3.6455852590541075e-06,
      "loss": 0.159,
      "step": 8384
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.388438417768651,
      "learning_rate": 3.6444702011252455e-06,
      "loss": 0.2019,
      "step": 8385
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.614838264291656,
      "learning_rate": 3.643355215955835e-06,
      "loss": 0.2023,
      "step": 8386
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5325605824165516,
      "learning_rate": 3.6422403036057287e-06,
      "loss": 0.1623,
      "step": 8387
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.551157103412736,
      "learning_rate": 3.641125464134768e-06,
      "loss": 0.2302,
      "step": 8388
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2898447666256956,
      "learning_rate": 3.640010697602795e-06,
      "loss": 0.1643,
      "step": 8389
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.243238201094228,
      "learning_rate": 3.638896004069644e-06,
      "loss": 0.1546,
      "step": 8390
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3250652271781276,
      "learning_rate": 3.6377813835951483e-06,
      "loss": 0.2234,
      "step": 8391
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3113914745820412,
      "learning_rate": 3.636666836239135e-06,
      "loss": 0.1629,
      "step": 8392
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5349374102385616,
      "learning_rate": 3.635552362061432e-06,
      "loss": 0.1941,
      "step": 8393
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2642342546498229,
      "learning_rate": 3.6344379611218557e-06,
      "loss": 0.1877,
      "step": 8394
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3365056072929644,
      "learning_rate": 3.633323633480226e-06,
      "loss": 0.1673,
      "step": 8395
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3971651059673489,
      "learning_rate": 3.632209379196352e-06,
      "loss": 0.2178,
      "step": 8396
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4960848921120704,
      "learning_rate": 3.631095198330047e-06,
      "loss": 0.2026,
      "step": 8397
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.380448071203612,
      "learning_rate": 3.6299810909411114e-06,
      "loss": 0.1791,
      "step": 8398
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.785850953093162,
      "learning_rate": 3.6288670570893505e-06,
      "loss": 0.5208,
      "step": 8399
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2476096016344955,
      "learning_rate": 3.6277530968345552e-06,
      "loss": 0.1757,
      "step": 8400
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3908193244569391,
      "learning_rate": 3.6266392102365245e-06,
      "loss": 0.1934,
      "step": 8401
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.790317570983507,
      "learning_rate": 3.625525397355044e-06,
      "loss": 0.5287,
      "step": 8402
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.513729924869061,
      "learning_rate": 3.624411658249901e-06,
      "loss": 0.2312,
      "step": 8403
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.8320252004375126,
      "learning_rate": 3.6232979929808735e-06,
      "loss": 0.3684,
      "step": 8404
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2894707673000105,
      "learning_rate": 3.6221844016077414e-06,
      "loss": 0.1443,
      "step": 8405
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2214325665922086,
      "learning_rate": 3.621070884190276e-06,
      "loss": 0.1726,
      "step": 8406
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.31295387530887,
      "learning_rate": 3.619957440788249e-06,
      "loss": 0.1964,
      "step": 8407
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3191468058107259,
      "learning_rate": 3.6188440714614228e-06,
      "loss": 0.1734,
      "step": 8408
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.665724523521432,
      "learning_rate": 3.61773077626956e-06,
      "loss": 0.5279,
      "step": 8409
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3785646298616203,
      "learning_rate": 3.6166175552724165e-06,
      "loss": 0.188,
      "step": 8410
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.555284487043558,
      "learning_rate": 3.615504408529749e-06,
      "loss": 0.1838,
      "step": 8411
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2856872267074997,
      "learning_rate": 3.6143913361013026e-06,
      "loss": 0.2211,
      "step": 8412
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6786345925203463,
      "learning_rate": 3.6132783380468263e-06,
      "loss": 0.2247,
      "step": 8413
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5993502112216864,
      "learning_rate": 3.6121654144260576e-06,
      "loss": 0.208,
      "step": 8414
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.689764043236425,
      "learning_rate": 3.611052565298738e-06,
      "loss": 0.4547,
      "step": 8415
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3706206541146488,
      "learning_rate": 3.609939790724597e-06,
      "loss": 0.1606,
      "step": 8416
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4037807889252851,
      "learning_rate": 3.6088270907633654e-06,
      "loss": 0.1632,
      "step": 8417
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.369250976882622,
      "learning_rate": 3.6077144654747696e-06,
      "loss": 0.1856,
      "step": 8418
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3894419463153862,
      "learning_rate": 3.6066019149185296e-06,
      "loss": 0.1974,
      "step": 8419
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1476116302668742,
      "learning_rate": 3.605489439154365e-06,
      "loss": 0.1515,
      "step": 8420
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.353712757508189,
      "learning_rate": 3.604377038241985e-06,
      "loss": 0.2001,
      "step": 8421
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.549066734720979,
      "learning_rate": 3.6032647122411018e-06,
      "loss": 0.1914,
      "step": 8422
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5528898029659586,
      "learning_rate": 3.602152461211419e-06,
      "loss": 0.2029,
      "step": 8423
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3803578173852133,
      "learning_rate": 3.6010402852126415e-06,
      "loss": 0.1916,
      "step": 8424
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.575511380649091,
      "learning_rate": 3.5999281843044607e-06,
      "loss": 0.5809,
      "step": 8425
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2146737696212906,
      "learning_rate": 3.598816158546574e-06,
      "loss": 0.1653,
      "step": 8426
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.29875464830727,
      "learning_rate": 3.597704207998669e-06,
      "loss": 0.1751,
      "step": 8427
    },
    {
      "epoch": 0.6,
      "grad_norm": 3.9294187641129517,
      "learning_rate": 3.5965923327204326e-06,
      "loss": 0.6068,
      "step": 8428
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.547543362719488,
      "learning_rate": 3.5954805327715426e-06,
      "loss": 0.1821,
      "step": 8429
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2397043860833754,
      "learning_rate": 3.5943688082116784e-06,
      "loss": 0.1594,
      "step": 8430
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2944129480680122,
      "learning_rate": 3.5932571591005116e-06,
      "loss": 0.1551,
      "step": 8431
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.426555619890655,
      "learning_rate": 3.592145585497713e-06,
      "loss": 0.5961,
      "step": 8432
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6747670621203012,
      "learning_rate": 3.591034087462945e-06,
      "loss": 0.2073,
      "step": 8433
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4303967749279372,
      "learning_rate": 3.5899226650558704e-06,
      "loss": 0.1904,
      "step": 8434
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4788903749098592,
      "learning_rate": 3.588811318336145e-06,
      "loss": 0.204,
      "step": 8435
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3598727880587567,
      "learning_rate": 3.5877000473634227e-06,
      "loss": 0.2277,
      "step": 8436
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5147111709789234,
      "learning_rate": 3.5865888521973493e-06,
      "loss": 0.1692,
      "step": 8437
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1423775609675448,
      "learning_rate": 3.585477732897572e-06,
      "loss": 0.1564,
      "step": 8438
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.222204772320094,
      "learning_rate": 3.58436668952373e-06,
      "loss": 0.1738,
      "step": 8439
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.252503334126405,
      "learning_rate": 3.583255722135462e-06,
      "loss": 0.172,
      "step": 8440
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.9193106987262265,
      "learning_rate": 3.5821448307923957e-06,
      "loss": 0.6111,
      "step": 8441
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3137855379053673,
      "learning_rate": 3.5810340155541636e-06,
      "loss": 0.1738,
      "step": 8442
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5445580979958053,
      "learning_rate": 3.579923276480387e-06,
      "loss": 0.2087,
      "step": 8443
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.681052262333492,
      "learning_rate": 3.57881261363069e-06,
      "loss": 0.5736,
      "step": 8444
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3926166746148214,
      "learning_rate": 3.5777020270646827e-06,
      "loss": 0.1871,
      "step": 8445
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2756241716841266,
      "learning_rate": 3.576591516841982e-06,
      "loss": 0.1498,
      "step": 8446
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1738736732685149,
      "learning_rate": 3.575481083022192e-06,
      "loss": 0.1728,
      "step": 8447
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4167673744009057,
      "learning_rate": 3.5743707256649184e-06,
      "loss": 0.1805,
      "step": 8448
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3696861528187663,
      "learning_rate": 3.573260444829763e-06,
      "loss": 0.2219,
      "step": 8449
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.423873178920901,
      "learning_rate": 3.5721502405763163e-06,
      "loss": 0.2212,
      "step": 8450
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.2514934794189445,
      "learning_rate": 3.571040112964174e-06,
      "loss": 0.7691,
      "step": 8451
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5438970613979808,
      "learning_rate": 3.569930062052919e-06,
      "loss": 0.1896,
      "step": 8452
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2103593409757405,
      "learning_rate": 3.5688200879021402e-06,
      "loss": 0.161,
      "step": 8453
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2014437946414354,
      "learning_rate": 3.5677101905714106e-06,
      "loss": 0.1626,
      "step": 8454
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.164408108726043,
      "learning_rate": 3.566600370120309e-06,
      "loss": 0.3726,
      "step": 8455
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3764834229506135,
      "learning_rate": 3.565490626608404e-06,
      "loss": 0.1938,
      "step": 8456
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4079927529388696,
      "learning_rate": 3.564380960095265e-06,
      "loss": 0.2147,
      "step": 8457
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2339033085450002,
      "learning_rate": 3.563271370640451e-06,
      "loss": 0.1644,
      "step": 8458
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4137453845990362,
      "learning_rate": 3.5621618583035223e-06,
      "loss": 0.1815,
      "step": 8459
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3358613037355063,
      "learning_rate": 3.5610524231440324e-06,
      "loss": 0.1691,
      "step": 8460
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4217867775110316,
      "learning_rate": 3.5599430652215337e-06,
      "loss": 0.2288,
      "step": 8461
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2527569521361974,
      "learning_rate": 3.5588337845955678e-06,
      "loss": 0.1835,
      "step": 8462
    },
    {
      "epoch": 0.61,
      "grad_norm": 12.05132144997392,
      "learning_rate": 3.5577245813256807e-06,
      "loss": 0.7908,
      "step": 8463
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2856943258653033,
      "learning_rate": 3.5566154554714067e-06,
      "loss": 0.175,
      "step": 8464
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.545320405462178,
      "learning_rate": 3.5555064070922827e-06,
      "loss": 0.2211,
      "step": 8465
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3599588398325186,
      "learning_rate": 3.5543974362478345e-06,
      "loss": 0.163,
      "step": 8466
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3744884093304237,
      "learning_rate": 3.5532885429975893e-06,
      "loss": 0.2166,
      "step": 8467
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.31987026837196,
      "learning_rate": 3.552179727401067e-06,
      "loss": 0.2,
      "step": 8468
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1821937905733866,
      "learning_rate": 3.5510709895177873e-06,
      "loss": 0.127,
      "step": 8469
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.386302636031659,
      "learning_rate": 3.5499623294072585e-06,
      "loss": 0.176,
      "step": 8470
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.203115265671527,
      "learning_rate": 3.5488537471289917e-06,
      "loss": 0.1587,
      "step": 8471
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2823763253514968,
      "learning_rate": 3.5477452427424895e-06,
      "loss": 0.1925,
      "step": 8472
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2766545067833932,
      "learning_rate": 3.5466368163072563e-06,
      "loss": 0.1645,
      "step": 8473
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.359090712367458,
      "learning_rate": 3.5455284678827818e-06,
      "loss": 0.1818,
      "step": 8474
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3844395691989348,
      "learning_rate": 3.544420197528562e-06,
      "loss": 0.1664,
      "step": 8475
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3977142316994342,
      "learning_rate": 3.5433120053040826e-06,
      "loss": 0.224,
      "step": 8476
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3129763370128653,
      "learning_rate": 3.5422038912688296e-06,
      "loss": 0.1801,
      "step": 8477
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.097706968772192,
      "learning_rate": 3.5410958554822783e-06,
      "loss": 0.441,
      "step": 8478
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4618236398664612,
      "learning_rate": 3.539987898003906e-06,
      "loss": 0.1994,
      "step": 8479
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.243414385200963,
      "learning_rate": 3.5388800188931825e-06,
      "loss": 0.2155,
      "step": 8480
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1963692063450893,
      "learning_rate": 3.5377722182095746e-06,
      "loss": 0.1787,
      "step": 8481
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2201024520052748,
      "learning_rate": 3.5366644960125474e-06,
      "loss": 0.1603,
      "step": 8482
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2555192276633345,
      "learning_rate": 3.535556852361554e-06,
      "loss": 0.1947,
      "step": 8483
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3147341562824693,
      "learning_rate": 3.534449287316052e-06,
      "loss": 0.1803,
      "step": 8484
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3494254478569783,
      "learning_rate": 3.533341800935489e-06,
      "loss": 0.2184,
      "step": 8485
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3230171056413007,
      "learning_rate": 3.5322343932793133e-06,
      "loss": 0.1777,
      "step": 8486
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.271745427557579,
      "learning_rate": 3.5311270644069624e-06,
      "loss": 0.1615,
      "step": 8487
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1149804083740633,
      "learning_rate": 3.5300198143778764e-06,
      "loss": 0.1443,
      "step": 8488
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4583341359880058,
      "learning_rate": 3.5289126432514854e-06,
      "loss": 0.1959,
      "step": 8489
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3441474074986868,
      "learning_rate": 3.527805551087221e-06,
      "loss": 0.1883,
      "step": 8490
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.6319334664071325,
      "learning_rate": 3.526698537944505e-06,
      "loss": 0.5411,
      "step": 8491
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3184478594161924,
      "learning_rate": 3.5255916038827586e-06,
      "loss": 0.1672,
      "step": 8492
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3189433111837376,
      "learning_rate": 3.5244847489613964e-06,
      "loss": 0.1977,
      "step": 8493
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4611787463512182,
      "learning_rate": 3.523377973239833e-06,
      "loss": 0.2262,
      "step": 8494
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1760501416446874,
      "learning_rate": 3.5222712767774718e-06,
      "loss": 0.16,
      "step": 8495
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3337550589986982,
      "learning_rate": 3.52116465963372e-06,
      "loss": 0.1706,
      "step": 8496
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5720199669906814,
      "learning_rate": 3.5200581218679708e-06,
      "loss": 0.2146,
      "step": 8497
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4128661548772858,
      "learning_rate": 3.5189516635396254e-06,
      "loss": 0.1749,
      "step": 8498
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4092287292457755,
      "learning_rate": 3.517845284708069e-06,
      "loss": 0.1964,
      "step": 8499
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.310898891779096,
      "learning_rate": 3.5167389854326907e-06,
      "loss": 0.4652,
      "step": 8500
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3307790284839118,
      "learning_rate": 3.5156327657728693e-06,
      "loss": 0.1867,
      "step": 8501
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4215756111294309,
      "learning_rate": 3.514526625787985e-06,
      "loss": 0.1982,
      "step": 8502
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.420069258325338,
      "learning_rate": 3.5134205655374094e-06,
      "loss": 0.413,
      "step": 8503
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4413048153707648,
      "learning_rate": 3.5123145850805135e-06,
      "loss": 0.1947,
      "step": 8504
    },
    {
      "epoch": 0.61,
      "grad_norm": 16.33850354796122,
      "learning_rate": 3.5112086844766586e-06,
      "loss": 0.5862,
      "step": 8505
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.161032185173859,
      "learning_rate": 3.5101028637852076e-06,
      "loss": 0.4368,
      "step": 8506
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.239626846076969,
      "learning_rate": 3.508997123065515e-06,
      "loss": 0.1809,
      "step": 8507
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2122490677260336,
      "learning_rate": 3.5078914623769357e-06,
      "loss": 0.1885,
      "step": 8508
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3576104378297111,
      "learning_rate": 3.5067858817788124e-06,
      "loss": 0.2293,
      "step": 8509
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2182292740448817,
      "learning_rate": 3.505680381330492e-06,
      "loss": 0.1358,
      "step": 8510
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.64415605720798,
      "learning_rate": 3.5045749610913106e-06,
      "loss": 0.5023,
      "step": 8511
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.374680624330204,
      "learning_rate": 3.503469621120604e-06,
      "loss": 0.1818,
      "step": 8512
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4244841651917741,
      "learning_rate": 3.5023643614777057e-06,
      "loss": 0.1718,
      "step": 8513
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5219431564764838,
      "learning_rate": 3.501259182221937e-06,
      "loss": 0.2314,
      "step": 8514
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.38404273856023,
      "learning_rate": 3.5001540834126215e-06,
      "loss": 0.6031,
      "step": 8515
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3256221189615416,
      "learning_rate": 3.499049065109075e-06,
      "loss": 0.1961,
      "step": 8516
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.468821265520208,
      "learning_rate": 3.497944127370614e-06,
      "loss": 0.5287,
      "step": 8517
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.387071117558416,
      "learning_rate": 3.4968392702565435e-06,
      "loss": 0.1961,
      "step": 8518
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.455914613523176,
      "learning_rate": 3.4957344938261696e-06,
      "loss": 0.1697,
      "step": 8519
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2260581014359513,
      "learning_rate": 3.4946297981387913e-06,
      "loss": 0.1469,
      "step": 8520
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1562137496669744,
      "learning_rate": 3.4935251832537063e-06,
      "loss": 0.1586,
      "step": 8521
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1620544698785438,
      "learning_rate": 3.4924206492302024e-06,
      "loss": 0.1697,
      "step": 8522
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3978741885194694,
      "learning_rate": 3.4913161961275706e-06,
      "loss": 0.2127,
      "step": 8523
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.233596701668589,
      "learning_rate": 3.4902118240050904e-06,
      "loss": 0.1766,
      "step": 8524
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3796451018382179,
      "learning_rate": 3.4891075329220436e-06,
      "loss": 0.1693,
      "step": 8525
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2404593049090307,
      "learning_rate": 3.4880033229376997e-06,
      "loss": 0.1482,
      "step": 8526
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3813321550992355,
      "learning_rate": 3.4868991941113317e-06,
      "loss": 0.2027,
      "step": 8527
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.286488521359285,
      "learning_rate": 3.485795146502202e-06,
      "loss": 0.1755,
      "step": 8528
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.387550183080702,
      "learning_rate": 3.484691180169576e-06,
      "loss": 0.2206,
      "step": 8529
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2831500810976135,
      "learning_rate": 3.4835872951727055e-06,
      "loss": 0.216,
      "step": 8530
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0867328869360358,
      "learning_rate": 3.4824834915708456e-06,
      "loss": 0.1325,
      "step": 8531
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3583162473875354,
      "learning_rate": 3.481379769423242e-06,
      "loss": 0.1856,
      "step": 8532
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.478942780581958,
      "learning_rate": 3.4802761287891405e-06,
      "loss": 0.1933,
      "step": 8533
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2106054381316174,
      "learning_rate": 3.479172569727778e-06,
      "loss": 0.1619,
      "step": 8534
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2199101890595334,
      "learning_rate": 3.4780690922983905e-06,
      "loss": 0.2098,
      "step": 8535
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3208745282789323,
      "learning_rate": 3.4769656965602073e-06,
      "loss": 0.1857,
      "step": 8536
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2997287844962955,
      "learning_rate": 3.475862382572456e-06,
      "loss": 0.1539,
      "step": 8537
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3925049095390296,
      "learning_rate": 3.4747591503943557e-06,
      "loss": 0.1751,
      "step": 8538
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2870860018118642,
      "learning_rate": 3.473656000085125e-06,
      "loss": 0.1762,
      "step": 8539
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2750415302241722,
      "learning_rate": 3.472552931703975e-06,
      "loss": 0.1694,
      "step": 8540
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1774431810087114,
      "learning_rate": 3.4714499453101192e-06,
      "loss": 0.14,
      "step": 8541
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.302893428721198,
      "learning_rate": 3.470347040962754e-06,
      "loss": 0.1974,
      "step": 8542
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3447498927341335,
      "learning_rate": 3.4692442187210845e-06,
      "loss": 0.1932,
      "step": 8543
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3665477688404022,
      "learning_rate": 3.4681414786443026e-06,
      "loss": 0.2049,
      "step": 8544
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4784932319332114,
      "learning_rate": 3.467038820791601e-06,
      "loss": 0.1989,
      "step": 8545
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2612465475073615,
      "learning_rate": 3.465936245222167e-06,
      "loss": 0.1869,
      "step": 8546
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3226690536907344,
      "learning_rate": 3.464833751995178e-06,
      "loss": 0.176,
      "step": 8547
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.199645469378118,
      "learning_rate": 3.4637313411698166e-06,
      "loss": 0.1309,
      "step": 8548
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.283174495951623,
      "learning_rate": 3.462629012805252e-06,
      "loss": 0.1925,
      "step": 8549
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2491659858308428,
      "learning_rate": 3.461526766960656e-06,
      "loss": 0.1648,
      "step": 8550
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.368293977817649,
      "learning_rate": 3.4604246036951895e-06,
      "loss": 0.1881,
      "step": 8551
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2143614843020516,
      "learning_rate": 3.4593225230680148e-06,
      "loss": 0.1681,
      "step": 8552
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3904016015035026,
      "learning_rate": 3.4582205251382853e-06,
      "loss": 0.1871,
      "step": 8553
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3847496043895973,
      "learning_rate": 3.457118609965155e-06,
      "loss": 0.1719,
      "step": 8554
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.394336909992013,
      "learning_rate": 3.456016777607766e-06,
      "loss": 0.4762,
      "step": 8555
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1876925014720798,
      "learning_rate": 3.4549150281252635e-06,
      "loss": 0.1466,
      "step": 8556
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1284980289732123,
      "learning_rate": 3.4538133615767832e-06,
      "loss": 0.1487,
      "step": 8557
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3749375526821372,
      "learning_rate": 3.45271177802146e-06,
      "loss": 0.1898,
      "step": 8558
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.551599736672071,
      "learning_rate": 3.4516102775184206e-06,
      "loss": 0.1954,
      "step": 8559
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.314450120501177,
      "learning_rate": 3.4505088601267913e-06,
      "loss": 0.5061,
      "step": 8560
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2392908694576763,
      "learning_rate": 3.4494075259056888e-06,
      "loss": 0.1457,
      "step": 8561
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4333085023945549,
      "learning_rate": 3.448306274914232e-06,
      "loss": 0.1876,
      "step": 8562
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2106012820665915,
      "learning_rate": 3.4472051072115285e-06,
      "loss": 0.1581,
      "step": 8563
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.474993049547695,
      "learning_rate": 3.4461040228566865e-06,
      "loss": 0.4627,
      "step": 8564
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4390453027233296,
      "learning_rate": 3.445003021908806e-06,
      "loss": 0.1912,
      "step": 8565
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3546882474831952,
      "learning_rate": 3.4439021044269882e-06,
      "loss": 0.1652,
      "step": 8566
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.299947803263112,
      "learning_rate": 3.442801270470321e-06,
      "loss": 0.7075,
      "step": 8567
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4338853630270099,
      "learning_rate": 3.4417005200978965e-06,
      "loss": 0.2169,
      "step": 8568
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.374269909410642,
      "learning_rate": 3.4405998533687955e-06,
      "loss": 0.6094,
      "step": 8569
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.330783331304729,
      "learning_rate": 3.4394992703421026e-06,
      "loss": 0.2077,
      "step": 8570
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.362972339875966,
      "learning_rate": 3.4383987710768863e-06,
      "loss": 0.1908,
      "step": 8571
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2567323106023296,
      "learning_rate": 3.4372983556322213e-06,
      "loss": 0.1745,
      "step": 8572
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3441258617137768,
      "learning_rate": 3.4361980240671713e-06,
      "loss": 0.1963,
      "step": 8573
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3541682223384297,
      "learning_rate": 3.4350977764408005e-06,
      "loss": 0.1875,
      "step": 8574
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2827997444277321,
      "learning_rate": 3.433997612812162e-06,
      "loss": 0.2051,
      "step": 8575
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.485093874537023,
      "learning_rate": 3.4328975332403105e-06,
      "loss": 0.2006,
      "step": 8576
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2770275800610154,
      "learning_rate": 3.431797537784295e-06,
      "loss": 0.1588,
      "step": 8577
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1916715099261135,
      "learning_rate": 3.4306976265031555e-06,
      "loss": 0.1652,
      "step": 8578
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0942575710213116,
      "learning_rate": 3.429597799455936e-06,
      "loss": 0.1474,
      "step": 8579
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.170145344610598,
      "learning_rate": 3.428498056701665e-06,
      "loss": 0.173,
      "step": 8580
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1989052366242785,
      "learning_rate": 3.4273983982993765e-06,
      "loss": 0.1261,
      "step": 8581
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3334679788573782,
      "learning_rate": 3.426298824308093e-06,
      "loss": 0.158,
      "step": 8582
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3749311868200442,
      "learning_rate": 3.4251993347868396e-06,
      "loss": 0.1645,
      "step": 8583
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.270133050243536,
      "learning_rate": 3.4240999297946266e-06,
      "loss": 0.1919,
      "step": 8584
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.088293758333263,
      "learning_rate": 3.42300060939047e-06,
      "loss": 0.1448,
      "step": 8585
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3454301371616595,
      "learning_rate": 3.421901373633374e-06,
      "loss": 0.1789,
      "step": 8586
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3861888472450776,
      "learning_rate": 3.4208022225823455e-06,
      "loss": 0.1912,
      "step": 8587
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5241479967175833,
      "learning_rate": 3.4197031562963773e-06,
      "loss": 0.2261,
      "step": 8588
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.354415413622633,
      "learning_rate": 3.4186041748344667e-06,
      "loss": 0.2034,
      "step": 8589
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3415751734943744,
      "learning_rate": 3.4175052782556004e-06,
      "loss": 0.1938,
      "step": 8590
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1854241162055552,
      "learning_rate": 3.416406466618767e-06,
      "loss": 0.1482,
      "step": 8591
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4344052298001966,
      "learning_rate": 3.4153077399829393e-06,
      "loss": 0.1928,
      "step": 8592
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.446903210475065,
      "learning_rate": 3.4142090984070997e-06,
      "loss": 0.1816,
      "step": 8593
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2650657161103647,
      "learning_rate": 3.413110541950212e-06,
      "loss": 0.1924,
      "step": 8594
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.34275825572772,
      "learning_rate": 3.41201207067125e-06,
      "loss": 0.1774,
      "step": 8595
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1968725439007109,
      "learning_rate": 3.4109136846291684e-06,
      "loss": 0.1655,
      "step": 8596
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3342252411708815,
      "learning_rate": 3.4098153838829297e-06,
      "loss": 0.1922,
      "step": 8597
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2585929207151196,
      "learning_rate": 3.4087171684914812e-06,
      "loss": 0.1854,
      "step": 8598
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2843500409183648,
      "learning_rate": 3.407619038513774e-06,
      "loss": 0.2041,
      "step": 8599
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3102132790374756,
      "learning_rate": 3.4065209940087507e-06,
      "loss": 0.164,
      "step": 8600
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2022603341184783,
      "learning_rate": 3.405423035035351e-06,
      "loss": 0.1469,
      "step": 8601
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2165682177119461,
      "learning_rate": 3.404325161652506e-06,
      "loss": 0.1853,
      "step": 8602
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3976224667382484,
      "learning_rate": 3.4032273739191484e-06,
      "loss": 0.1832,
      "step": 8603
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2073596262191757,
      "learning_rate": 3.4021296718942006e-06,
      "loss": 0.1825,
      "step": 8604
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5514805707547807,
      "learning_rate": 3.401032055636586e-06,
      "loss": 0.2075,
      "step": 8605
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3860784010625111,
      "learning_rate": 3.3999345252052162e-06,
      "loss": 0.1832,
      "step": 8606
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.025647968863036,
      "learning_rate": 3.3988370806590054e-06,
      "loss": 0.6009,
      "step": 8607
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6252164696348075,
      "learning_rate": 3.3977397220568575e-06,
      "loss": 0.1905,
      "step": 8608
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2903631407541527,
      "learning_rate": 3.396642449457677e-06,
      "loss": 0.1854,
      "step": 8609
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2459143873617287,
      "learning_rate": 3.395545262920361e-06,
      "loss": 0.2184,
      "step": 8610
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.364433543838583,
      "learning_rate": 3.3944481625037996e-06,
      "loss": 0.2071,
      "step": 8611
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2409111497032799,
      "learning_rate": 3.3933511482668834e-06,
      "loss": 0.1307,
      "step": 8612
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.300492055125391,
      "learning_rate": 3.3922542202684934e-06,
      "loss": 0.1837,
      "step": 8613
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.937449676392841,
      "learning_rate": 3.3911573785675117e-06,
      "loss": 0.6464,
      "step": 8614
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3565271169299857,
      "learning_rate": 3.3900606232228086e-06,
      "loss": 0.1765,
      "step": 8615
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.276844932018481,
      "learning_rate": 3.3889639542932563e-06,
      "loss": 0.1798,
      "step": 8616
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3132890154316432,
      "learning_rate": 3.387867371837717e-06,
      "loss": 0.1746,
      "step": 8617
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.374320787321818,
      "learning_rate": 3.3867708759150554e-06,
      "loss": 0.1991,
      "step": 8618
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3112896799821265,
      "learning_rate": 3.385674466584121e-06,
      "loss": 0.1821,
      "step": 8619
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6715102859251778,
      "learning_rate": 3.3845781439037695e-06,
      "loss": 0.2184,
      "step": 8620
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.340800479331654,
      "learning_rate": 3.3834819079328446e-06,
      "loss": 0.1812,
      "step": 8621
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3325544478535725,
      "learning_rate": 3.3823857587301913e-06,
      "loss": 0.177,
      "step": 8622
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.339146441401021,
      "learning_rate": 3.381289696354641e-06,
      "loss": 0.213,
      "step": 8623
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.340317293636075,
      "learning_rate": 3.38019372086503e-06,
      "loss": 0.1515,
      "step": 8624
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4349589869108337,
      "learning_rate": 3.379097832320185e-06,
      "loss": 0.2018,
      "step": 8625
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4228090997176617,
      "learning_rate": 3.3780020307789303e-06,
      "loss": 0.1666,
      "step": 8626
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.164790870175347,
      "learning_rate": 3.3769063163000803e-06,
      "loss": 0.1584,
      "step": 8627
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1100530915661122,
      "learning_rate": 3.3758106889424526e-06,
      "loss": 0.1557,
      "step": 8628
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5657948720855377,
      "learning_rate": 3.3747151487648533e-06,
      "loss": 0.1978,
      "step": 8629
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.4045824313513,
      "learning_rate": 3.3736196958260902e-06,
      "loss": 0.6103,
      "step": 8630
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.383326993430047,
      "learning_rate": 3.372524330184958e-06,
      "loss": 0.1916,
      "step": 8631
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3498457385592173,
      "learning_rate": 3.371429051900256e-06,
      "loss": 0.1817,
      "step": 8632
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2339949154542118,
      "learning_rate": 3.3703338610307707e-06,
      "loss": 0.1685,
      "step": 8633
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.10639255004049,
      "learning_rate": 3.3692387576352914e-06,
      "loss": 0.1515,
      "step": 8634
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.398060173340639,
      "learning_rate": 3.3681437417725954e-06,
      "loss": 0.1881,
      "step": 8635
    },
    {
      "epoch": 0.62,
      "grad_norm": 17.782464745225116,
      "learning_rate": 3.3670488135014613e-06,
      "loss": 0.5967,
      "step": 8636
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.175138701246457,
      "learning_rate": 3.365953972880658e-06,
      "loss": 0.184,
      "step": 8637
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.494031360321678,
      "learning_rate": 3.3648592199689567e-06,
      "loss": 0.195,
      "step": 8638
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3148881394671816,
      "learning_rate": 3.3637645548251137e-06,
      "loss": 0.1788,
      "step": 8639
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3346393964753769,
      "learning_rate": 3.3626699775078884e-06,
      "loss": 0.1711,
      "step": 8640
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2853281487364556,
      "learning_rate": 3.3615754880760358e-06,
      "loss": 0.1833,
      "step": 8641
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3891862089959266,
      "learning_rate": 3.360481086588301e-06,
      "loss": 0.1794,
      "step": 8642
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3623478850126345,
      "learning_rate": 3.3593867731034297e-06,
      "loss": 0.1668,
      "step": 8643
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3867393329906643,
      "learning_rate": 3.3582925476801563e-06,
      "loss": 0.1928,
      "step": 8644
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3041854721246153,
      "learning_rate": 3.3571984103772183e-06,
      "loss": 0.1728,
      "step": 8645
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1347276111334628,
      "learning_rate": 3.3561043612533416e-06,
      "loss": 0.1496,
      "step": 8646
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.738822657597331,
      "learning_rate": 3.3550104003672545e-06,
      "loss": 0.6189,
      "step": 8647
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3465038860276453,
      "learning_rate": 3.353916527777672e-06,
      "loss": 0.1718,
      "step": 8648
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1751736808380955,
      "learning_rate": 3.3528227435433114e-06,
      "loss": 0.1571,
      "step": 8649
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2187797434548493,
      "learning_rate": 3.3517290477228804e-06,
      "loss": 0.1634,
      "step": 8650
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2430212514361907,
      "learning_rate": 3.3506354403750886e-06,
      "loss": 0.1888,
      "step": 8651
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4736642082777354,
      "learning_rate": 3.3495419215586324e-06,
      "loss": 0.1964,
      "step": 8652
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2958159435682504,
      "learning_rate": 3.348448491332209e-06,
      "loss": 0.1774,
      "step": 8653
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.40179239589177,
      "learning_rate": 3.3473551497545087e-06,
      "loss": 0.2551,
      "step": 8654
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1538580964288585,
      "learning_rate": 3.3462618968842197e-06,
      "loss": 0.1865,
      "step": 8655
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.364450028534952,
      "learning_rate": 3.345168732780021e-06,
      "loss": 0.18,
      "step": 8656
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3086104426993366,
      "learning_rate": 3.3440756575005905e-06,
      "loss": 0.1646,
      "step": 8657
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.407745359289409,
      "learning_rate": 3.3429826711046e-06,
      "loss": 0.1852,
      "step": 8658
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2063331389397043,
      "learning_rate": 3.3418897736507184e-06,
      "loss": 0.1844,
      "step": 8659
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3126321331279238,
      "learning_rate": 3.3407969651976045e-06,
      "loss": 0.1606,
      "step": 8660
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.789337471685698,
      "learning_rate": 3.3397042458039186e-06,
      "loss": 0.7628,
      "step": 8661
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3567526927523057,
      "learning_rate": 3.3386116155283122e-06,
      "loss": 0.1744,
      "step": 8662
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3635409368629408,
      "learning_rate": 3.3375190744294357e-06,
      "loss": 0.201,
      "step": 8663
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4102701693184723,
      "learning_rate": 3.3364266225659283e-06,
      "loss": 0.1877,
      "step": 8664
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3239105153571677,
      "learning_rate": 3.335334259996432e-06,
      "loss": 0.1661,
      "step": 8665
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.037669634476579,
      "learning_rate": 3.3342419867795785e-06,
      "loss": 0.6535,
      "step": 8666
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.258586904227841,
      "learning_rate": 3.333149802973999e-06,
      "loss": 0.1748,
      "step": 8667
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2303489922391702,
      "learning_rate": 3.3320577086383144e-06,
      "loss": 0.1665,
      "step": 8668
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.951440224966758,
      "learning_rate": 3.330965703831146e-06,
      "loss": 0.6723,
      "step": 8669
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3095504626776195,
      "learning_rate": 3.3298737886111075e-06,
      "loss": 0.1728,
      "step": 8670
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.412882263641033,
      "learning_rate": 3.3287819630368113e-06,
      "loss": 0.1597,
      "step": 8671
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3659364983316864,
      "learning_rate": 3.327690227166857e-06,
      "loss": 0.1822,
      "step": 8672
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3249393799676707,
      "learning_rate": 3.326598581059848e-06,
      "loss": 0.1927,
      "step": 8673
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3610729051344845,
      "learning_rate": 3.3255070247743802e-06,
      "loss": 0.1794,
      "step": 8674
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4451311227039474,
      "learning_rate": 3.324415558369042e-06,
      "loss": 0.1971,
      "step": 8675
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4213584204350374,
      "learning_rate": 3.323324181902422e-06,
      "loss": 0.1619,
      "step": 8676
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3907900192148737,
      "learning_rate": 3.322232895433096e-06,
      "loss": 0.1667,
      "step": 8677
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4097187806244211,
      "learning_rate": 3.3211416990196444e-06,
      "loss": 0.171,
      "step": 8678
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5090879791713012,
      "learning_rate": 3.3200505927206363e-06,
      "loss": 0.2063,
      "step": 8679
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.336233840270013,
      "learning_rate": 3.3189595765946394e-06,
      "loss": 0.187,
      "step": 8680
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3438193215760068,
      "learning_rate": 3.3178686507002117e-06,
      "loss": 0.1925,
      "step": 8681
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5187105238626142,
      "learning_rate": 3.316777815095914e-06,
      "loss": 0.2203,
      "step": 8682
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1577282388810495,
      "learning_rate": 3.3156870698402942e-06,
      "loss": 0.1566,
      "step": 8683
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2500257738216274,
      "learning_rate": 3.3145964149919034e-06,
      "loss": 0.1431,
      "step": 8684
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3162291838810491,
      "learning_rate": 3.31350585060928e-06,
      "loss": 0.1863,
      "step": 8685
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2031501120221884,
      "learning_rate": 3.3124153767509626e-06,
      "loss": 0.1749,
      "step": 8686
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.352768427778713,
      "learning_rate": 3.3113249934754823e-06,
      "loss": 0.5964,
      "step": 8687
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.001087248451887,
      "learning_rate": 3.3102347008413703e-06,
      "loss": 0.6842,
      "step": 8688
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.597846118714508,
      "learning_rate": 3.309144498907144e-06,
      "loss": 0.5868,
      "step": 8689
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.566956700977014,
      "learning_rate": 3.3080543877313244e-06,
      "loss": 0.4193,
      "step": 8690
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.827586582477459,
      "learning_rate": 3.306964367372423e-06,
      "loss": 0.6278,
      "step": 8691
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2530552735413754,
      "learning_rate": 3.30587443788895e-06,
      "loss": 0.1519,
      "step": 8692
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.175532372256806,
      "learning_rate": 3.3047845993394047e-06,
      "loss": 0.1467,
      "step": 8693
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3687162231977779,
      "learning_rate": 3.30369485178229e-06,
      "loss": 0.1955,
      "step": 8694
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3478984479182248,
      "learning_rate": 3.3026051952760928e-06,
      "loss": 0.1635,
      "step": 8695
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3179963864441426,
      "learning_rate": 3.301515629879309e-06,
      "loss": 0.1966,
      "step": 8696
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.293834634343991,
      "learning_rate": 3.3004261556504157e-06,
      "loss": 0.1319,
      "step": 8697
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.408175906736706,
      "learning_rate": 3.299336772647897e-06,
      "loss": 0.5648,
      "step": 8698
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4764772893137355,
      "learning_rate": 3.2982474809302216e-06,
      "loss": 0.1832,
      "step": 8699
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3388818864796406,
      "learning_rate": 3.2971582805558622e-06,
      "loss": 0.1996,
      "step": 8700
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3668078187000845,
      "learning_rate": 3.2960691715832793e-06,
      "loss": 0.1624,
      "step": 8701
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3104181352541406,
      "learning_rate": 3.294980154070937e-06,
      "loss": 0.1753,
      "step": 8702
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4032214548601905,
      "learning_rate": 3.293891228077284e-06,
      "loss": 0.16,
      "step": 8703
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.334905072313115,
      "learning_rate": 3.292802393660771e-06,
      "loss": 0.5789,
      "step": 8704
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2919248886937025,
      "learning_rate": 3.2917136508798452e-06,
      "loss": 0.1479,
      "step": 8705
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6307168402080758,
      "learning_rate": 3.2906249997929427e-06,
      "loss": 0.2102,
      "step": 8706
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.338258011451746,
      "learning_rate": 3.289536440458501e-06,
      "loss": 0.1802,
      "step": 8707
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.529269840966855,
      "learning_rate": 3.288447972934946e-06,
      "loss": 0.1868,
      "step": 8708
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4843884856604834,
      "learning_rate": 3.2873595972807054e-06,
      "loss": 0.1805,
      "step": 8709
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3975063464460078,
      "learning_rate": 3.286271313554196e-06,
      "loss": 0.2047,
      "step": 8710
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3884893827525335,
      "learning_rate": 3.2851831218138373e-06,
      "loss": 0.2142,
      "step": 8711
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3780908452096792,
      "learning_rate": 3.2840950221180336e-06,
      "loss": 0.1927,
      "step": 8712
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1841892506778477,
      "learning_rate": 3.2830070145251926e-06,
      "loss": 0.1427,
      "step": 8713
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2846183383816219,
      "learning_rate": 3.2819190990937134e-06,
      "loss": 0.1898,
      "step": 8714
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.448947175387091,
      "learning_rate": 3.2808312758819926e-06,
      "loss": 0.5468,
      "step": 8715
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2113134088730473,
      "learning_rate": 3.2797435449484182e-06,
      "loss": 0.1835,
      "step": 8716
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2880639486876093,
      "learning_rate": 3.2786559063513763e-06,
      "loss": 0.1825,
      "step": 8717
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.950657622118116,
      "learning_rate": 3.2775683601492457e-06,
      "loss": 0.4313,
      "step": 8718
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.431133487427325,
      "learning_rate": 3.2764809064004055e-06,
      "loss": 0.2174,
      "step": 8719
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3023604295307964,
      "learning_rate": 3.27539354516322e-06,
      "loss": 0.1718,
      "step": 8720
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3631230861078312,
      "learning_rate": 3.2743062764960594e-06,
      "loss": 0.1796,
      "step": 8721
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4696938709677243,
      "learning_rate": 3.2732191004572806e-06,
      "loss": 0.2179,
      "step": 8722
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2570828363237014,
      "learning_rate": 3.272132017105242e-06,
      "loss": 0.1763,
      "step": 8723
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5099290606694302,
      "learning_rate": 3.2710450264982906e-06,
      "loss": 0.1718,
      "step": 8724
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.276085479829146,
      "learning_rate": 3.2699581286947747e-06,
      "loss": 0.1577,
      "step": 8725
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.253008245438098,
      "learning_rate": 3.268871323753032e-06,
      "loss": 0.1708,
      "step": 8726
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.260099351378492,
      "learning_rate": 3.2677846117314016e-06,
      "loss": 0.179,
      "step": 8727
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.287806581986106,
      "learning_rate": 3.266697992688209e-06,
      "loss": 0.1715,
      "step": 8728
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4057828993213628,
      "learning_rate": 3.265611466681784e-06,
      "loss": 0.2027,
      "step": 8729
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5198456269055516,
      "learning_rate": 3.2645250337704437e-06,
      "loss": 0.2214,
      "step": 8730
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.544071585132708,
      "learning_rate": 3.2634386940125064e-06,
      "loss": 0.7532,
      "step": 8731
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3841289846779117,
      "learning_rate": 3.2623524474662792e-06,
      "loss": 0.1755,
      "step": 8732
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3763059404463562,
      "learning_rate": 3.2612662941900707e-06,
      "loss": 0.1913,
      "step": 8733
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2745271305190085,
      "learning_rate": 3.2601802342421784e-06,
      "loss": 0.1881,
      "step": 8734
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4221743138015968,
      "learning_rate": 3.2590942676809017e-06,
      "loss": 0.1853,
      "step": 8735
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1745237432487898,
      "learning_rate": 3.2580083945645264e-06,
      "loss": 0.1426,
      "step": 8736
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3106647306835637,
      "learning_rate": 3.2569226149513393e-06,
      "loss": 0.1523,
      "step": 8737
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2036955580314888,
      "learning_rate": 3.2558369288996226e-06,
      "loss": 0.1834,
      "step": 8738
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4457982528688325,
      "learning_rate": 3.2547513364676497e-06,
      "loss": 0.2283,
      "step": 8739
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1927022607588957,
      "learning_rate": 3.253665837713694e-06,
      "loss": 0.1373,
      "step": 8740
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4644863457255979,
      "learning_rate": 3.252580432696015e-06,
      "loss": 0.1791,
      "step": 8741
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2411593019173135,
      "learning_rate": 3.2514951214728787e-06,
      "loss": 0.1775,
      "step": 8742
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6312403767158474,
      "learning_rate": 3.250409904102536e-06,
      "loss": 0.2337,
      "step": 8743
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.397134312988982,
      "learning_rate": 3.2493247806432403e-06,
      "loss": 0.7022,
      "step": 8744
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2072114269845415,
      "learning_rate": 3.248239751153235e-06,
      "loss": 0.1655,
      "step": 8745
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2428445910544852,
      "learning_rate": 3.24715481569076e-06,
      "loss": 0.1563,
      "step": 8746
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.229613614904168,
      "learning_rate": 3.24606997431405e-06,
      "loss": 0.5724,
      "step": 8747
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4130183422064464,
      "learning_rate": 3.2449852270813386e-06,
      "loss": 0.158,
      "step": 8748
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2522743976032682,
      "learning_rate": 3.243900574050845e-06,
      "loss": 0.1919,
      "step": 8749
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3657312217598976,
      "learning_rate": 3.2428160152807934e-06,
      "loss": 0.183,
      "step": 8750
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2258537032764811,
      "learning_rate": 3.241731550829396e-06,
      "loss": 0.1916,
      "step": 8751
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.284224399031125,
      "learning_rate": 3.240647180754866e-06,
      "loss": 0.1295,
      "step": 8752
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.304655458964003,
      "learning_rate": 3.239562905115404e-06,
      "loss": 0.1539,
      "step": 8753
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0721800562513246,
      "learning_rate": 3.2384787239692115e-06,
      "loss": 0.1605,
      "step": 8754
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.225840395998788,
      "learning_rate": 3.2373946373744826e-06,
      "loss": 0.1968,
      "step": 8755
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4181937716000355,
      "learning_rate": 3.236310645389409e-06,
      "loss": 0.1724,
      "step": 8756
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.446415193838621,
      "learning_rate": 3.235226748072171e-06,
      "loss": 0.183,
      "step": 8757
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3321038927362396,
      "learning_rate": 3.2341429454809516e-06,
      "loss": 0.1614,
      "step": 8758
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2657181253538574,
      "learning_rate": 3.233059237673922e-06,
      "loss": 0.1661,
      "step": 8759
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3646892380370144,
      "learning_rate": 3.2319756247092552e-06,
      "loss": 0.1895,
      "step": 8760
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.361351781035548,
      "learning_rate": 3.230892106645111e-06,
      "loss": 0.1674,
      "step": 8761
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.199405732815531,
      "learning_rate": 3.229808683539651e-06,
      "loss": 0.1511,
      "step": 8762
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2769623701382478,
      "learning_rate": 3.2287253554510267e-06,
      "loss": 0.1514,
      "step": 8763
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5259361861573943,
      "learning_rate": 3.2276421224373912e-06,
      "loss": 0.1808,
      "step": 8764
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.387472104348445,
      "learning_rate": 3.2265589845568835e-06,
      "loss": 0.1902,
      "step": 8765
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6739096738981114,
      "learning_rate": 3.225475941867644e-06,
      "loss": 0.1907,
      "step": 8766
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4426136569362213,
      "learning_rate": 3.2243929944278065e-06,
      "loss": 0.1627,
      "step": 8767
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.365149131194602,
      "learning_rate": 3.2233101422954975e-06,
      "loss": 0.177,
      "step": 8768
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4075727220349132,
      "learning_rate": 3.2222273855288446e-06,
      "loss": 0.2027,
      "step": 8769
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4576376140699303,
      "learning_rate": 3.2211447241859606e-06,
      "loss": 0.2046,
      "step": 8770
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.24032341729528,
      "learning_rate": 3.2200621583249613e-06,
      "loss": 0.1636,
      "step": 8771
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.306733781425664,
      "learning_rate": 3.2189796880039535e-06,
      "loss": 0.1922,
      "step": 8772
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.28233687522276,
      "learning_rate": 3.217897313281043e-06,
      "loss": 0.1702,
      "step": 8773
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3472196666251426,
      "learning_rate": 3.216815034214322e-06,
      "loss": 0.1856,
      "step": 8774
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.563654025155306,
      "learning_rate": 3.2157328508618873e-06,
      "loss": 0.2125,
      "step": 8775
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3637711791766904,
      "learning_rate": 3.2146507632818237e-06,
      "loss": 0.1681,
      "step": 8776
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5331361845218694,
      "learning_rate": 3.213568771532216e-06,
      "loss": 0.2084,
      "step": 8777
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3370895010188608,
      "learning_rate": 3.212486875671137e-06,
      "loss": 0.1673,
      "step": 8778
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4181346824719638,
      "learning_rate": 3.2114050757566627e-06,
      "loss": 0.2232,
      "step": 8779
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.160070264111628,
      "learning_rate": 3.2103233718468574e-06,
      "loss": 0.1576,
      "step": 8780
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2071853737726663,
      "learning_rate": 3.2092417639997853e-06,
      "loss": 0.1783,
      "step": 8781
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3269724623284693,
      "learning_rate": 3.2081602522734987e-06,
      "loss": 0.2258,
      "step": 8782
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5556833384652586,
      "learning_rate": 3.207078836726053e-06,
      "loss": 0.2007,
      "step": 8783
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.779378576324171,
      "learning_rate": 3.20599751741549e-06,
      "loss": 0.5426,
      "step": 8784
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2270274943483026,
      "learning_rate": 3.2049162943998563e-06,
      "loss": 0.1742,
      "step": 8785
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.729921937651621,
      "learning_rate": 3.203835167737182e-06,
      "loss": 0.5412,
      "step": 8786
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.253131238926825,
      "learning_rate": 3.202754137485501e-06,
      "loss": 0.1575,
      "step": 8787
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2806861899280495,
      "learning_rate": 3.2016732037028375e-06,
      "loss": 0.1808,
      "step": 8788
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3960081756921914,
      "learning_rate": 3.2005923664472134e-06,
      "loss": 0.2011,
      "step": 8789
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.287404793138487,
      "learning_rate": 3.19951162577664e-06,
      "loss": 0.5753,
      "step": 8790
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.420283866307087,
      "learning_rate": 3.1984309817491327e-06,
      "loss": 0.191,
      "step": 8791
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2611759945066752,
      "learning_rate": 3.197350434422689e-06,
      "loss": 0.1842,
      "step": 8792
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5605813478108956,
      "learning_rate": 3.1962699838553164e-06,
      "loss": 0.2045,
      "step": 8793
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3841780143079696,
      "learning_rate": 3.1951896301050022e-06,
      "loss": 0.2224,
      "step": 8794
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3826302719958257,
      "learning_rate": 3.1941093732297412e-06,
      "loss": 0.2181,
      "step": 8795
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.0323033810152245,
      "learning_rate": 3.193029213287513e-06,
      "loss": 0.5028,
      "step": 8796
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4614444225235361,
      "learning_rate": 3.191949150336298e-06,
      "loss": 0.1926,
      "step": 8797
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2397674469683848,
      "learning_rate": 3.190869184434069e-06,
      "loss": 0.1651,
      "step": 8798
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2515749926056632,
      "learning_rate": 3.189789315638797e-06,
      "loss": 0.1915,
      "step": 8799
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2018222616851735,
      "learning_rate": 3.1887095440084402e-06,
      "loss": 0.1902,
      "step": 8800
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2495030326546441,
      "learning_rate": 3.187629869600959e-06,
      "loss": 0.2035,
      "step": 8801
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.485802813696995,
      "learning_rate": 3.18655029247431e-06,
      "loss": 0.1884,
      "step": 8802
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2097727020580316,
      "learning_rate": 3.185470812686433e-06,
      "loss": 0.1844,
      "step": 8803
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.424383783767484,
      "learning_rate": 3.1843914302952774e-06,
      "loss": 0.7474,
      "step": 8804
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2873612395061638,
      "learning_rate": 3.183312145358774e-06,
      "loss": 0.1994,
      "step": 8805
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6001685767150289,
      "learning_rate": 3.1822329579348586e-06,
      "loss": 0.2102,
      "step": 8806
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.371527594517568,
      "learning_rate": 3.181153868081456e-06,
      "loss": 0.1673,
      "step": 8807
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1593608057837652,
      "learning_rate": 3.180074875856489e-06,
      "loss": 0.1435,
      "step": 8808
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.182849996743878,
      "learning_rate": 3.1789959813178704e-06,
      "loss": 0.6055,
      "step": 8809
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2528606345333098,
      "learning_rate": 3.1779171845235144e-06,
      "loss": 0.1618,
      "step": 8810
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3444459482260436,
      "learning_rate": 3.176838485531325e-06,
      "loss": 0.1766,
      "step": 8811
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4907913264859274,
      "learning_rate": 3.175759884399203e-06,
      "loss": 0.1859,
      "step": 8812
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.210942553801653,
      "learning_rate": 3.1746813811850424e-06,
      "loss": 0.1732,
      "step": 8813
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4356574807055744,
      "learning_rate": 3.1736029759467345e-06,
      "loss": 0.2085,
      "step": 8814
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.274379200816446,
      "learning_rate": 3.1725246687421616e-06,
      "loss": 0.1634,
      "step": 8815
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.398396826892128,
      "learning_rate": 3.1714464596292065e-06,
      "loss": 0.1837,
      "step": 8816
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2668053272079471,
      "learning_rate": 3.170368348665739e-06,
      "loss": 0.1708,
      "step": 8817
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.430912822381253,
      "learning_rate": 3.1692903359096305e-06,
      "loss": 0.1973,
      "step": 8818
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2718128796216763,
      "learning_rate": 3.1682124214187427e-06,
      "loss": 0.1467,
      "step": 8819
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2482868574170807,
      "learning_rate": 3.167134605250938e-06,
      "loss": 0.1706,
      "step": 8820
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.452240409976205,
      "learning_rate": 3.1660568874640627e-06,
      "loss": 0.4341,
      "step": 8821
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.8710785984880625,
      "learning_rate": 3.164979268115969e-06,
      "loss": 0.6081,
      "step": 8822
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4718850313100704,
      "learning_rate": 3.1639017472644975e-06,
      "loss": 0.1868,
      "step": 8823
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.460272002991077,
      "learning_rate": 3.162824324967487e-06,
      "loss": 0.2029,
      "step": 8824
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3050281730843296,
      "learning_rate": 3.1617470012827674e-06,
      "loss": 0.2015,
      "step": 8825
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3304566485991751,
      "learning_rate": 3.160669776268166e-06,
      "loss": 0.1735,
      "step": 8826
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2428681803698354,
      "learning_rate": 3.159592649981502e-06,
      "loss": 0.2113,
      "step": 8827
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2991718587101573,
      "learning_rate": 3.1585156224805964e-06,
      "loss": 0.1869,
      "step": 8828
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1840332850568884,
      "learning_rate": 3.157438693823254e-06,
      "loss": 0.1346,
      "step": 8829
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4375418467726195,
      "learning_rate": 3.156361864067283e-06,
      "loss": 0.2102,
      "step": 8830
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2793589348710266,
      "learning_rate": 3.1552851332704816e-06,
      "loss": 0.1572,
      "step": 8831
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2140866881572394,
      "learning_rate": 3.1542085014906456e-06,
      "loss": 0.1692,
      "step": 8832
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.3033030527223115,
      "learning_rate": 3.1531319687855666e-06,
      "loss": 0.5462,
      "step": 8833
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2811081884191098,
      "learning_rate": 3.1520555352130246e-06,
      "loss": 0.1791,
      "step": 8834
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2849381569596956,
      "learning_rate": 3.1509792008308003e-06,
      "loss": 0.1473,
      "step": 8835
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1820783798089056,
      "learning_rate": 3.149902965696666e-06,
      "loss": 0.1445,
      "step": 8836
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.495352450514506,
      "learning_rate": 3.148826829868392e-06,
      "loss": 0.1934,
      "step": 8837
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4403379489320527,
      "learning_rate": 3.147750793403739e-06,
      "loss": 0.2087,
      "step": 8838
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5206568211527367,
      "learning_rate": 3.1466748563604643e-06,
      "loss": 0.1702,
      "step": 8839
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.301550706919972,
      "learning_rate": 3.14559901879632e-06,
      "loss": 0.1461,
      "step": 8840
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.372285156932272,
      "learning_rate": 3.144523280769055e-06,
      "loss": 0.1758,
      "step": 8841
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2554276442076213,
      "learning_rate": 3.143447642336407e-06,
      "loss": 0.1796,
      "step": 8842
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4685929166706162,
      "learning_rate": 3.142372103556115e-06,
      "loss": 0.177,
      "step": 8843
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4564830910823923,
      "learning_rate": 3.1412966644859073e-06,
      "loss": 0.2193,
      "step": 8844
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3098117723453786,
      "learning_rate": 3.1402213251835124e-06,
      "loss": 0.1788,
      "step": 8845
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.422002744513389,
      "learning_rate": 3.1391460857066465e-06,
      "loss": 0.1999,
      "step": 8846
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3161119472032723,
      "learning_rate": 3.1380709461130267e-06,
      "loss": 0.1653,
      "step": 8847
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1830774487804931,
      "learning_rate": 3.1369959064603607e-06,
      "loss": 0.1839,
      "step": 8848
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3196350012228997,
      "learning_rate": 3.135920966806355e-06,
      "loss": 0.2055,
      "step": 8849
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.260750789462727,
      "learning_rate": 3.134846127208704e-06,
      "loss": 0.151,
      "step": 8850
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1997800540004422,
      "learning_rate": 3.1337713877251043e-06,
      "loss": 0.1551,
      "step": 8851
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.3036874147466,
      "learning_rate": 3.132696748413241e-06,
      "loss": 0.5738,
      "step": 8852
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.155824888330129,
      "learning_rate": 3.1316222093308e-06,
      "loss": 0.17,
      "step": 8853
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3491935745093266,
      "learning_rate": 3.1305477705354546e-06,
      "loss": 0.1626,
      "step": 8854
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.725674005346951,
      "learning_rate": 3.1294734320848787e-06,
      "loss": 0.4747,
      "step": 8855
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3706054944490205,
      "learning_rate": 3.1283991940367365e-06,
      "loss": 0.1614,
      "step": 8856
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3957512667871042,
      "learning_rate": 3.1273250564486924e-06,
      "loss": 0.2073,
      "step": 8857
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1989228552822873,
      "learning_rate": 3.1262510193783966e-06,
      "loss": 0.154,
      "step": 8858
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.387621023597622,
      "learning_rate": 3.1251770828835043e-06,
      "loss": 0.1983,
      "step": 8859
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2702175449365516,
      "learning_rate": 3.1241032470216564e-06,
      "loss": 0.2008,
      "step": 8860
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3295065603780214,
      "learning_rate": 3.123029511850496e-06,
      "loss": 0.2183,
      "step": 8861
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4504848165717281,
      "learning_rate": 3.121955877427652e-06,
      "loss": 0.198,
      "step": 8862
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3593693491593615,
      "learning_rate": 3.1208823438107567e-06,
      "loss": 0.1876,
      "step": 8863
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.426057254864369,
      "learning_rate": 3.119808911057431e-06,
      "loss": 0.5131,
      "step": 8864
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1970410910589355,
      "learning_rate": 3.118735579225293e-06,
      "loss": 0.1796,
      "step": 8865
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3472276644147088,
      "learning_rate": 3.1176623483719576e-06,
      "loss": 0.2035,
      "step": 8866
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5903488520738773,
      "learning_rate": 3.1165892185550272e-06,
      "loss": 0.1749,
      "step": 8867
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.340465416073398,
      "learning_rate": 3.1155161898321064e-06,
      "loss": 0.573,
      "step": 8868
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2482834591975278,
      "learning_rate": 3.1144432622607877e-06,
      "loss": 0.1697,
      "step": 8869
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3581271568406463,
      "learning_rate": 3.1133704358986674e-06,
      "loss": 0.1806,
      "step": 8870
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3667894662750992,
      "learning_rate": 3.1122977108033237e-06,
      "loss": 0.19,
      "step": 8871
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2702473398470158,
      "learning_rate": 3.1112250870323406e-06,
      "loss": 0.1988,
      "step": 8872
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1920508736353224,
      "learning_rate": 3.110152564643291e-06,
      "loss": 0.1595,
      "step": 8873
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3842444809170014,
      "learning_rate": 3.1090801436937446e-06,
      "loss": 0.1713,
      "step": 8874
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.170675181142841,
      "learning_rate": 3.1080078242412627e-06,
      "loss": 0.6643,
      "step": 8875
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3145728982252256,
      "learning_rate": 3.106935606343404e-06,
      "loss": 0.1493,
      "step": 8876
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3405440973322789,
      "learning_rate": 3.1058634900577216e-06,
      "loss": 0.1683,
      "step": 8877
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3068131161695118,
      "learning_rate": 3.104791475441764e-06,
      "loss": 0.1933,
      "step": 8878
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3068759017012703,
      "learning_rate": 3.103719562553068e-06,
      "loss": 0.1948,
      "step": 8879
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6389880783838642,
      "learning_rate": 3.102647751449174e-06,
      "loss": 0.2163,
      "step": 8880
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2874779699311212,
      "learning_rate": 3.1015760421876095e-06,
      "loss": 0.1876,
      "step": 8881
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1211522667827096,
      "learning_rate": 3.100504434825903e-06,
      "loss": 0.1754,
      "step": 8882
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.30324175203119,
      "learning_rate": 3.0994329294215707e-06,
      "loss": 0.1571,
      "step": 8883
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2801131400738004,
      "learning_rate": 3.0983615260321286e-06,
      "loss": 0.1778,
      "step": 8884
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3919939624156619,
      "learning_rate": 3.097290224715084e-06,
      "loss": 0.2051,
      "step": 8885
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4459454853178553,
      "learning_rate": 3.0962190255279433e-06,
      "loss": 0.1754,
      "step": 8886
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.246065239189784,
      "learning_rate": 3.0951479285282005e-06,
      "loss": 0.1826,
      "step": 8887
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2788513383978244,
      "learning_rate": 3.094076933773349e-06,
      "loss": 0.1759,
      "step": 8888
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2755330168232222,
      "learning_rate": 3.093006041320876e-06,
      "loss": 0.1957,
      "step": 8889
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2536842737368088,
      "learning_rate": 3.091935251228264e-06,
      "loss": 0.1815,
      "step": 8890
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.348033848778666,
      "learning_rate": 3.0908645635529857e-06,
      "loss": 0.1638,
      "step": 8891
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2771167337999285,
      "learning_rate": 3.0897939783525156e-06,
      "loss": 0.1625,
      "step": 8892
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.297569785907774,
      "learning_rate": 3.0887234956843115e-06,
      "loss": 0.1791,
      "step": 8893
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.348054492099745,
      "learning_rate": 3.0876531156058407e-06,
      "loss": 0.631,
      "step": 8894
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3612267561762736,
      "learning_rate": 3.0865828381745515e-06,
      "loss": 0.1778,
      "step": 8895
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4527640516674205,
      "learning_rate": 3.085512663447894e-06,
      "loss": 0.2174,
      "step": 8896
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3765184528476222,
      "learning_rate": 3.0844425914833133e-06,
      "loss": 0.1893,
      "step": 8897
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.798712977944269,
      "learning_rate": 3.0833726223382397e-06,
      "loss": 0.712,
      "step": 8898
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2591647315777175,
      "learning_rate": 3.0823027560701135e-06,
      "loss": 0.2076,
      "step": 8899
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.487563793603447,
      "learning_rate": 3.081232992736355e-06,
      "loss": 0.5308,
      "step": 8900
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3901669595080166,
      "learning_rate": 3.080163332394388e-06,
      "loss": 0.2091,
      "step": 8901
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.342217962347694,
      "learning_rate": 3.0790937751016243e-06,
      "loss": 0.1731,
      "step": 8902
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5637813921643717,
      "learning_rate": 3.0780243209154763e-06,
      "loss": 0.1932,
      "step": 8903
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.317242390607728,
      "learning_rate": 3.0769549698933465e-06,
      "loss": 0.2203,
      "step": 8904
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3878957543756512,
      "learning_rate": 3.075885722092635e-06,
      "loss": 0.1745,
      "step": 8905
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1348643222694037,
      "learning_rate": 3.0748165775707334e-06,
      "loss": 0.151,
      "step": 8906
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1849760355659502,
      "learning_rate": 3.07374753638503e-06,
      "loss": 0.1486,
      "step": 8907
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.377211838259763,
      "learning_rate": 3.0726785985929048e-06,
      "loss": 0.186,
      "step": 8908
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5084746512590044,
      "learning_rate": 3.071609764251738e-06,
      "loss": 0.1906,
      "step": 8909
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3346883548232737,
      "learning_rate": 3.0705410334188957e-06,
      "loss": 0.1808,
      "step": 8910
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3721942395043114,
      "learning_rate": 3.069472406151747e-06,
      "loss": 0.1798,
      "step": 8911
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3603723918144288,
      "learning_rate": 3.068403882507648e-06,
      "loss": 0.1841,
      "step": 8912
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3566573889379783,
      "learning_rate": 3.0673354625439577e-06,
      "loss": 0.1915,
      "step": 8913
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3970179252006574,
      "learning_rate": 3.066267146318019e-06,
      "loss": 0.2056,
      "step": 8914
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3502037299082505,
      "learning_rate": 3.065198933887179e-06,
      "loss": 0.1861,
      "step": 8915
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3451761324297506,
      "learning_rate": 3.0641308253087722e-06,
      "loss": 0.2042,
      "step": 8916
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.2358359870780635,
      "learning_rate": 3.0630628206401347e-06,
      "loss": 0.4907,
      "step": 8917
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2881979292596126,
      "learning_rate": 3.0619949199385866e-06,
      "loss": 0.1797,
      "step": 8918
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1577687135728536,
      "learning_rate": 3.0609271232614545e-06,
      "loss": 0.1747,
      "step": 8919
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2601056733692304,
      "learning_rate": 3.059859430666049e-06,
      "loss": 0.1951,
      "step": 8920
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5329657652634363,
      "learning_rate": 3.0587918422096835e-06,
      "loss": 0.1786,
      "step": 8921
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.242628099751199,
      "learning_rate": 3.0577243579496583e-06,
      "loss": 0.1623,
      "step": 8922
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6859495916837106,
      "learning_rate": 3.056656977943274e-06,
      "loss": 0.2316,
      "step": 8923
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3299118202892588,
      "learning_rate": 3.0555897022478217e-06,
      "loss": 0.1731,
      "step": 8924
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1911641552700072,
      "learning_rate": 3.0545225309205918e-06,
      "loss": 0.1615,
      "step": 8925
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2296697063326794,
      "learning_rate": 3.0534554640188608e-06,
      "loss": 0.1558,
      "step": 8926
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3038618299436948,
      "learning_rate": 3.0523885015999076e-06,
      "loss": 0.1616,
      "step": 8927
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2461747968884576,
      "learning_rate": 3.051321643721003e-06,
      "loss": 0.1729,
      "step": 8928
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3000685290478593,
      "learning_rate": 3.05025489043941e-06,
      "loss": 0.1548,
      "step": 8929
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.209858723154252,
      "learning_rate": 3.049188241812391e-06,
      "loss": 0.1698,
      "step": 8930
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3707010463983653,
      "learning_rate": 3.0481216978971937e-06,
      "loss": 0.214,
      "step": 8931
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3565091191272802,
      "learning_rate": 3.0470552587510714e-06,
      "loss": 0.1824,
      "step": 8932
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.86158429004687,
      "learning_rate": 3.0459889244312636e-06,
      "loss": 0.6688,
      "step": 8933
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2400955520526493,
      "learning_rate": 3.0449226949950086e-06,
      "loss": 0.1771,
      "step": 8934
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4074892990993473,
      "learning_rate": 3.0438565704995343e-06,
      "loss": 0.2203,
      "step": 8935
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2586931432721307,
      "learning_rate": 3.0427905510020693e-06,
      "loss": 0.1606,
      "step": 8936
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.64271705484394,
      "learning_rate": 3.041724636559831e-06,
      "loss": 0.5625,
      "step": 8937
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.35176895919859,
      "learning_rate": 3.0406588272300357e-06,
      "loss": 0.1713,
      "step": 8938
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2083590681503207,
      "learning_rate": 3.0395931230698893e-06,
      "loss": 0.1327,
      "step": 8939
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2007659454264854,
      "learning_rate": 3.0385275241365965e-06,
      "loss": 0.1602,
      "step": 8940
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5824947839303294,
      "learning_rate": 3.0374620304873525e-06,
      "loss": 0.2264,
      "step": 8941
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3038507737673648,
      "learning_rate": 3.0363966421793518e-06,
      "loss": 0.1803,
      "step": 8942
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4668400931674868,
      "learning_rate": 3.0353313592697765e-06,
      "loss": 0.1921,
      "step": 8943
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3165143140182756,
      "learning_rate": 3.03426618181581e-06,
      "loss": 0.1989,
      "step": 8944
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4032072675005498,
      "learning_rate": 3.033201109874624e-06,
      "loss": 0.1965,
      "step": 8945
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1787711933468785,
      "learning_rate": 3.032136143503391e-06,
      "loss": 0.1651,
      "step": 8946
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.207891212759981,
      "learning_rate": 3.0310712827592696e-06,
      "loss": 0.5839,
      "step": 8947
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.432004812184872,
      "learning_rate": 3.030006527699421e-06,
      "loss": 0.2086,
      "step": 8948
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3290496851023192,
      "learning_rate": 3.028941878380994e-06,
      "loss": 0.1378,
      "step": 8949
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.317892079568536,
      "learning_rate": 3.027877334861139e-06,
      "loss": 0.1445,
      "step": 8950
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2476046313781441,
      "learning_rate": 3.026812897196992e-06,
      "loss": 0.1833,
      "step": 8951
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3325306475077288,
      "learning_rate": 3.0257485654456897e-06,
      "loss": 0.1937,
      "step": 8952
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3724534163945883,
      "learning_rate": 3.024684339664361e-06,
      "loss": 0.1922,
      "step": 8953
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.341381314770079,
      "learning_rate": 3.0236202199101313e-06,
      "loss": 0.1858,
      "step": 8954
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.260299981318571,
      "learning_rate": 3.0225562062401138e-06,
      "loss": 0.152,
      "step": 8955
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4059760816998603,
      "learning_rate": 3.0214922987114247e-06,
      "loss": 0.1754,
      "step": 8956
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.527796990144992,
      "learning_rate": 3.020428497381168e-06,
      "loss": 0.5417,
      "step": 8957
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.242402686040406,
      "learning_rate": 3.019364802306447e-06,
      "loss": 0.1606,
      "step": 8958
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2048144410810482,
      "learning_rate": 3.0183012135443525e-06,
      "loss": 0.1609,
      "step": 8959
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3517809032625816,
      "learning_rate": 3.017237731151976e-06,
      "loss": 0.1691,
      "step": 8960
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2806943695435553,
      "learning_rate": 3.016174355186402e-06,
      "loss": 0.1759,
      "step": 8961
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2983962180398247,
      "learning_rate": 3.015111085704707e-06,
      "loss": 0.1902,
      "step": 8962
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4696046077114788,
      "learning_rate": 3.0140479227639643e-06,
      "loss": 0.2091,
      "step": 8963
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.273598698202574,
      "learning_rate": 3.012984866421238e-06,
      "loss": 0.1679,
      "step": 8964
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2531709262376356,
      "learning_rate": 3.0119219167335913e-06,
      "loss": 0.1624,
      "step": 8965
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.314800932334077,
      "learning_rate": 3.010859073758076e-06,
      "loss": 0.1699,
      "step": 8966
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.346193728445068,
      "learning_rate": 3.009796337551746e-06,
      "loss": 0.1526,
      "step": 8967
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.36693171164487,
      "learning_rate": 3.0087337081716393e-06,
      "loss": 0.1807,
      "step": 8968
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.5239613128363265,
      "learning_rate": 3.0076711856747977e-06,
      "loss": 0.4946,
      "step": 8969
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2577052715276438,
      "learning_rate": 3.0066087701182513e-06,
      "loss": 0.1412,
      "step": 8970
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1530185894253566,
      "learning_rate": 3.005546461559028e-06,
      "loss": 0.1667,
      "step": 8971
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4711722568069157,
      "learning_rate": 3.0044842600541447e-06,
      "loss": 0.129,
      "step": 8972
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5519954306432573,
      "learning_rate": 3.00342216566062e-06,
      "loss": 0.1496,
      "step": 8973
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3178250918891812,
      "learning_rate": 3.0023601784354605e-06,
      "loss": 0.181,
      "step": 8974
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3073732517174987,
      "learning_rate": 3.001298298435673e-06,
      "loss": 0.2126,
      "step": 8975
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4359627401075035,
      "learning_rate": 3.0002365257182496e-06,
      "loss": 0.1924,
      "step": 8976
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3799339486399935,
      "learning_rate": 2.999174860340186e-06,
      "loss": 0.1562,
      "step": 8977
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2088711192617874,
      "learning_rate": 2.9981133023584664e-06,
      "loss": 0.1718,
      "step": 8978
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3244857076038692,
      "learning_rate": 2.997051851830074e-06,
      "loss": 0.199,
      "step": 8979
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2427599981070234,
      "learning_rate": 2.9959905088119777e-06,
      "loss": 0.1566,
      "step": 8980
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4249666536763284,
      "learning_rate": 2.9949292733611512e-06,
      "loss": 0.1968,
      "step": 8981
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2100262496124679,
      "learning_rate": 2.9938681455345542e-06,
      "loss": 0.1577,
      "step": 8982
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3810841964683973,
      "learning_rate": 2.9928071253891476e-06,
      "loss": 0.1722,
      "step": 8983
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3688845375925671,
      "learning_rate": 2.9917462129818785e-06,
      "loss": 0.1841,
      "step": 8984
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2647315028134762,
      "learning_rate": 2.9906854083696947e-06,
      "loss": 0.1625,
      "step": 8985
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2471020906403547,
      "learning_rate": 2.989624711609535e-06,
      "loss": 0.1847,
      "step": 8986
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2833137779985297,
      "learning_rate": 2.988564122758336e-06,
      "loss": 0.1822,
      "step": 8987
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2538857701666606,
      "learning_rate": 2.9875036418730218e-06,
      "loss": 0.2055,
      "step": 8988
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4175681266325126,
      "learning_rate": 2.986443269010519e-06,
      "loss": 0.2021,
      "step": 8989
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4247700466053956,
      "learning_rate": 2.985383004227739e-06,
      "loss": 0.1613,
      "step": 8990
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3569114641050322,
      "learning_rate": 2.9843228475815965e-06,
      "loss": 0.1875,
      "step": 8991
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4549816871284327,
      "learning_rate": 2.9832627991289973e-06,
      "loss": 0.185,
      "step": 8992
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2768614512045962,
      "learning_rate": 2.9822028589268383e-06,
      "loss": 0.1786,
      "step": 8993
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.31323651838234,
      "learning_rate": 2.981143027032013e-06,
      "loss": 0.1996,
      "step": 8994
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2041169749711287,
      "learning_rate": 2.9800833035014098e-06,
      "loss": 0.2167,
      "step": 8995
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2477360425933492,
      "learning_rate": 2.979023688391911e-06,
      "loss": 0.1434,
      "step": 8996
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2769529395078603,
      "learning_rate": 2.977964181760391e-06,
      "loss": 0.2005,
      "step": 8997
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4065839791151251,
      "learning_rate": 2.976904783663722e-06,
      "loss": 0.1961,
      "step": 8998
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.95403982604042,
      "learning_rate": 2.9758454941587632e-06,
      "loss": 0.5831,
      "step": 8999
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.116017101658174,
      "learning_rate": 2.9747863133023803e-06,
      "loss": 0.1446,
      "step": 9000
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2536273900457988,
      "learning_rate": 2.9737272411514207e-06,
      "loss": 0.1553,
      "step": 9001
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.322677442441434,
      "learning_rate": 2.972668277762735e-06,
      "loss": 0.1822,
      "step": 9002
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3925139595523468,
      "learning_rate": 2.971609423193159e-06,
      "loss": 0.1786,
      "step": 9003
    },
    {
      "epoch": 0.64,
      "grad_norm": 9.047854841153445,
      "learning_rate": 2.970550677499532e-06,
      "loss": 0.6008,
      "step": 9004
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5273823418147014,
      "learning_rate": 2.969492040738682e-06,
      "loss": 0.2036,
      "step": 9005
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3815385345536884,
      "learning_rate": 2.9684335129674334e-06,
      "loss": 0.1894,
      "step": 9006
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5948521083294915,
      "learning_rate": 2.9673750942426015e-06,
      "loss": 0.1929,
      "step": 9007
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1879092678524734,
      "learning_rate": 2.966316784621e-06,
      "loss": 0.1244,
      "step": 9008
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1778206912142821,
      "learning_rate": 2.9652585841594327e-06,
      "loss": 0.1611,
      "step": 9009
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.334322094811316,
      "learning_rate": 2.9642004929147037e-06,
      "loss": 0.1662,
      "step": 9010
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3680612768090985,
      "learning_rate": 2.9631425109436022e-06,
      "loss": 0.1372,
      "step": 9011
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.299897327149625,
      "learning_rate": 2.96208463830292e-06,
      "loss": 0.1636,
      "step": 9012
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1906746982821015,
      "learning_rate": 2.961026875049437e-06,
      "loss": 0.1291,
      "step": 9013
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.502762352376142,
      "learning_rate": 2.959969221239933e-06,
      "loss": 0.191,
      "step": 9014
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2899640855085186,
      "learning_rate": 2.9589116769311744e-06,
      "loss": 0.1581,
      "step": 9015
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.430895570738022,
      "learning_rate": 2.9578542421799296e-06,
      "loss": 0.2104,
      "step": 9016
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3954055199146087,
      "learning_rate": 2.9567969170429543e-06,
      "loss": 0.1579,
      "step": 9017
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4727090079464,
      "learning_rate": 2.9557397015770063e-06,
      "loss": 0.2301,
      "step": 9018
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2900463999664125,
      "learning_rate": 2.954682595838827e-06,
      "loss": 0.1566,
      "step": 9019
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3134563279302998,
      "learning_rate": 2.9536255998851615e-06,
      "loss": 0.149,
      "step": 9020
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2630352005867114,
      "learning_rate": 2.952568713772743e-06,
      "loss": 0.161,
      "step": 9021
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3921550032241177,
      "learning_rate": 2.9515119375583036e-06,
      "loss": 0.1977,
      "step": 9022
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3444820193727198,
      "learning_rate": 2.9504552712985635e-06,
      "loss": 0.2174,
      "step": 9023
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2562094881397003,
      "learning_rate": 2.949398715050242e-06,
      "loss": 0.165,
      "step": 9024
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2811385299677427,
      "learning_rate": 2.9483422688700513e-06,
      "loss": 0.1741,
      "step": 9025
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.921001978261553,
      "learning_rate": 2.947285932814697e-06,
      "loss": 0.5729,
      "step": 9026
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3143679551853817,
      "learning_rate": 2.9462297069408798e-06,
      "loss": 0.1907,
      "step": 9027
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3492723006377716,
      "learning_rate": 2.9451735913052916e-06,
      "loss": 0.177,
      "step": 9028
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.138793696274279,
      "learning_rate": 2.944117585964622e-06,
      "loss": 0.589,
      "step": 9029
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3779049601013813,
      "learning_rate": 2.9430616909755515e-06,
      "loss": 0.2178,
      "step": 9030
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1290230378198411,
      "learning_rate": 2.9420059063947605e-06,
      "loss": 0.1613,
      "step": 9031
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2745577473478662,
      "learning_rate": 2.940950232278914e-06,
      "loss": 0.1617,
      "step": 9032
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3087596392562098,
      "learning_rate": 2.9398946686846797e-06,
      "loss": 0.1752,
      "step": 9033
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.506951213251745,
      "learning_rate": 2.9388392156687145e-06,
      "loss": 0.2133,
      "step": 9034
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.350795108913614,
      "learning_rate": 2.937783873287674e-06,
      "loss": 0.2114,
      "step": 9035
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1806920529447555,
      "learning_rate": 2.9367286415982e-06,
      "loss": 0.1663,
      "step": 9036
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3584325124931473,
      "learning_rate": 2.9356735206569365e-06,
      "loss": 0.1392,
      "step": 9037
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1690467394505801,
      "learning_rate": 2.934618510520516e-06,
      "loss": 0.1405,
      "step": 9038
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3011541070499342,
      "learning_rate": 2.9335636112455714e-06,
      "loss": 0.1717,
      "step": 9039
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2464153596213308,
      "learning_rate": 2.93250882288872e-06,
      "loss": 0.1418,
      "step": 9040
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6590000467747565,
      "learning_rate": 2.9314541455065827e-06,
      "loss": 0.1878,
      "step": 9041
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2129390096594173,
      "learning_rate": 2.930399579155767e-06,
      "loss": 0.181,
      "step": 9042
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4688911300417744,
      "learning_rate": 2.9293451238928828e-06,
      "loss": 0.2074,
      "step": 9043
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2847687418643121,
      "learning_rate": 2.9282907797745236e-06,
      "loss": 0.1448,
      "step": 9044
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3190927229070775,
      "learning_rate": 2.927236546857286e-06,
      "loss": 0.1724,
      "step": 9045
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1006115878611313,
      "learning_rate": 2.926182425197755e-06,
      "loss": 0.129,
      "step": 9046
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9919982585784483,
      "learning_rate": 2.925128414852515e-06,
      "loss": 0.1058,
      "step": 9047
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4020618619342518,
      "learning_rate": 2.9240745158781357e-06,
      "loss": 0.2155,
      "step": 9048
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.383480464557657,
      "learning_rate": 2.9230207283311908e-06,
      "loss": 0.2162,
      "step": 9049
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4240563442996321,
      "learning_rate": 2.921967052268241e-06,
      "loss": 0.1918,
      "step": 9050
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4480067377313433,
      "learning_rate": 2.9209134877458457e-06,
      "loss": 0.179,
      "step": 9051
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.728263846645501,
      "learning_rate": 2.919860034820553e-06,
      "loss": 0.6956,
      "step": 9052
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.312509461221239,
      "learning_rate": 2.91880669354891e-06,
      "loss": 0.5956,
      "step": 9053
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2563410389005534,
      "learning_rate": 2.917753463987455e-06,
      "loss": 0.1447,
      "step": 9054
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6161052401027491,
      "learning_rate": 2.916700346192722e-06,
      "loss": 0.1818,
      "step": 9055
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4791989778215424,
      "learning_rate": 2.9156473402212394e-06,
      "loss": 0.1974,
      "step": 9056
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4014265023359394,
      "learning_rate": 2.914594446129525e-06,
      "loss": 0.1728,
      "step": 9057
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2746216672457429,
      "learning_rate": 2.9135416639740987e-06,
      "loss": 0.1781,
      "step": 9058
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3753562975862914,
      "learning_rate": 2.912488993811464e-06,
      "loss": 0.2074,
      "step": 9059
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2971239740076461,
      "learning_rate": 2.9114364356981274e-06,
      "loss": 0.1685,
      "step": 9060
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.537578315723535,
      "learning_rate": 2.910383989690585e-06,
      "loss": 0.1862,
      "step": 9061
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.296001205770126,
      "learning_rate": 2.90933165584533e-06,
      "loss": 0.1657,
      "step": 9062
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.187497057145445,
      "learning_rate": 2.908279434218845e-06,
      "loss": 0.1431,
      "step": 9063
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3406397043285756,
      "learning_rate": 2.9072273248676107e-06,
      "loss": 0.1961,
      "step": 9064
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2837856155197962,
      "learning_rate": 2.9061753278480987e-06,
      "loss": 0.1689,
      "step": 9065
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.177606595634527,
      "learning_rate": 2.9051234432167786e-06,
      "loss": 0.1936,
      "step": 9066
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1449900086891058,
      "learning_rate": 2.904071671030107e-06,
      "loss": 0.1611,
      "step": 9067
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.39725572436349,
      "learning_rate": 2.9030200113445415e-06,
      "loss": 0.1912,
      "step": 9068
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.690426718452229,
      "learning_rate": 2.901968464216531e-06,
      "loss": 0.6479,
      "step": 9069
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2779487289619416,
      "learning_rate": 2.900917029702521e-06,
      "loss": 0.1517,
      "step": 9070
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3601654246466315,
      "learning_rate": 2.8998657078589424e-06,
      "loss": 0.2039,
      "step": 9071
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5263630373187766,
      "learning_rate": 2.8988144987422325e-06,
      "loss": 0.2035,
      "step": 9072
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2719445102531923,
      "learning_rate": 2.89776340240881e-06,
      "loss": 0.1824,
      "step": 9073
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.909376100349962,
      "learning_rate": 2.8967124189150985e-06,
      "loss": 0.2139,
      "step": 9074
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.749218754821759,
      "learning_rate": 2.895661548317506e-06,
      "loss": 0.5513,
      "step": 9075
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2558987075772945,
      "learning_rate": 2.894610790672442e-06,
      "loss": 0.1612,
      "step": 9076
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4662531947988435,
      "learning_rate": 2.893560146036306e-06,
      "loss": 0.2098,
      "step": 9077
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4402046517144116,
      "learning_rate": 2.8925096144654956e-06,
      "loss": 0.1849,
      "step": 9078
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2424667475979911,
      "learning_rate": 2.891459196016394e-06,
      "loss": 0.1827,
      "step": 9079
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1925889059076282,
      "learning_rate": 2.8904088907453887e-06,
      "loss": 0.1557,
      "step": 9080
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5071488789489207,
      "learning_rate": 2.8893586987088502e-06,
      "loss": 0.1963,
      "step": 9081
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4459193823704106,
      "learning_rate": 2.888308619963154e-06,
      "loss": 0.1829,
      "step": 9082
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4476566453499513,
      "learning_rate": 2.8872586545646597e-06,
      "loss": 0.1851,
      "step": 9083
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3245811867298163,
      "learning_rate": 2.886208802569728e-06,
      "loss": 0.1841,
      "step": 9084
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4738373197165826,
      "learning_rate": 2.8851590640347094e-06,
      "loss": 0.2039,
      "step": 9085
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2732079501974507,
      "learning_rate": 2.884109439015952e-06,
      "loss": 0.2151,
      "step": 9086
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2783253958145018,
      "learning_rate": 2.883059927569793e-06,
      "loss": 0.1564,
      "step": 9087
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2752686393749104,
      "learning_rate": 2.8820105297525664e-06,
      "loss": 0.1549,
      "step": 9088
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3687513795261879,
      "learning_rate": 2.880961245620602e-06,
      "loss": 0.1876,
      "step": 9089
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4022633759405605,
      "learning_rate": 2.879912075230218e-06,
      "loss": 0.1781,
      "step": 9090
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.307045836353741,
      "learning_rate": 2.878863018637733e-06,
      "loss": 0.1985,
      "step": 9091
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3820230734764989,
      "learning_rate": 2.877814075899452e-06,
      "loss": 0.1791,
      "step": 9092
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2296235227739558,
      "learning_rate": 2.8767652470716805e-06,
      "loss": 0.1497,
      "step": 9093
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3645595110393514,
      "learning_rate": 2.8757165322107156e-06,
      "loss": 0.1732,
      "step": 9094
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.232719967045591,
      "learning_rate": 2.87466793137285e-06,
      "loss": 0.1538,
      "step": 9095
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3449034208731279,
      "learning_rate": 2.873619444614364e-06,
      "loss": 0.1837,
      "step": 9096
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3438380185872054,
      "learning_rate": 2.8725710719915406e-06,
      "loss": 0.208,
      "step": 9097
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6930454655398925,
      "learning_rate": 2.871522813560648e-06,
      "loss": 0.1669,
      "step": 9098
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.112427403316681,
      "learning_rate": 2.8704746693779573e-06,
      "loss": 0.5536,
      "step": 9099
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2130588102525823,
      "learning_rate": 2.8694266394997238e-06,
      "loss": 0.1514,
      "step": 9100
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4631426558518188,
      "learning_rate": 2.868378723982206e-06,
      "loss": 0.1789,
      "step": 9101
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.614652285387004,
      "learning_rate": 2.867330922881648e-06,
      "loss": 0.2069,
      "step": 9102
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2544767784979112,
      "learning_rate": 2.8662832362542934e-06,
      "loss": 0.1725,
      "step": 9103
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.583331225964863,
      "learning_rate": 2.8652356641563783e-06,
      "loss": 0.2405,
      "step": 9104
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2748638191999488,
      "learning_rate": 2.8641882066441336e-06,
      "loss": 0.1553,
      "step": 9105
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3181552108068055,
      "learning_rate": 2.8631408637737784e-06,
      "loss": 0.1817,
      "step": 9106
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2913770516319223,
      "learning_rate": 2.862093635601535e-06,
      "loss": 0.1582,
      "step": 9107
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.310057142195116,
      "learning_rate": 2.8610465221836094e-06,
      "loss": 0.6924,
      "step": 9108
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3511973881496233,
      "learning_rate": 2.85999952357621e-06,
      "loss": 0.1821,
      "step": 9109
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3148709285229379,
      "learning_rate": 2.858952639835534e-06,
      "loss": 0.1987,
      "step": 9110
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.24549234908076,
      "learning_rate": 2.8579058710177733e-06,
      "loss": 0.1253,
      "step": 9111
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.733868103061433,
      "learning_rate": 2.8568592171791156e-06,
      "loss": 0.5707,
      "step": 9112
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1403666249966755,
      "learning_rate": 2.8558126783757424e-06,
      "loss": 0.1475,
      "step": 9113
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5130716516099876,
      "learning_rate": 2.8547662546638248e-06,
      "loss": 0.1922,
      "step": 9114
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3140950004612044,
      "learning_rate": 2.853719946099534e-06,
      "loss": 0.1616,
      "step": 9115
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2622954116082894,
      "learning_rate": 2.852673752739028e-06,
      "loss": 0.1759,
      "step": 9116
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.218053777558217,
      "learning_rate": 2.8516276746384656e-06,
      "loss": 0.1535,
      "step": 9117
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.222865556753346,
      "learning_rate": 2.850581711853993e-06,
      "loss": 0.1332,
      "step": 9118
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1454488605550988,
      "learning_rate": 2.849535864441755e-06,
      "loss": 0.154,
      "step": 9119
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4973720088212403,
      "learning_rate": 2.8484901324578883e-06,
      "loss": 0.221,
      "step": 9120
    },
    {
      "epoch": 0.65,
      "grad_norm": 10.722205533689605,
      "learning_rate": 2.8474445159585235e-06,
      "loss": 0.5296,
      "step": 9121
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3466521190940732,
      "learning_rate": 2.8463990149997877e-06,
      "loss": 0.1998,
      "step": 9122
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5686678226019848,
      "learning_rate": 2.8453536296377948e-06,
      "loss": 0.2467,
      "step": 9123
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4203469811446856,
      "learning_rate": 2.8443083599286615e-06,
      "loss": 0.1818,
      "step": 9124
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3290866520438898,
      "learning_rate": 2.8432632059284893e-06,
      "loss": 0.178,
      "step": 9125
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5623409548955918,
      "learning_rate": 2.842218167693381e-06,
      "loss": 0.2708,
      "step": 9126
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4246139366756787,
      "learning_rate": 2.8411732452794284e-06,
      "loss": 0.1496,
      "step": 9127
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2972264778860447,
      "learning_rate": 2.8401284387427185e-06,
      "loss": 0.1708,
      "step": 9128
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3209367647959487,
      "learning_rate": 2.8390837481393335e-06,
      "loss": 0.136,
      "step": 9129
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2561911045433416,
      "learning_rate": 2.83803917352535e-06,
      "loss": 0.178,
      "step": 9130
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.174680667521737,
      "learning_rate": 2.836994714956833e-06,
      "loss": 0.1475,
      "step": 9131
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.732567710659345,
      "learning_rate": 2.8359503724898485e-06,
      "loss": 0.5529,
      "step": 9132
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4348805872214954,
      "learning_rate": 2.8349061461804483e-06,
      "loss": 0.1601,
      "step": 9133
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4528042419704412,
      "learning_rate": 2.833862036084687e-06,
      "loss": 0.1781,
      "step": 9134
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2746644933427882,
      "learning_rate": 2.8328180422586036e-06,
      "loss": 0.1695,
      "step": 9135
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4331759495519956,
      "learning_rate": 2.8317741647582385e-06,
      "loss": 0.1553,
      "step": 9136
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.9043904215217404,
      "learning_rate": 2.8307304036396222e-06,
      "loss": 0.3483,
      "step": 9137
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.374350982287738,
      "learning_rate": 2.829686758958782e-06,
      "loss": 0.1613,
      "step": 9138
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.390443321937747,
      "learning_rate": 2.8286432307717315e-06,
      "loss": 0.5265,
      "step": 9139
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2727971354479681,
      "learning_rate": 2.827599819134489e-06,
      "loss": 0.1713,
      "step": 9140
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.285165735343625,
      "learning_rate": 2.8265565241030547e-06,
      "loss": 0.1679,
      "step": 9141
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.440593107010905,
      "learning_rate": 2.8255133457334343e-06,
      "loss": 0.1953,
      "step": 9142
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.446254804985315,
      "learning_rate": 2.8244702840816163e-06,
      "loss": 0.1885,
      "step": 9143
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1311931396842339,
      "learning_rate": 2.823427339203591e-06,
      "loss": 0.1357,
      "step": 9144
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.531695091559079,
      "learning_rate": 2.8223845111553388e-06,
      "loss": 0.2032,
      "step": 9145
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.340532428074617,
      "learning_rate": 2.821341799992837e-06,
      "loss": 0.1908,
      "step": 9146
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.213585702827845,
      "learning_rate": 2.82029920577205e-06,
      "loss": 0.124,
      "step": 9147
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2553022926886956,
      "learning_rate": 2.8192567285489434e-06,
      "loss": 0.1536,
      "step": 9148
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4772475540394465,
      "learning_rate": 2.818214368379471e-06,
      "loss": 0.2383,
      "step": 9149
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4770072465137571,
      "learning_rate": 2.817172125319585e-06,
      "loss": 0.1643,
      "step": 9150
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.300381001786014,
      "learning_rate": 2.8161299994252255e-06,
      "loss": 0.157,
      "step": 9151
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2452223147923527,
      "learning_rate": 2.815087990752332e-06,
      "loss": 0.1561,
      "step": 9152
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1776467473936596,
      "learning_rate": 2.8140460993568348e-06,
      "loss": 0.139,
      "step": 9153
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.578830121928663,
      "learning_rate": 2.813004325294658e-06,
      "loss": 0.6898,
      "step": 9154
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.308938250782106,
      "learning_rate": 2.8119626686217235e-06,
      "loss": 0.1578,
      "step": 9155
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2630429962404797,
      "learning_rate": 2.8109211293939376e-06,
      "loss": 0.2092,
      "step": 9156
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.346525557804274,
      "learning_rate": 2.8098797076672114e-06,
      "loss": 0.168,
      "step": 9157
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.435339094381345,
      "learning_rate": 2.8088384034974385e-06,
      "loss": 0.1855,
      "step": 9158
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4088798707848207,
      "learning_rate": 2.8077972169405175e-06,
      "loss": 0.2067,
      "step": 9159
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.177345398638196,
      "learning_rate": 2.8067561480523315e-06,
      "loss": 0.167,
      "step": 9160
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.696041322348422,
      "learning_rate": 2.8057151968887618e-06,
      "loss": 0.6083,
      "step": 9161
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.265258120934531,
      "learning_rate": 2.8046743635056828e-06,
      "loss": 0.1888,
      "step": 9162
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3942502139023836,
      "learning_rate": 2.8036336479589644e-06,
      "loss": 0.1653,
      "step": 9163
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.384709645024829,
      "learning_rate": 2.802593050304464e-06,
      "loss": 0.1835,
      "step": 9164
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.212484651119731,
      "learning_rate": 2.8015525705980405e-06,
      "loss": 0.1737,
      "step": 9165
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.286298711861579,
      "learning_rate": 2.8005122088955395e-06,
      "loss": 0.183,
      "step": 9166
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4490900745105453,
      "learning_rate": 2.799471965252807e-06,
      "loss": 0.2291,
      "step": 9167
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3634770143179573,
      "learning_rate": 2.798431839725675e-06,
      "loss": 0.1945,
      "step": 9168
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2895287728446805,
      "learning_rate": 2.7973918323699756e-06,
      "loss": 0.1488,
      "step": 9169
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4185321326779774,
      "learning_rate": 2.796351943241532e-06,
      "loss": 0.2199,
      "step": 9170
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3432723184361033,
      "learning_rate": 2.795312172396164e-06,
      "loss": 0.1527,
      "step": 9171
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.412181900464206,
      "learning_rate": 2.794272519889677e-06,
      "loss": 0.1634,
      "step": 9172
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3334300851699827,
      "learning_rate": 2.7932329857778807e-06,
      "loss": 0.1874,
      "step": 9173
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4641846326819397,
      "learning_rate": 2.7921935701165686e-06,
      "loss": 0.1524,
      "step": 9174
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.360484250754988,
      "learning_rate": 2.791154272961537e-06,
      "loss": 0.1512,
      "step": 9175
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5298355751829058,
      "learning_rate": 2.790115094368566e-06,
      "loss": 0.1794,
      "step": 9176
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6649370543017161,
      "learning_rate": 2.7890760343934377e-06,
      "loss": 0.1544,
      "step": 9177
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2767568785959826,
      "learning_rate": 2.7880370930919243e-06,
      "loss": 0.1533,
      "step": 9178
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3520757758417188,
      "learning_rate": 2.786998270519794e-06,
      "loss": 0.2153,
      "step": 9179
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4642949250540587,
      "learning_rate": 2.7859595667328027e-06,
      "loss": 0.1927,
      "step": 9180
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5527659305911454,
      "learning_rate": 2.7849209817867085e-06,
      "loss": 0.1869,
      "step": 9181
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2480577221849667,
      "learning_rate": 2.7838825157372527e-06,
      "loss": 0.1688,
      "step": 9182
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3030919870143174,
      "learning_rate": 2.7828441686401807e-06,
      "loss": 0.1707,
      "step": 9183
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4262679733622055,
      "learning_rate": 2.7818059405512267e-06,
      "loss": 0.1884,
      "step": 9184
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2281805717888794,
      "learning_rate": 2.7807678315261155e-06,
      "loss": 0.1805,
      "step": 9185
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3617106125711933,
      "learning_rate": 2.77972984162057e-06,
      "loss": 0.1759,
      "step": 9186
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3542284249040688,
      "learning_rate": 2.778691970890307e-06,
      "loss": 0.1989,
      "step": 9187
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2899810532209837,
      "learning_rate": 2.777654219391036e-06,
      "loss": 0.1654,
      "step": 9188
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.426899245331698,
      "learning_rate": 2.776616587178455e-06,
      "loss": 0.1804,
      "step": 9189
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3285160228805808,
      "learning_rate": 2.775579074308265e-06,
      "loss": 0.2092,
      "step": 9190
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4613867719211835,
      "learning_rate": 2.774541680836151e-06,
      "loss": 0.1712,
      "step": 9191
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1779294228018107,
      "learning_rate": 2.7735044068178e-06,
      "loss": 0.138,
      "step": 9192
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1790873204804098,
      "learning_rate": 2.7724672523088857e-06,
      "loss": 0.1298,
      "step": 9193
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3240577487046459,
      "learning_rate": 2.771430217365081e-06,
      "loss": 0.1978,
      "step": 9194
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3361666719078507,
      "learning_rate": 2.7703933020420453e-06,
      "loss": 0.1552,
      "step": 9195
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.449438649050825,
      "learning_rate": 2.769356506395443e-06,
      "loss": 0.1892,
      "step": 9196
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.396391739398802,
      "learning_rate": 2.76831983048092e-06,
      "loss": 0.1726,
      "step": 9197
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3060722383166927,
      "learning_rate": 2.767283274354124e-06,
      "loss": 0.1446,
      "step": 9198
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.89992247238176,
      "learning_rate": 2.7662468380706897e-06,
      "loss": 0.5954,
      "step": 9199
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1793376818854633,
      "learning_rate": 2.7652105216862536e-06,
      "loss": 0.1465,
      "step": 9200
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.834673135848603,
      "learning_rate": 2.7641743252564355e-06,
      "loss": 0.5286,
      "step": 9201
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3412421253854436,
      "learning_rate": 2.7631382488368598e-06,
      "loss": 0.1694,
      "step": 9202
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.487649754527599,
      "learning_rate": 2.7621022924831337e-06,
      "loss": 0.2221,
      "step": 9203
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1986840182196448,
      "learning_rate": 2.761066456250866e-06,
      "loss": 0.1717,
      "step": 9204
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5059505157009987,
      "learning_rate": 2.7600307401956565e-06,
      "loss": 0.2336,
      "step": 9205
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.346391640069979,
      "learning_rate": 2.7589951443731e-06,
      "loss": 0.1972,
      "step": 9206
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5097187839877142,
      "learning_rate": 2.757959668838779e-06,
      "loss": 0.2404,
      "step": 9207
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3159413497478796,
      "learning_rate": 2.756924313648277e-06,
      "loss": 0.1851,
      "step": 9208
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4321556356958238,
      "learning_rate": 2.7558890788571657e-06,
      "loss": 0.1942,
      "step": 9209
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.461317887693462,
      "learning_rate": 2.754853964521015e-06,
      "loss": 0.4871,
      "step": 9210
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3715587534979456,
      "learning_rate": 2.7538189706953812e-06,
      "loss": 0.2329,
      "step": 9211
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4394626112279234,
      "learning_rate": 2.752784097435822e-06,
      "loss": 0.1903,
      "step": 9212
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.616960026516626,
      "learning_rate": 2.751749344797885e-06,
      "loss": 0.4282,
      "step": 9213
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2756494240633804,
      "learning_rate": 2.750714712837113e-06,
      "loss": 0.1727,
      "step": 9214
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3693924493472256,
      "learning_rate": 2.749680201609037e-06,
      "loss": 0.1667,
      "step": 9215
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3591515202456343,
      "learning_rate": 2.748645811169189e-06,
      "loss": 0.169,
      "step": 9216
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2740030145018901,
      "learning_rate": 2.7476115415730906e-06,
      "loss": 0.207,
      "step": 9217
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5061819690401377,
      "learning_rate": 2.7465773928762546e-06,
      "loss": 0.146,
      "step": 9218
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3130471229954106,
      "learning_rate": 2.745543365134194e-06,
      "loss": 0.1776,
      "step": 9219
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3722361122194406,
      "learning_rate": 2.7445094584024067e-06,
      "loss": 0.1792,
      "step": 9220
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3214313491408873,
      "learning_rate": 2.743475672736391e-06,
      "loss": 0.1562,
      "step": 9221
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.216806364554223,
      "learning_rate": 2.742442008191637e-06,
      "loss": 0.1566,
      "step": 9222
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5274689212737964,
      "learning_rate": 2.741408464823629e-06,
      "loss": 0.1697,
      "step": 9223
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3116436787574242,
      "learning_rate": 2.7403750426878394e-06,
      "loss": 0.1733,
      "step": 9224
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.400038341916177,
      "learning_rate": 2.739341741839743e-06,
      "loss": 0.2095,
      "step": 9225
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3957557592259002,
      "learning_rate": 2.7383085623347983e-06,
      "loss": 0.178,
      "step": 9226
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4805199707811998,
      "learning_rate": 2.737275504228467e-06,
      "loss": 0.1885,
      "step": 9227
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.409596295065184,
      "learning_rate": 2.7362425675761955e-06,
      "loss": 0.1808,
      "step": 9228
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.208954979052102,
      "learning_rate": 2.735209752433429e-06,
      "loss": 0.122,
      "step": 9229
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4108146234348289,
      "learning_rate": 2.734177058855606e-06,
      "loss": 0.1874,
      "step": 9230
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2695486863214673,
      "learning_rate": 2.733144486898159e-06,
      "loss": 0.1549,
      "step": 9231
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2980650207421978,
      "learning_rate": 2.7321120366165075e-06,
      "loss": 0.1871,
      "step": 9232
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.39149004498744,
      "learning_rate": 2.7310797080660746e-06,
      "loss": 0.1848,
      "step": 9233
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3966372159945377,
      "learning_rate": 2.7300475013022666e-06,
      "loss": 0.1968,
      "step": 9234
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2854290403330828,
      "learning_rate": 2.7290154163804928e-06,
      "loss": 0.189,
      "step": 9235
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.27864775455434,
      "learning_rate": 2.7279834533561477e-06,
      "loss": 0.6036,
      "step": 9236
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.281735888003445,
      "learning_rate": 2.726951612284624e-06,
      "loss": 0.1967,
      "step": 9237
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2513717817598022,
      "learning_rate": 2.7259198932213078e-06,
      "loss": 0.1764,
      "step": 9238
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.373489243001319,
      "learning_rate": 2.7248882962215794e-06,
      "loss": 0.4761,
      "step": 9239
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.318781985113303,
      "learning_rate": 2.723856821340806e-06,
      "loss": 0.1839,
      "step": 9240
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2637637233589352,
      "learning_rate": 2.722825468634358e-06,
      "loss": 0.1879,
      "step": 9241
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.517800726371777,
      "learning_rate": 2.7217942381575913e-06,
      "loss": 0.7799,
      "step": 9242
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.526346952652297,
      "learning_rate": 2.72076312996586e-06,
      "loss": 0.1882,
      "step": 9243
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3204087250759695,
      "learning_rate": 2.719732144114507e-06,
      "loss": 0.2194,
      "step": 9244
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.163283456153252,
      "learning_rate": 2.718701280658873e-06,
      "loss": 0.1904,
      "step": 9245
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4052492410158965,
      "learning_rate": 2.7176705396542925e-06,
      "loss": 0.2031,
      "step": 9246
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.104724609128661,
      "learning_rate": 2.716639921156089e-06,
      "loss": 0.1386,
      "step": 9247
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.034995552793996,
      "learning_rate": 2.7156094252195858e-06,
      "loss": 0.572,
      "step": 9248
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.490733837958997,
      "learning_rate": 2.7145790519000912e-06,
      "loss": 0.1751,
      "step": 9249
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.329690949967403,
      "learning_rate": 2.7135488012529154e-06,
      "loss": 0.1545,
      "step": 9250
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1595356314013299,
      "learning_rate": 2.712518673333354e-06,
      "loss": 0.154,
      "step": 9251
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3548915620337294,
      "learning_rate": 2.711488668196706e-06,
      "loss": 0.1801,
      "step": 9252
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3175768296350248,
      "learning_rate": 2.7104587858982514e-06,
      "loss": 0.1944,
      "step": 9253
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2982247766862691,
      "learning_rate": 2.7094290264932733e-06,
      "loss": 0.172,
      "step": 9254
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2446071473635107,
      "learning_rate": 2.7083993900370453e-06,
      "loss": 0.143,
      "step": 9255
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1977745423597743,
      "learning_rate": 2.7073698765848357e-06,
      "loss": 0.1374,
      "step": 9256
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3248679409289712,
      "learning_rate": 2.706340486191901e-06,
      "loss": 0.2029,
      "step": 9257
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3913776645603948,
      "learning_rate": 2.7053112189134988e-06,
      "loss": 0.187,
      "step": 9258
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4605588310662811,
      "learning_rate": 2.7042820748048724e-06,
      "loss": 0.1859,
      "step": 9259
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4480777026998808,
      "learning_rate": 2.703253053921266e-06,
      "loss": 0.1926,
      "step": 9260
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.224398025382799,
      "learning_rate": 2.7022241563179085e-06,
      "loss": 0.1946,
      "step": 9261
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4120986783646317,
      "learning_rate": 2.7011953820500305e-06,
      "loss": 0.2026,
      "step": 9262
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3533241482151315,
      "learning_rate": 2.7001667311728507e-06,
      "loss": 0.1797,
      "step": 9263
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.405525446702556,
      "learning_rate": 2.699138203741587e-06,
      "loss": 0.1892,
      "step": 9264
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4180960069873771,
      "learning_rate": 2.6981097998114416e-06,
      "loss": 0.1887,
      "step": 9265
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3051652773211828,
      "learning_rate": 2.6970815194376187e-06,
      "loss": 0.1638,
      "step": 9266
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3653319811806104,
      "learning_rate": 2.696053362675309e-06,
      "loss": 0.1749,
      "step": 9267
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5026885968292292,
      "learning_rate": 2.6950253295797033e-06,
      "loss": 0.1877,
      "step": 9268
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.512445884946505,
      "learning_rate": 2.6939974202059793e-06,
      "loss": 0.2127,
      "step": 9269
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3343622463117142,
      "learning_rate": 2.6929696346093128e-06,
      "loss": 0.1737,
      "step": 9270
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4869490897346702,
      "learning_rate": 2.6919419728448703e-06,
      "loss": 0.2147,
      "step": 9271
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6034803164105245,
      "learning_rate": 2.690914434967815e-06,
      "loss": 0.1741,
      "step": 9272
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.205290952666303,
      "learning_rate": 2.6898870210332982e-06,
      "loss": 0.1607,
      "step": 9273
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2927306173376851,
      "learning_rate": 2.68885973109647e-06,
      "loss": 0.1783,
      "step": 9274
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1995964389815783,
      "learning_rate": 2.6878325652124683e-06,
      "loss": 0.1788,
      "step": 9275
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1697118062243048,
      "learning_rate": 2.6868055234364304e-06,
      "loss": 0.1586,
      "step": 9276
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.817803226555485,
      "learning_rate": 2.6857786058234803e-06,
      "loss": 0.1664,
      "step": 9277
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4316280401644745,
      "learning_rate": 2.6847518124287403e-06,
      "loss": 0.1878,
      "step": 9278
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5599982382612807,
      "learning_rate": 2.6837251433073252e-06,
      "loss": 0.1649,
      "step": 9279
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.967772934417691,
      "learning_rate": 2.682698598514343e-06,
      "loss": 0.4092,
      "step": 9280
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3190403475864327,
      "learning_rate": 2.681672178104895e-06,
      "loss": 0.1636,
      "step": 9281
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3481183260083802,
      "learning_rate": 2.6806458821340726e-06,
      "loss": 0.172,
      "step": 9282
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4273857136394643,
      "learning_rate": 2.6796197106569667e-06,
      "loss": 0.1904,
      "step": 9283
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.355000729326456,
      "learning_rate": 2.6785936637286546e-06,
      "loss": 0.1675,
      "step": 9284
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.566463122407972,
      "learning_rate": 2.677567741404214e-06,
      "loss": 0.5335,
      "step": 9285
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6682991609050168,
      "learning_rate": 2.6765419437387088e-06,
      "loss": 0.2171,
      "step": 9286
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.332531661889876,
      "learning_rate": 2.675516270787201e-06,
      "loss": 0.168,
      "step": 9287
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.299136020006505,
      "learning_rate": 2.6744907226047455e-06,
      "loss": 0.1454,
      "step": 9288
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.225761706161005,
      "learning_rate": 2.673465299246391e-06,
      "loss": 0.1576,
      "step": 9289
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0842168558994822,
      "learning_rate": 2.6724400007671745e-06,
      "loss": 0.1326,
      "step": 9290
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2244833517813778,
      "learning_rate": 2.671414827222134e-06,
      "loss": 0.1614,
      "step": 9291
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4513025346771953,
      "learning_rate": 2.6703897786662925e-06,
      "loss": 0.1556,
      "step": 9292
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3402095259989195,
      "learning_rate": 2.669364855154675e-06,
      "loss": 0.1906,
      "step": 9293
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.310623714205806,
      "learning_rate": 2.66834005674229e-06,
      "loss": 0.1679,
      "step": 9294
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.337539931906442,
      "learning_rate": 2.6673153834841503e-06,
      "loss": 0.1953,
      "step": 9295
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3903304616172176,
      "learning_rate": 2.666290835435249e-06,
      "loss": 0.1688,
      "step": 9296
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2773407558574614,
      "learning_rate": 2.665266412650589e-06,
      "loss": 0.1838,
      "step": 9297
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3843137570338333,
      "learning_rate": 2.6642421151851506e-06,
      "loss": 0.1789,
      "step": 9298
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.254203113204737,
      "learning_rate": 2.6632179430939175e-06,
      "loss": 0.1675,
      "step": 9299
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.341905399106123,
      "learning_rate": 2.6621938964318593e-06,
      "loss": 0.1515,
      "step": 9300
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2534036113592772,
      "learning_rate": 2.6611699752539477e-06,
      "loss": 0.1755,
      "step": 9301
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2606640330360908,
      "learning_rate": 2.6601461796151383e-06,
      "loss": 0.1517,
      "step": 9302
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3065428778443706,
      "learning_rate": 2.6591225095703876e-06,
      "loss": 0.1971,
      "step": 9303
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4171440044672396,
      "learning_rate": 2.6580989651746395e-06,
      "loss": 0.1725,
      "step": 9304
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2898160533609875,
      "learning_rate": 2.6570755464828346e-06,
      "loss": 0.1811,
      "step": 9305
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1624951769933016,
      "learning_rate": 2.6560522535499067e-06,
      "loss": 0.1482,
      "step": 9306
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.467477688349392,
      "learning_rate": 2.655029086430784e-06,
      "loss": 0.203,
      "step": 9307
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4449126070963914,
      "learning_rate": 2.6540060451803817e-06,
      "loss": 0.2012,
      "step": 9308
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.314717863675286,
      "learning_rate": 2.6529831298536165e-06,
      "loss": 0.1709,
      "step": 9309
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.378840632807368,
      "learning_rate": 2.6519603405053906e-06,
      "loss": 0.2039,
      "step": 9310
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2368441788263926,
      "learning_rate": 2.650937677190606e-06,
      "loss": 0.1497,
      "step": 9311
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2516041645441565,
      "learning_rate": 2.6499151399641554e-06,
      "loss": 0.1595,
      "step": 9312
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1908802864332015,
      "learning_rate": 2.6488927288809228e-06,
      "loss": 0.1716,
      "step": 9313
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.190018255022369,
      "learning_rate": 2.647870443995788e-06,
      "loss": 0.1604,
      "step": 9314
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.7783849537471483,
      "learning_rate": 2.6468482853636223e-06,
      "loss": 0.4645,
      "step": 9315
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.388480098679166,
      "learning_rate": 2.645826253039295e-06,
      "loss": 0.1917,
      "step": 9316
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.290163946696938,
      "learning_rate": 2.6448043470776596e-06,
      "loss": 0.1578,
      "step": 9317
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3711633249526303,
      "learning_rate": 2.6437825675335714e-06,
      "loss": 0.1716,
      "step": 9318
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.130156475775542,
      "learning_rate": 2.6427609144618737e-06,
      "loss": 0.6032,
      "step": 9319
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.352036236926486,
      "learning_rate": 2.6417393879174056e-06,
      "loss": 0.1949,
      "step": 9320
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2439730595921765,
      "learning_rate": 2.6407179879549976e-06,
      "loss": 0.2039,
      "step": 9321
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.847093245452757,
      "learning_rate": 2.639696714629475e-06,
      "loss": 0.3558,
      "step": 9322
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5186189482682566,
      "learning_rate": 2.6386755679956557e-06,
      "loss": 0.1724,
      "step": 9323
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4583920556492413,
      "learning_rate": 2.637654548108352e-06,
      "loss": 0.1894,
      "step": 9324
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4507341880288265,
      "learning_rate": 2.636633655022366e-06,
      "loss": 0.2165,
      "step": 9325
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.818422412431467,
      "learning_rate": 2.6356128887924975e-06,
      "loss": 0.5487,
      "step": 9326
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.152994506410259,
      "learning_rate": 2.6345922494735345e-06,
      "loss": 0.5725,
      "step": 9327
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3785943050660376,
      "learning_rate": 2.6335717371202642e-06,
      "loss": 0.1931,
      "step": 9328
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2858839415292656,
      "learning_rate": 2.632551351787459e-06,
      "loss": 0.1242,
      "step": 9329
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3391311034556699,
      "learning_rate": 2.6315310935298925e-06,
      "loss": 0.1697,
      "step": 9330
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5412038893174917,
      "learning_rate": 2.6305109624023273e-06,
      "loss": 0.2047,
      "step": 9331
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1845457482871438,
      "learning_rate": 2.629490958459523e-06,
      "loss": 0.1446,
      "step": 9332
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.236969026655756,
      "learning_rate": 2.6284710817562233e-06,
      "loss": 0.1569,
      "step": 9333
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5001832513724567,
      "learning_rate": 2.627451332347177e-06,
      "loss": 0.2038,
      "step": 9334
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2993687079211058,
      "learning_rate": 2.626431710287115e-06,
      "loss": 0.1858,
      "step": 9335
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3976779735873606,
      "learning_rate": 2.6254122156307717e-06,
      "loss": 0.2116,
      "step": 9336
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.825096375892765,
      "learning_rate": 2.6243928484328653e-06,
      "loss": 0.4562,
      "step": 9337
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2199147986963674,
      "learning_rate": 2.623373608748112e-06,
      "loss": 0.1721,
      "step": 9338
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3550745283656216,
      "learning_rate": 2.622354496631222e-06,
      "loss": 0.175,
      "step": 9339
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2927934456195045,
      "learning_rate": 2.621335512136899e-06,
      "loss": 0.1991,
      "step": 9340
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3467794721091668,
      "learning_rate": 2.6203166553198334e-06,
      "loss": 0.1871,
      "step": 9341
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3932350941373284,
      "learning_rate": 2.6192979262347174e-06,
      "loss": 0.1741,
      "step": 9342
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2323234385641546,
      "learning_rate": 2.6182793249362293e-06,
      "loss": 0.1971,
      "step": 9343
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4026438291440717,
      "learning_rate": 2.617260851479045e-06,
      "loss": 0.211,
      "step": 9344
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2435716531540053,
      "learning_rate": 2.616242505917834e-06,
      "loss": 0.1501,
      "step": 9345
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1901959717746668,
      "learning_rate": 2.615224288307253e-06,
      "loss": 0.1505,
      "step": 9346
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.366117603031097,
      "learning_rate": 2.614206198701958e-06,
      "loss": 0.1989,
      "step": 9347
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.627917647099022,
      "learning_rate": 2.613188237156596e-06,
      "loss": 0.5708,
      "step": 9348
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1522649440190307,
      "learning_rate": 2.612170403725809e-06,
      "loss": 0.1571,
      "step": 9349
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.402475462827547,
      "learning_rate": 2.6111526984642264e-06,
      "loss": 0.1915,
      "step": 9350
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1711313286606628,
      "learning_rate": 2.6101351214264786e-06,
      "loss": 0.1524,
      "step": 9351
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3257106917788217,
      "learning_rate": 2.6091176726671815e-06,
      "loss": 0.1779,
      "step": 9352
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2419111184467833,
      "learning_rate": 2.608100352240951e-06,
      "loss": 0.1603,
      "step": 9353
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.430704688108737,
      "learning_rate": 2.6070831602023882e-06,
      "loss": 0.2167,
      "step": 9354
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2324689143075884,
      "learning_rate": 2.606066096606096e-06,
      "loss": 0.1482,
      "step": 9355
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.313772675883521,
      "learning_rate": 2.6050491615066646e-06,
      "loss": 0.5332,
      "step": 9356
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4623855661703495,
      "learning_rate": 2.6040323549586814e-06,
      "loss": 0.154,
      "step": 9357
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.321971314757792,
      "learning_rate": 2.6030156770167216e-06,
      "loss": 0.1889,
      "step": 9358
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4218177859606302,
      "learning_rate": 2.6019991277353586e-06,
      "loss": 0.1759,
      "step": 9359
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2714487148094773,
      "learning_rate": 2.600982707169154e-06,
      "loss": 0.158,
      "step": 9360
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.194443203620762,
      "learning_rate": 2.599966415372669e-06,
      "loss": 0.155,
      "step": 9361
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4351956499202008,
      "learning_rate": 2.5989502524004496e-06,
      "loss": 0.1839,
      "step": 9362
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3973737761653315,
      "learning_rate": 2.5979342183070422e-06,
      "loss": 0.1772,
      "step": 9363
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.326328677963942,
      "learning_rate": 2.5969183131469834e-06,
      "loss": 0.2161,
      "step": 9364
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5916166484798997,
      "learning_rate": 2.5959025369748035e-06,
      "loss": 0.2114,
      "step": 9365
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4282663435773646,
      "learning_rate": 2.594886889845023e-06,
      "loss": 0.1848,
      "step": 9366
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2776917314592318,
      "learning_rate": 2.5938713718121617e-06,
      "loss": 0.1712,
      "step": 9367
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4994098307899435,
      "learning_rate": 2.5928559829307236e-06,
      "loss": 0.2216,
      "step": 9368
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2203642787089797,
      "learning_rate": 2.5918407232552156e-06,
      "loss": 0.1763,
      "step": 9369
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6569940859029384,
      "learning_rate": 2.5908255928401283e-06,
      "loss": 0.1887,
      "step": 9370
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.531836052188907,
      "learning_rate": 2.589810591739952e-06,
      "loss": 0.2132,
      "step": 9371
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2740067404076434,
      "learning_rate": 2.588795720009168e-06,
      "loss": 0.203,
      "step": 9372
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2295950964849394,
      "learning_rate": 2.587780977702252e-06,
      "loss": 0.164,
      "step": 9373
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4588205683030269,
      "learning_rate": 2.5867663648736687e-06,
      "loss": 0.2045,
      "step": 9374
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.217350171890381,
      "learning_rate": 2.5857518815778795e-06,
      "loss": 0.607,
      "step": 9375
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2405239398352612,
      "learning_rate": 2.58473752786934e-06,
      "loss": 0.1541,
      "step": 9376
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.604736293080333,
      "learning_rate": 2.5837233038024924e-06,
      "loss": 0.2058,
      "step": 9377
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3667483378536727,
      "learning_rate": 2.5827092094317807e-06,
      "loss": 0.1783,
      "step": 9378
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3664999839484415,
      "learning_rate": 2.581695244811633e-06,
      "loss": 0.1571,
      "step": 9379
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1894738211791513,
      "learning_rate": 2.580681409996477e-06,
      "loss": 0.1799,
      "step": 9380
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.597504834857048,
      "learning_rate": 2.5796677050407327e-06,
      "loss": 0.4943,
      "step": 9381
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3622817014108628,
      "learning_rate": 2.578654129998811e-06,
      "loss": 0.1668,
      "step": 9382
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2763570083498477,
      "learning_rate": 2.5776406849251144e-06,
      "loss": 0.1925,
      "step": 9383
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3678536993405142,
      "learning_rate": 2.5766273698740454e-06,
      "loss": 0.1758,
      "step": 9384
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1801426475082042,
      "learning_rate": 2.5756141848999883e-06,
      "loss": 0.1577,
      "step": 9385
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3299629284589918,
      "learning_rate": 2.5746011300573327e-06,
      "loss": 0.1484,
      "step": 9386
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.383337167043854,
      "learning_rate": 2.5735882054004515e-06,
      "loss": 0.556,
      "step": 9387
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.192455243772222,
      "learning_rate": 2.572575410983715e-06,
      "loss": 0.1532,
      "step": 9388
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2823850473967455,
      "learning_rate": 2.571562746861487e-06,
      "loss": 0.1766,
      "step": 9389
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.333794744793949,
      "learning_rate": 2.570550213088125e-06,
      "loss": 0.1693,
      "step": 9390
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.345812001483703,
      "learning_rate": 2.5695378097179735e-06,
      "loss": 0.1972,
      "step": 9391
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3587442780064092,
      "learning_rate": 2.5685255368053796e-06,
      "loss": 0.1814,
      "step": 9392
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1889564598425788,
      "learning_rate": 2.5675133944046726e-06,
      "loss": 0.1495,
      "step": 9393
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.193778657595414,
      "learning_rate": 2.5665013825701845e-06,
      "loss": 0.569,
      "step": 9394
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4228530323214668,
      "learning_rate": 2.565489501356233e-06,
      "loss": 0.1919,
      "step": 9395
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2441754907306442,
      "learning_rate": 2.564477750817135e-06,
      "loss": 0.1645,
      "step": 9396
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1577420526330335,
      "learning_rate": 2.563466131007194e-06,
      "loss": 0.1331,
      "step": 9397
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7258741505990391,
      "learning_rate": 2.5624546419807105e-06,
      "loss": 0.1927,
      "step": 9398
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.303096822534713,
      "learning_rate": 2.5614432837919785e-06,
      "loss": 0.1867,
      "step": 9399
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4182391584554086,
      "learning_rate": 2.5604320564952846e-06,
      "loss": 0.2092,
      "step": 9400
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3879500152722326,
      "learning_rate": 2.559420960144905e-06,
      "loss": 0.1555,
      "step": 9401
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3574713096735207,
      "learning_rate": 2.5584099947951136e-06,
      "loss": 0.1849,
      "step": 9402
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.762614528303701,
      "learning_rate": 2.5573991605001715e-06,
      "loss": 0.6438,
      "step": 9403
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4038797438286432,
      "learning_rate": 2.556388457314341e-06,
      "loss": 0.1769,
      "step": 9404
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.912980337263262,
      "learning_rate": 2.5553778852918683e-06,
      "loss": 0.5079,
      "step": 9405
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3184240667136915,
      "learning_rate": 2.5543674444869983e-06,
      "loss": 0.1728,
      "step": 9406
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2286490769659115,
      "learning_rate": 2.553357134953968e-06,
      "loss": 0.1362,
      "step": 9407
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2661342113241145,
      "learning_rate": 2.552346956747007e-06,
      "loss": 0.1402,
      "step": 9408
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3023213503530433,
      "learning_rate": 2.551336909920339e-06,
      "loss": 0.1787,
      "step": 9409
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2676728376641704,
      "learning_rate": 2.550326994528176e-06,
      "loss": 0.144,
      "step": 9410
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2470361119800326,
      "learning_rate": 2.54931721062473e-06,
      "loss": 0.2013,
      "step": 9411
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3686661924701848,
      "learning_rate": 2.548307558264197e-06,
      "loss": 0.1922,
      "step": 9412
    },
    {
      "epoch": 0.67,
      "grad_norm": 8.895753112918506,
      "learning_rate": 2.547298037500777e-06,
      "loss": 0.5998,
      "step": 9413
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3691302868122082,
      "learning_rate": 2.546288648388653e-06,
      "loss": 0.1922,
      "step": 9414
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.327826954883381,
      "learning_rate": 2.545279390982005e-06,
      "loss": 0.2045,
      "step": 9415
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1196954692619139,
      "learning_rate": 2.5442702653350083e-06,
      "loss": 0.1256,
      "step": 9416
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3256193739894035,
      "learning_rate": 2.543261271501829e-06,
      "loss": 0.2431,
      "step": 9417
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.251508021641177,
      "learning_rate": 2.5422524095366223e-06,
      "loss": 0.1342,
      "step": 9418
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.636520434693917,
      "learning_rate": 2.5412436794935446e-06,
      "loss": 0.1899,
      "step": 9419
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2532637964310225,
      "learning_rate": 2.5402350814267364e-06,
      "loss": 0.1432,
      "step": 9420
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.160802048867107,
      "learning_rate": 2.5392266153903376e-06,
      "loss": 0.1725,
      "step": 9421
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.222153408678953,
      "learning_rate": 2.538218281438477e-06,
      "loss": 0.1584,
      "step": 9422
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.332858945629204,
      "learning_rate": 2.537210079625279e-06,
      "loss": 0.1517,
      "step": 9423
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.205923606162704,
      "learning_rate": 2.53620201000486e-06,
      "loss": 0.1791,
      "step": 9424
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.285492880336232,
      "learning_rate": 2.53519407263133e-06,
      "loss": 0.2022,
      "step": 9425
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.441178238393899,
      "learning_rate": 2.5341862675587886e-06,
      "loss": 0.7059,
      "step": 9426
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.184285262087191,
      "learning_rate": 2.5331785948413348e-06,
      "loss": 0.1601,
      "step": 9427
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.009035267654339,
      "learning_rate": 2.5321710545330504e-06,
      "loss": 0.6278,
      "step": 9428
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2901917690526674,
      "learning_rate": 2.531163646688023e-06,
      "loss": 0.1895,
      "step": 9429
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4617353265688011,
      "learning_rate": 2.5301563713603195e-06,
      "loss": 0.2099,
      "step": 9430
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2331575803259158,
      "learning_rate": 2.529149228604011e-06,
      "loss": 0.1689,
      "step": 9431
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1666139802057132,
      "learning_rate": 2.528142218473155e-06,
      "loss": 0.1801,
      "step": 9432
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3805029973256087,
      "learning_rate": 2.527135341021807e-06,
      "loss": 0.1832,
      "step": 9433
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.31696521585725,
      "learning_rate": 2.5261285963040076e-06,
      "loss": 0.2019,
      "step": 9434
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.96895352628261,
      "learning_rate": 2.5251219843737993e-06,
      "loss": 0.5936,
      "step": 9435
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1252506865319964,
      "learning_rate": 2.5241155052852086e-06,
      "loss": 0.1388,
      "step": 9436
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3938873326305772,
      "learning_rate": 2.523109159092263e-06,
      "loss": 0.2086,
      "step": 9437
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2916593597002277,
      "learning_rate": 2.5221029458489756e-06,
      "loss": 0.1718,
      "step": 9438
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2036627619734719,
      "learning_rate": 2.5210968656093583e-06,
      "loss": 0.1698,
      "step": 9439
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4072654144888759,
      "learning_rate": 2.5200909184274125e-06,
      "loss": 0.1922,
      "step": 9440
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.573798950341566,
      "learning_rate": 2.5190851043571344e-06,
      "loss": 0.6186,
      "step": 9441
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.424787531461755,
      "learning_rate": 2.5180794234525143e-06,
      "loss": 0.1912,
      "step": 9442
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1986551870897442,
      "learning_rate": 2.5170738757675277e-06,
      "loss": 0.1593,
      "step": 9443
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5058013828757564,
      "learning_rate": 2.516068461356154e-06,
      "loss": 0.1886,
      "step": 9444
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.255133515063899,
      "learning_rate": 2.5150631802723547e-06,
      "loss": 0.1374,
      "step": 9445
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4423207873489587,
      "learning_rate": 2.5140580325700934e-06,
      "loss": 0.1939,
      "step": 9446
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2951851515705977,
      "learning_rate": 2.5130530183033197e-06,
      "loss": 0.1673,
      "step": 9447
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.494209470562147,
      "learning_rate": 2.51204813752598e-06,
      "loss": 0.1891,
      "step": 9448
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2655596634464201,
      "learning_rate": 2.5110433902920106e-06,
      "loss": 0.1789,
      "step": 9449
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.388063477107571,
      "learning_rate": 2.510038776655347e-06,
      "loss": 0.1962,
      "step": 9450
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5563532591953824,
      "learning_rate": 2.5090342966699077e-06,
      "loss": 0.1764,
      "step": 9451
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.220676935674887,
      "learning_rate": 2.508029950389613e-06,
      "loss": 0.5474,
      "step": 9452
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.329741177054563,
      "learning_rate": 2.5070257378683692e-06,
      "loss": 0.1586,
      "step": 9453
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2589800126763218,
      "learning_rate": 2.5060216591600817e-06,
      "loss": 0.1458,
      "step": 9454
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3146463159371862,
      "learning_rate": 2.5050177143186405e-06,
      "loss": 0.1659,
      "step": 9455
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3628922224618658,
      "learning_rate": 2.504013903397937e-06,
      "loss": 0.1938,
      "step": 9456
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3852549875067506,
      "learning_rate": 2.5030102264518515e-06,
      "loss": 0.1728,
      "step": 9457
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3744170065686803,
      "learning_rate": 2.502006683534258e-06,
      "loss": 0.1984,
      "step": 9458
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3613976128515801,
      "learning_rate": 2.501003274699021e-06,
      "loss": 0.174,
      "step": 9459
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.455684614279227,
      "learning_rate": 2.5000000000000015e-06,
      "loss": 0.1973,
      "step": 9460
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4092578066508759,
      "learning_rate": 2.498996859491048e-06,
      "loss": 0.1685,
      "step": 9461
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2687481141064756,
      "learning_rate": 2.497993853226009e-06,
      "loss": 0.1816,
      "step": 9462
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.360109787545344,
      "learning_rate": 2.4969909812587184e-06,
      "loss": 0.1436,
      "step": 9463
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2735829582039413,
      "learning_rate": 2.495988243643008e-06,
      "loss": 0.1448,
      "step": 9464
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3937288717018805,
      "learning_rate": 2.4949856404327007e-06,
      "loss": 0.1952,
      "step": 9465
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3140238149532244,
      "learning_rate": 2.4939831716816136e-06,
      "loss": 0.1959,
      "step": 9466
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.444659848827377,
      "learning_rate": 2.4929808374435526e-06,
      "loss": 0.2006,
      "step": 9467
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3392593441187322,
      "learning_rate": 2.4919786377723225e-06,
      "loss": 0.1875,
      "step": 9468
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3340744555736854,
      "learning_rate": 2.4909765727217132e-06,
      "loss": 0.1585,
      "step": 9469
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2504621477337763,
      "learning_rate": 2.4899746423455158e-06,
      "loss": 0.156,
      "step": 9470
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4548149456950323,
      "learning_rate": 2.4889728466975056e-06,
      "loss": 0.1317,
      "step": 9471
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2499324520767812,
      "learning_rate": 2.487971185831457e-06,
      "loss": 0.1518,
      "step": 9472
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2749843872325894,
      "learning_rate": 2.4869696598011354e-06,
      "loss": 0.1869,
      "step": 9473
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3900200163774987,
      "learning_rate": 2.485968268660299e-06,
      "loss": 0.1517,
      "step": 9474
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3032265095461188,
      "learning_rate": 2.4849670124627e-06,
      "loss": 0.1726,
      "step": 9475
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.294659071149979,
      "learning_rate": 2.4839658912620782e-06,
      "loss": 0.2013,
      "step": 9476
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3896783519645675,
      "learning_rate": 2.482964905112174e-06,
      "loss": 0.1868,
      "step": 9477
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4051772761160892,
      "learning_rate": 2.481964054066712e-06,
      "loss": 0.1577,
      "step": 9478
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5351161988158686,
      "learning_rate": 2.4809633381794175e-06,
      "loss": 0.1997,
      "step": 9479
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.421929577193885,
      "learning_rate": 2.4799627575040014e-06,
      "loss": 0.1847,
      "step": 9480
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.632105163448895,
      "learning_rate": 2.478962312094173e-06,
      "loss": 0.577,
      "step": 9481
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3233472911457882,
      "learning_rate": 2.4779620020036328e-06,
      "loss": 0.212,
      "step": 9482
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4495694547483886,
      "learning_rate": 2.4769618272860742e-06,
      "loss": 0.1898,
      "step": 9483
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.251711329981804,
      "learning_rate": 2.475961787995179e-06,
      "loss": 0.1467,
      "step": 9484
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5035248994312191,
      "learning_rate": 2.4749618841846297e-06,
      "loss": 0.1829,
      "step": 9485
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.219082891583727,
      "learning_rate": 2.473962115908093e-06,
      "loss": 0.1504,
      "step": 9486
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2129967224136002,
      "learning_rate": 2.4729624832192365e-06,
      "loss": 0.1311,
      "step": 9487
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5828077011642536,
      "learning_rate": 2.4719629861717126e-06,
      "loss": 0.2106,
      "step": 9488
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3868597614783982,
      "learning_rate": 2.4709636248191727e-06,
      "loss": 0.2075,
      "step": 9489
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4235150300827484,
      "learning_rate": 2.469964399215258e-06,
      "loss": 0.1897,
      "step": 9490
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3344444004687213,
      "learning_rate": 2.468965309413604e-06,
      "loss": 0.1742,
      "step": 9491
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3688543567069344,
      "learning_rate": 2.4679663554678357e-06,
      "loss": 0.1759,
      "step": 9492
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3834593693099588,
      "learning_rate": 2.466967537431576e-06,
      "loss": 0.1664,
      "step": 9493
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.336560301208897,
      "learning_rate": 2.4659688553584342e-06,
      "loss": 0.6521,
      "step": 9494
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2037183356606593,
      "learning_rate": 2.464970309302019e-06,
      "loss": 0.1566,
      "step": 9495
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.220859189978877,
      "learning_rate": 2.463971899315924e-06,
      "loss": 0.1756,
      "step": 9496
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2568516985268094,
      "learning_rate": 2.462973625453744e-06,
      "loss": 0.1647,
      "step": 9497
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.688822089006416,
      "learning_rate": 2.4619754877690593e-06,
      "loss": 0.7172,
      "step": 9498
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.394205855984081,
      "learning_rate": 2.4609774863154474e-06,
      "loss": 0.2194,
      "step": 9499
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3164653331753404,
      "learning_rate": 2.4599796211464772e-06,
      "loss": 0.1605,
      "step": 9500
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4060312633532654,
      "learning_rate": 2.4589818923157117e-06,
      "loss": 0.1648,
      "step": 9501
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2162788429382967,
      "learning_rate": 2.457984299876701e-06,
      "loss": 0.1402,
      "step": 9502
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3076689569243973,
      "learning_rate": 2.4569868438829948e-06,
      "loss": 0.1575,
      "step": 9503
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2764283887880365,
      "learning_rate": 2.4559895243881333e-06,
      "loss": 0.1542,
      "step": 9504
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3877067994408514,
      "learning_rate": 2.4549923414456455e-06,
      "loss": 0.1633,
      "step": 9505
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3024964980712612,
      "learning_rate": 2.4539952951090607e-06,
      "loss": 0.2202,
      "step": 9506
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4326962381366886,
      "learning_rate": 2.452998385431891e-06,
      "loss": 0.1804,
      "step": 9507
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.445550988927256,
      "learning_rate": 2.4520016124676493e-06,
      "loss": 0.2032,
      "step": 9508
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3695328356604932,
      "learning_rate": 2.451004976269839e-06,
      "loss": 0.2041,
      "step": 9509
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2906646431967768,
      "learning_rate": 2.4500084768919563e-06,
      "loss": 0.1888,
      "step": 9510
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3497308458581203,
      "learning_rate": 2.449012114387486e-06,
      "loss": 0.19,
      "step": 9511
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1930495781368502,
      "learning_rate": 2.448015888809913e-06,
      "loss": 0.1641,
      "step": 9512
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.33863872982848,
      "learning_rate": 2.447019800212707e-06,
      "loss": 0.1662,
      "step": 9513
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4377090155090906,
      "learning_rate": 2.4460238486493372e-06,
      "loss": 0.1769,
      "step": 9514
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6028935901016272,
      "learning_rate": 2.445028034173259e-06,
      "loss": 0.1802,
      "step": 9515
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.7922137598579253,
      "learning_rate": 2.4440323568379255e-06,
      "loss": 0.4954,
      "step": 9516
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.182783380660088,
      "learning_rate": 2.4430368166967806e-06,
      "loss": 0.1387,
      "step": 9517
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4545432327652243,
      "learning_rate": 2.4420414138032632e-06,
      "loss": 0.194,
      "step": 9518
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.258249588613014,
      "learning_rate": 2.441046148210799e-06,
      "loss": 0.1686,
      "step": 9519
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.75849137008114,
      "learning_rate": 2.4400510199728123e-06,
      "loss": 0.5279,
      "step": 9520
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.258812033061077,
      "learning_rate": 2.439056029142716e-06,
      "loss": 0.1658,
      "step": 9521
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4257034995983668,
      "learning_rate": 2.4380611757739186e-06,
      "loss": 0.169,
      "step": 9522
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5325296600330616,
      "learning_rate": 2.437066459919818e-06,
      "loss": 0.1687,
      "step": 9523
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.292259345136729,
      "learning_rate": 2.436071881633808e-06,
      "loss": 0.4338,
      "step": 9524
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1769596361936145,
      "learning_rate": 2.435077440969273e-06,
      "loss": 0.1798,
      "step": 9525
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.489933317507412,
      "learning_rate": 2.434083137979593e-06,
      "loss": 0.2304,
      "step": 9526
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3552780428631421,
      "learning_rate": 2.433088972718135e-06,
      "loss": 0.1513,
      "step": 9527
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.283262168029897,
      "learning_rate": 2.4320949452382646e-06,
      "loss": 0.5374,
      "step": 9528
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4900735313325197,
      "learning_rate": 2.431101055593334e-06,
      "loss": 0.2067,
      "step": 9529
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.334086340807652,
      "learning_rate": 2.430107303836695e-06,
      "loss": 0.2008,
      "step": 9530
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2883223805723385,
      "learning_rate": 2.4291136900216845e-06,
      "loss": 0.1515,
      "step": 9531
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5280252776473957,
      "learning_rate": 2.428120214201637e-06,
      "loss": 0.2167,
      "step": 9532
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2621615766897027,
      "learning_rate": 2.4271268764298795e-06,
      "loss": 0.1889,
      "step": 9533
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.224430695059103,
      "learning_rate": 2.426133676759731e-06,
      "loss": 0.1298,
      "step": 9534
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2327989639630421,
      "learning_rate": 2.4251406152445002e-06,
      "loss": 0.1662,
      "step": 9535
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.274958052224102,
      "learning_rate": 2.424147691937491e-06,
      "loss": 0.1647,
      "step": 9536
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4628545205932186,
      "learning_rate": 2.423154906892003e-06,
      "loss": 0.1916,
      "step": 9537
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2578554589432847,
      "learning_rate": 2.4221622601613197e-06,
      "loss": 0.1792,
      "step": 9538
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4167442112575475,
      "learning_rate": 2.4211697517987278e-06,
      "loss": 0.177,
      "step": 9539
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.393326625321518,
      "learning_rate": 2.4201773818574956e-06,
      "loss": 0.1719,
      "step": 9540
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.231525474208533,
      "learning_rate": 2.419185150390893e-06,
      "loss": 0.1508,
      "step": 9541
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.360380482853282,
      "learning_rate": 2.418193057452179e-06,
      "loss": 0.1716,
      "step": 9542
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2789153732114065,
      "learning_rate": 2.4172011030946057e-06,
      "loss": 0.1634,
      "step": 9543
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.481508343520224,
      "learning_rate": 2.416209287371415e-06,
      "loss": 0.5845,
      "step": 9544
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4085161678270786,
      "learning_rate": 2.4152176103358468e-06,
      "loss": 0.1948,
      "step": 9545
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2984463025381572,
      "learning_rate": 2.414226072041126e-06,
      "loss": 0.2037,
      "step": 9546
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.259660295186739,
      "learning_rate": 2.4132346725404795e-06,
      "loss": 0.1794,
      "step": 9547
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2880896301145979,
      "learning_rate": 2.4122434118871162e-06,
      "loss": 0.1659,
      "step": 9548
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2507517360154965,
      "learning_rate": 2.4112522901342467e-06,
      "loss": 0.1793,
      "step": 9549
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2664580496144495,
      "learning_rate": 2.41026130733507e-06,
      "loss": 0.1667,
      "step": 9550
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.18301868968222,
      "learning_rate": 2.409270463542779e-06,
      "loss": 0.1518,
      "step": 9551
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.574399271597619,
      "learning_rate": 2.4082797588105554e-06,
      "loss": 0.1907,
      "step": 9552
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.323093302716809,
      "learning_rate": 2.4072891931915805e-06,
      "loss": 0.1622,
      "step": 9553
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.371821637289227,
      "learning_rate": 2.4062987667390186e-06,
      "loss": 0.2031,
      "step": 9554
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.349611035479337,
      "learning_rate": 2.405308479506037e-06,
      "loss": 0.1971,
      "step": 9555
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4158294566891614,
      "learning_rate": 2.4043183315457862e-06,
      "loss": 0.1797,
      "step": 9556
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2103979872541717,
      "learning_rate": 2.403328322911416e-06,
      "loss": 0.1622,
      "step": 9557
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.295627483288364,
      "learning_rate": 2.402338453656064e-06,
      "loss": 0.1909,
      "step": 9558
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5097208732141465,
      "learning_rate": 2.4013487238328663e-06,
      "loss": 0.1812,
      "step": 9559
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.358160288033,
      "learning_rate": 2.400359133494944e-06,
      "loss": 0.1854,
      "step": 9560
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4850782248528784,
      "learning_rate": 2.3993696826954168e-06,
      "loss": 0.1809,
      "step": 9561
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.451460854456984,
      "learning_rate": 2.398380371487391e-06,
      "loss": 0.1809,
      "step": 9562
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3840279145599783,
      "learning_rate": 2.3973911999239737e-06,
      "loss": 0.1737,
      "step": 9563
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.381715306101897,
      "learning_rate": 2.396402168058255e-06,
      "loss": 0.1703,
      "step": 9564
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.382419570520723,
      "learning_rate": 2.3954132759433244e-06,
      "loss": 0.2302,
      "step": 9565
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.459974944879367,
      "learning_rate": 2.3944245236322623e-06,
      "loss": 0.5708,
      "step": 9566
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4032251710027135,
      "learning_rate": 2.3934359111781397e-06,
      "loss": 0.2099,
      "step": 9567
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2887263859508393,
      "learning_rate": 2.392447438634024e-06,
      "loss": 0.1829,
      "step": 9568
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2733876959471513,
      "learning_rate": 2.3914591060529683e-06,
      "loss": 0.1804,
      "step": 9569
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4487286894472564,
      "learning_rate": 2.390470913488027e-06,
      "loss": 0.1932,
      "step": 9570
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.061827390708526,
      "learning_rate": 2.3894828609922384e-06,
      "loss": 0.7468,
      "step": 9571
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.56720031848221,
      "learning_rate": 2.38849494861864e-06,
      "loss": 0.5525,
      "step": 9572
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2348093869460817,
      "learning_rate": 2.387507176420256e-06,
      "loss": 0.1843,
      "step": 9573
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.340631015273686,
      "learning_rate": 2.3865195444501083e-06,
      "loss": 0.2131,
      "step": 9574
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.347413371841547,
      "learning_rate": 2.3855320527612087e-06,
      "loss": 0.2193,
      "step": 9575
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5813625051119005,
      "learning_rate": 2.384544701406563e-06,
      "loss": 0.2122,
      "step": 9576
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3994996041294225,
      "learning_rate": 2.3835574904391655e-06,
      "loss": 0.1887,
      "step": 9577
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2557062125241178,
      "learning_rate": 2.382570419912009e-06,
      "loss": 0.1153,
      "step": 9578
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1901274711006653,
      "learning_rate": 2.381583489878073e-06,
      "loss": 0.1597,
      "step": 9579
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2554220870604667,
      "learning_rate": 2.3805967003903336e-06,
      "loss": 0.1592,
      "step": 9580
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.242064754122138,
      "learning_rate": 2.3796100515017554e-06,
      "loss": 0.7462,
      "step": 9581
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4938901903656598,
      "learning_rate": 2.3786235432653e-06,
      "loss": 0.2004,
      "step": 9582
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2941732877084384,
      "learning_rate": 2.377637175733918e-06,
      "loss": 0.1683,
      "step": 9583
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2226125232845264,
      "learning_rate": 2.376650948960556e-06,
      "loss": 0.1641,
      "step": 9584
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1464644357026295,
      "learning_rate": 2.3756648629981476e-06,
      "loss": 0.151,
      "step": 9585
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2913757222535163,
      "learning_rate": 2.3746789178996248e-06,
      "loss": 0.1614,
      "step": 9586
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.310951053771698,
      "learning_rate": 2.3736931137179063e-06,
      "loss": 0.5001,
      "step": 9587
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3945321557538508,
      "learning_rate": 2.37270745050591e-06,
      "loss": 0.1654,
      "step": 9588
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4449032529699466,
      "learning_rate": 2.3717219283165375e-06,
      "loss": 0.219,
      "step": 9589
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.315662133987693,
      "learning_rate": 2.3707365472026926e-06,
      "loss": 0.1736,
      "step": 9590
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2897628579930753,
      "learning_rate": 2.36975130721726e-06,
      "loss": 0.1695,
      "step": 9591
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5824439239115817,
      "learning_rate": 2.3687662084131326e-06,
      "loss": 0.146,
      "step": 9592
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2913828139769141,
      "learning_rate": 2.367781250843179e-06,
      "loss": 0.1701,
      "step": 9593
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.276603265066647,
      "learning_rate": 2.366796434560273e-06,
      "loss": 0.135,
      "step": 9594
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.486992602464422,
      "learning_rate": 2.365811759617271e-06,
      "loss": 0.1921,
      "step": 9595
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4428170710564467,
      "learning_rate": 2.364827226067031e-06,
      "loss": 0.2036,
      "step": 9596
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3725179037986874,
      "learning_rate": 2.363842833962394e-06,
      "loss": 0.1904,
      "step": 9597
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1572152585931663,
      "learning_rate": 2.362858583356203e-06,
      "loss": 0.1424,
      "step": 9598
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4370164608425777,
      "learning_rate": 2.3618744743012843e-06,
      "loss": 0.1811,
      "step": 9599
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6166589740705166,
      "learning_rate": 2.360890506850464e-06,
      "loss": 0.1698,
      "step": 9600
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4120658181921466,
      "learning_rate": 2.3599066810565556e-06,
      "loss": 0.2117,
      "step": 9601
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.166276234696767,
      "learning_rate": 2.358922996972369e-06,
      "loss": 0.1582,
      "step": 9602
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3990501716524772,
      "learning_rate": 2.3579394546507043e-06,
      "loss": 0.1719,
      "step": 9603
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2322491575843164,
      "learning_rate": 2.356956054144352e-06,
      "loss": 0.1554,
      "step": 9604
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4300032138994128,
      "learning_rate": 2.3559727955061e-06,
      "loss": 0.2179,
      "step": 9605
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.271275713294997,
      "learning_rate": 2.3549896787887227e-06,
      "loss": 0.2032,
      "step": 9606
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.3170387518029685,
      "learning_rate": 2.354006704044993e-06,
      "loss": 0.4816,
      "step": 9607
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.19384141499077,
      "learning_rate": 2.3530238713276697e-06,
      "loss": 0.5959,
      "step": 9608
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4020350853909755,
      "learning_rate": 2.3520411806895094e-06,
      "loss": 0.2032,
      "step": 9609
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.408594129093204,
      "learning_rate": 2.3510586321832586e-06,
      "loss": 0.185,
      "step": 9610
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.319491006252254,
      "learning_rate": 2.3500762258616585e-06,
      "loss": 0.1617,
      "step": 9611
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4119304945685256,
      "learning_rate": 2.349093961777437e-06,
      "loss": 0.1668,
      "step": 9612
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3828278393312037,
      "learning_rate": 2.348111839983322e-06,
      "loss": 0.1974,
      "step": 9613
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3350745849245143,
      "learning_rate": 2.3471298605320263e-06,
      "loss": 0.1834,
      "step": 9614
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3318689998534465,
      "learning_rate": 2.346148023476262e-06,
      "loss": 0.1735,
      "step": 9615
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3889895134448862,
      "learning_rate": 2.345166328868727e-06,
      "loss": 0.1885,
      "step": 9616
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0415049470454416,
      "learning_rate": 2.344184776762117e-06,
      "loss": 0.1226,
      "step": 9617
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.614843922460169,
      "learning_rate": 2.3432033672091167e-06,
      "loss": 0.483,
      "step": 9618
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.361517812561361,
      "learning_rate": 2.342222100262406e-06,
      "loss": 0.1679,
      "step": 9619
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3337294422224266,
      "learning_rate": 2.341240975974653e-06,
      "loss": 0.1421,
      "step": 9620
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2618537815658224,
      "learning_rate": 2.3402599943985233e-06,
      "loss": 0.1687,
      "step": 9621
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3071918743921869,
      "learning_rate": 2.3392791555866682e-06,
      "loss": 0.1446,
      "step": 9622
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5100228688968884,
      "learning_rate": 2.3382984595917403e-06,
      "loss": 0.2021,
      "step": 9623
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2537668357837448,
      "learning_rate": 2.3373179064663742e-06,
      "loss": 0.1619,
      "step": 9624
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3725846921721856,
      "learning_rate": 2.3363374962632054e-06,
      "loss": 0.1994,
      "step": 9625
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2877026098755515,
      "learning_rate": 2.335357229034857e-06,
      "loss": 0.1697,
      "step": 9626
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7060020295720955,
      "learning_rate": 2.334377104833948e-06,
      "loss": 0.166,
      "step": 9627
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3044122683292672,
      "learning_rate": 2.3333971237130847e-06,
      "loss": 0.207,
      "step": 9628
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2975771909108422,
      "learning_rate": 2.3324172857248715e-06,
      "loss": 0.1692,
      "step": 9629
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3663624162254207,
      "learning_rate": 2.3314375909218997e-06,
      "loss": 0.1988,
      "step": 9630
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.106730218126169,
      "learning_rate": 2.3304580393567554e-06,
      "loss": 0.1743,
      "step": 9631
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.572069082973457,
      "learning_rate": 2.32947863108202e-06,
      "loss": 0.6817,
      "step": 9632
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4511487074092666,
      "learning_rate": 2.328499366150261e-06,
      "loss": 0.1923,
      "step": 9633
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.953096068918021,
      "learning_rate": 2.327520244614042e-06,
      "loss": 0.6517,
      "step": 9634
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3088291430614793,
      "learning_rate": 2.3265412665259194e-06,
      "loss": 0.1671,
      "step": 9635
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2518362639334584,
      "learning_rate": 2.325562431938442e-06,
      "loss": 0.1825,
      "step": 9636
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3953291939815788,
      "learning_rate": 2.324583740904147e-06,
      "loss": 0.188,
      "step": 9637
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3208537437516812,
      "learning_rate": 2.32360519347557e-06,
      "loss": 0.1747,
      "step": 9638
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4094861164622756,
      "learning_rate": 2.3226267897052313e-06,
      "loss": 0.1972,
      "step": 9639
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2997457753032087,
      "learning_rate": 2.3216485296456514e-06,
      "loss": 0.1778,
      "step": 9640
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1383595761387246,
      "learning_rate": 2.320670413349336e-06,
      "loss": 0.1519,
      "step": 9641
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.34202827430766,
      "learning_rate": 2.3196924408687887e-06,
      "loss": 0.1947,
      "step": 9642
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3386712008370174,
      "learning_rate": 2.318714612256503e-06,
      "loss": 0.1821,
      "step": 9643
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5602744842561822,
      "learning_rate": 2.3177369275649664e-06,
      "loss": 0.1659,
      "step": 9644
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4499245076413771,
      "learning_rate": 2.3167593868466537e-06,
      "loss": 0.1735,
      "step": 9645
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3265672274413594,
      "learning_rate": 2.315781990154039e-06,
      "loss": 0.1631,
      "step": 9646
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4210072248178438,
      "learning_rate": 2.3148047375395816e-06,
      "loss": 0.1911,
      "step": 9647
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3034695677690444,
      "learning_rate": 2.3138276290557394e-06,
      "loss": 0.1634,
      "step": 9648
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3722130315334324,
      "learning_rate": 2.312850664754957e-06,
      "loss": 0.2091,
      "step": 9649
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3035920861902845,
      "learning_rate": 2.3118738446896756e-06,
      "loss": 0.165,
      "step": 9650
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4044459720042444,
      "learning_rate": 2.3108971689123266e-06,
      "loss": 0.2282,
      "step": 9651
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.480192140490272,
      "learning_rate": 2.309920637475336e-06,
      "loss": 0.2281,
      "step": 9652
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2182601778361872,
      "learning_rate": 2.308944250431117e-06,
      "loss": 0.1736,
      "step": 9653
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4367435704557008,
      "learning_rate": 2.3079680078320814e-06,
      "loss": 0.1933,
      "step": 9654
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2192072422236415,
      "learning_rate": 2.3069919097306264e-06,
      "loss": 0.1446,
      "step": 9655
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.739784062486623,
      "learning_rate": 2.306015956179149e-06,
      "loss": 0.5005,
      "step": 9656
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3365742581605828,
      "learning_rate": 2.3050401472300304e-06,
      "loss": 0.1823,
      "step": 9657
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3917417583636542,
      "learning_rate": 2.30406448293565e-06,
      "loss": 0.1923,
      "step": 9658
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4665081824002415,
      "learning_rate": 2.303088963348378e-06,
      "loss": 0.1815,
      "step": 9659
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.934251264734299,
      "learning_rate": 2.302113588520578e-06,
      "loss": 0.4464,
      "step": 9660
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2269594425932628,
      "learning_rate": 2.3011383585046003e-06,
      "loss": 0.144,
      "step": 9661
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4242193818502173,
      "learning_rate": 2.3001632733527957e-06,
      "loss": 0.1796,
      "step": 9662
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2978960983933665,
      "learning_rate": 2.299188333117499e-06,
      "loss": 0.1625,
      "step": 9663
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3927477868875529,
      "learning_rate": 2.298213537851042e-06,
      "loss": 0.1784,
      "step": 9664
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1932552959588192,
      "learning_rate": 2.2972388876057517e-06,
      "loss": 0.155,
      "step": 9665
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3922745081488688,
      "learning_rate": 2.296264382433938e-06,
      "loss": 0.2045,
      "step": 9666
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.120035771211254,
      "learning_rate": 2.295290022387911e-06,
      "loss": 0.1283,
      "step": 9667
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4698697809323575,
      "learning_rate": 2.29431580751997e-06,
      "loss": 0.2104,
      "step": 9668
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4000771729505515,
      "learning_rate": 2.293341737882409e-06,
      "loss": 0.1723,
      "step": 9669
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3278408255016803,
      "learning_rate": 2.2923678135275096e-06,
      "loss": 0.1928,
      "step": 9670
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2506308987349293,
      "learning_rate": 2.29139403450755e-06,
      "loss": 0.1432,
      "step": 9671
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3173164957948924,
      "learning_rate": 2.2904204008747966e-06,
      "loss": 0.1832,
      "step": 9672
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.232586956895497,
      "learning_rate": 2.2894469126815134e-06,
      "loss": 0.1921,
      "step": 9673
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2640227449292543,
      "learning_rate": 2.2884735699799493e-06,
      "loss": 0.1395,
      "step": 9674
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.180722217334474,
      "learning_rate": 2.2875003728223515e-06,
      "loss": 0.1597,
      "step": 9675
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.433887040145558,
      "learning_rate": 2.286527321260958e-06,
      "loss": 0.1967,
      "step": 9676
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2583215096976526,
      "learning_rate": 2.2855544153479998e-06,
      "loss": 0.1685,
      "step": 9677
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.299870474557435,
      "learning_rate": 2.284581655135694e-06,
      "loss": 0.1812,
      "step": 9678
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4802137282370857,
      "learning_rate": 2.28360904067626e-06,
      "loss": 0.2006,
      "step": 9679
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3653653252914038,
      "learning_rate": 2.2826365720218984e-06,
      "loss": 0.2183,
      "step": 9680
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3538937561986935,
      "learning_rate": 2.2816642492248126e-06,
      "loss": 0.1857,
      "step": 9681
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1532971580799278,
      "learning_rate": 2.2806920723371882e-06,
      "loss": 0.1477,
      "step": 9682
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.367533812790115,
      "learning_rate": 2.2797200414112105e-06,
      "loss": 0.1751,
      "step": 9683
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.6011990748603315,
      "learning_rate": 2.2787481564990533e-06,
      "loss": 0.5811,
      "step": 9684
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.478295520358733,
      "learning_rate": 2.277776417652886e-06,
      "loss": 0.6629,
      "step": 9685
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4054731692061906,
      "learning_rate": 2.2768048249248648e-06,
      "loss": 0.2177,
      "step": 9686
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2453904173837793,
      "learning_rate": 2.2758333783671427e-06,
      "loss": 0.1397,
      "step": 9687
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.300086568585657,
      "learning_rate": 2.274862078031861e-06,
      "loss": 0.1844,
      "step": 9688
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.166774210259524,
      "learning_rate": 2.2738909239711586e-06,
      "loss": 0.1422,
      "step": 9689
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3262475475546804,
      "learning_rate": 2.272919916237159e-06,
      "loss": 0.1578,
      "step": 9690
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3283178285061432,
      "learning_rate": 2.2719490548819863e-06,
      "loss": 0.1521,
      "step": 9691
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2811957066415476,
      "learning_rate": 2.2709783399577463e-06,
      "loss": 0.1736,
      "step": 9692
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4105127662138184,
      "learning_rate": 2.270007771516551e-06,
      "loss": 0.2067,
      "step": 9693
    },
    {
      "epoch": 0.69,
      "grad_norm": 8.513005955007769,
      "learning_rate": 2.2690373496104906e-06,
      "loss": 0.5899,
      "step": 9694
    },
    {
      "epoch": 0.69,
      "grad_norm": 8.930753585509171,
      "learning_rate": 2.2680670742916565e-06,
      "loss": 0.5254,
      "step": 9695
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.332373171802525,
      "learning_rate": 2.26709694561213e-06,
      "loss": 0.1779,
      "step": 9696
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3162728915981785,
      "learning_rate": 2.2661269636239802e-06,
      "loss": 0.1543,
      "step": 9697
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3838313522337564,
      "learning_rate": 2.2651571283792758e-06,
      "loss": 0.1437,
      "step": 9698
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.333960544850729,
      "learning_rate": 2.2641874399300703e-06,
      "loss": 0.2292,
      "step": 9699
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2234946377382132,
      "learning_rate": 2.263217898328415e-06,
      "loss": 0.1697,
      "step": 9700
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5535356202117834,
      "learning_rate": 2.26224850362635e-06,
      "loss": 0.1778,
      "step": 9701
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4305158718920306,
      "learning_rate": 2.261279255875908e-06,
      "loss": 0.2066,
      "step": 9702
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.398051605771974,
      "learning_rate": 2.2603101551291144e-06,
      "loss": 0.1543,
      "step": 9703
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2080128727622723,
      "learning_rate": 2.2593412014379905e-06,
      "loss": 0.174,
      "step": 9704
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3103284854668777,
      "learning_rate": 2.2583723948545405e-06,
      "loss": 0.1655,
      "step": 9705
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3848654478525144,
      "learning_rate": 2.2574037354307703e-06,
      "loss": 0.1801,
      "step": 9706
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2605202889970932,
      "learning_rate": 2.2564352232186705e-06,
      "loss": 0.1783,
      "step": 9707
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2914575689928482,
      "learning_rate": 2.2554668582702294e-06,
      "loss": 0.1633,
      "step": 9708
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3747841020060527,
      "learning_rate": 2.2544986406374227e-06,
      "loss": 0.1896,
      "step": 9709
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4251162239176987,
      "learning_rate": 2.253530570372221e-06,
      "loss": 0.1921,
      "step": 9710
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.554411122996993,
      "learning_rate": 2.2525626475265872e-06,
      "loss": 0.1882,
      "step": 9711
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4508543175292883,
      "learning_rate": 2.2515948721524776e-06,
      "loss": 0.1696,
      "step": 9712
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4938149483985999,
      "learning_rate": 2.2506272443018345e-06,
      "loss": 0.1745,
      "step": 9713
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3205219015522827,
      "learning_rate": 2.2496597640266006e-06,
      "loss": 0.188,
      "step": 9714
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.303149261159802,
      "learning_rate": 2.2486924313787016e-06,
      "loss": 0.1834,
      "step": 9715
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3436937129281168,
      "learning_rate": 2.2477252464100647e-06,
      "loss": 0.1934,
      "step": 9716
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.757359500335642,
      "learning_rate": 2.2467582091726005e-06,
      "loss": 0.546,
      "step": 9717
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.276507786716258,
      "learning_rate": 2.2457913197182176e-06,
      "loss": 0.1815,
      "step": 9718
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2724609407467427,
      "learning_rate": 2.244824578098814e-06,
      "loss": 0.1684,
      "step": 9719
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3120618546666403,
      "learning_rate": 2.243857984366284e-06,
      "loss": 0.1853,
      "step": 9720
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.43249676899009,
      "learning_rate": 2.242891538572506e-06,
      "loss": 0.2002,
      "step": 9721
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5567195632798496,
      "learning_rate": 2.241925240769358e-06,
      "loss": 0.2131,
      "step": 9722
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.390862661319984,
      "learning_rate": 2.2409590910087043e-06,
      "loss": 0.193,
      "step": 9723
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.442157007922685,
      "learning_rate": 2.2399930893424075e-06,
      "loss": 0.1677,
      "step": 9724
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5379227399043693,
      "learning_rate": 2.2390272358223147e-06,
      "loss": 0.2169,
      "step": 9725
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4466667868936998,
      "learning_rate": 2.238061530500271e-06,
      "loss": 0.1782,
      "step": 9726
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3192603645555172,
      "learning_rate": 2.2370959734281123e-06,
      "loss": 0.1781,
      "step": 9727
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3863468411308855,
      "learning_rate": 2.236130564657665e-06,
      "loss": 0.17,
      "step": 9728
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1720704759578813,
      "learning_rate": 2.2351653042407506e-06,
      "loss": 0.1601,
      "step": 9729
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4201437717355005,
      "learning_rate": 2.2342001922291766e-06,
      "loss": 0.1919,
      "step": 9730
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3056523221471235,
      "learning_rate": 2.2332352286747506e-06,
      "loss": 0.1707,
      "step": 9731
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.433928092347942,
      "learning_rate": 2.2322704136292632e-06,
      "loss": 0.2172,
      "step": 9732
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3068410167898366,
      "learning_rate": 2.2313057471445067e-06,
      "loss": 0.1615,
      "step": 9733
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2976581236634164,
      "learning_rate": 2.2303412292722563e-06,
      "loss": 0.1755,
      "step": 9734
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.363446725123765,
      "learning_rate": 2.2293768600642857e-06,
      "loss": 0.2049,
      "step": 9735
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2796484475894485,
      "learning_rate": 2.2284126395723574e-06,
      "loss": 0.175,
      "step": 9736
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4633672223582117,
      "learning_rate": 2.2274485678482304e-06,
      "loss": 0.2145,
      "step": 9737
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.532321659074599,
      "learning_rate": 2.226484644943647e-06,
      "loss": 0.2231,
      "step": 9738
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.499386988206639,
      "learning_rate": 2.2255208709103514e-06,
      "loss": 0.1848,
      "step": 9739
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3196621951517866,
      "learning_rate": 2.2245572458000714e-06,
      "loss": 0.1861,
      "step": 9740
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2874800065123078,
      "learning_rate": 2.2235937696645337e-06,
      "loss": 0.1965,
      "step": 9741
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4868735999903773,
      "learning_rate": 2.2226304425554505e-06,
      "loss": 0.1956,
      "step": 9742
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.248187980098897,
      "learning_rate": 2.221667264524532e-06,
      "loss": 0.5484,
      "step": 9743
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.271445456407996,
      "learning_rate": 2.220704235623476e-06,
      "loss": 0.1615,
      "step": 9744
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5059273015754457,
      "learning_rate": 2.2197413559039776e-06,
      "loss": 0.2196,
      "step": 9745
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.702554128704514,
      "learning_rate": 2.2187786254177157e-06,
      "loss": 0.5361,
      "step": 9746
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3330002534738616,
      "learning_rate": 2.21781604421637e-06,
      "loss": 0.1557,
      "step": 9747
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1802860971891644,
      "learning_rate": 2.216853612351604e-06,
      "loss": 0.1865,
      "step": 9748
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2008768382834334,
      "learning_rate": 2.2158913298750808e-06,
      "loss": 0.1528,
      "step": 9749
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.209721038773094,
      "learning_rate": 2.214929196838449e-06,
      "loss": 0.1912,
      "step": 9750
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2997292422817348,
      "learning_rate": 2.213967213293353e-06,
      "loss": 0.1606,
      "step": 9751
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.138940089917714,
      "learning_rate": 2.2130053792914293e-06,
      "loss": 0.1854,
      "step": 9752
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2929227307368167,
      "learning_rate": 2.2120436948843054e-06,
      "loss": 0.1564,
      "step": 9753
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2777302272797602,
      "learning_rate": 2.2110821601235994e-06,
      "loss": 0.1798,
      "step": 9754
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.046816416246104,
      "learning_rate": 2.2101207750609244e-06,
      "loss": 0.1493,
      "step": 9755
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4664400815897602,
      "learning_rate": 2.209159539747881e-06,
      "loss": 0.1948,
      "step": 9756
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3631804036081288,
      "learning_rate": 2.208198454236068e-06,
      "loss": 0.1864,
      "step": 9757
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.349327172156145,
      "learning_rate": 2.2072375185770685e-06,
      "loss": 0.1669,
      "step": 9758
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3543415064924562,
      "learning_rate": 2.206276732822464e-06,
      "loss": 0.2043,
      "step": 9759
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.789483604727544,
      "learning_rate": 2.205316097023826e-06,
      "loss": 0.5836,
      "step": 9760
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1719311132301335,
      "learning_rate": 2.204355611232717e-06,
      "loss": 0.1951,
      "step": 9761
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.493055191698878,
      "learning_rate": 2.203395275500694e-06,
      "loss": 0.2191,
      "step": 9762
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3140231440065886,
      "learning_rate": 2.2024350898793007e-06,
      "loss": 0.1704,
      "step": 9763
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.228806126251975,
      "learning_rate": 2.201475054420079e-06,
      "loss": 0.1789,
      "step": 9764
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3956869046372913,
      "learning_rate": 2.2005151691745576e-06,
      "loss": 0.1751,
      "step": 9765
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.271396321940789,
      "learning_rate": 2.199555434194261e-06,
      "loss": 0.1606,
      "step": 9766
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3897115554178203,
      "learning_rate": 2.198595849530702e-06,
      "loss": 0.2011,
      "step": 9767
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2003788274107248,
      "learning_rate": 2.197636415235388e-06,
      "loss": 0.1534,
      "step": 9768
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2540910878332205,
      "learning_rate": 2.1966771313598177e-06,
      "loss": 0.1924,
      "step": 9769
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.284108328479996,
      "learning_rate": 2.1957179979554845e-06,
      "loss": 0.1714,
      "step": 9770
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.340130576373397,
      "learning_rate": 2.1947590150738657e-06,
      "loss": 0.1803,
      "step": 9771
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3238317064201461,
      "learning_rate": 2.193800182766441e-06,
      "loss": 0.1933,
      "step": 9772
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4028670337036626,
      "learning_rate": 2.1928415010846716e-06,
      "loss": 0.1714,
      "step": 9773
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3364499395157738,
      "learning_rate": 2.1918829700800205e-06,
      "loss": 0.2072,
      "step": 9774
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3260300115673664,
      "learning_rate": 2.190924589803933e-06,
      "loss": 0.1924,
      "step": 9775
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3889342063246712,
      "learning_rate": 2.1899663603078547e-06,
      "loss": 0.1712,
      "step": 9776
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.845651577365227,
      "learning_rate": 2.1890082816432178e-06,
      "loss": 0.5323,
      "step": 9777
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5081977872128027,
      "learning_rate": 2.188050353861451e-06,
      "loss": 0.1837,
      "step": 9778
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3551426045966632,
      "learning_rate": 2.187092577013968e-06,
      "loss": 0.1806,
      "step": 9779
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1858690375513175,
      "learning_rate": 2.1861349511521817e-06,
      "loss": 0.1596,
      "step": 9780
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6670239326283454,
      "learning_rate": 2.185177476327491e-06,
      "loss": 0.2124,
      "step": 9781
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.356287409109955,
      "learning_rate": 2.1842201525912926e-06,
      "loss": 0.1789,
      "step": 9782
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3372966736255054,
      "learning_rate": 2.1832629799949677e-06,
      "loss": 0.1799,
      "step": 9783
    },
    {
      "epoch": 0.7,
      "grad_norm": 8.179881864475025,
      "learning_rate": 2.182305958589896e-06,
      "loss": 0.4768,
      "step": 9784
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4942260252605708,
      "learning_rate": 2.1813490884274462e-06,
      "loss": 0.1978,
      "step": 9785
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.230835077684288,
      "learning_rate": 2.1803923695589813e-06,
      "loss": 0.1808,
      "step": 9786
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4082042883415542,
      "learning_rate": 2.179435802035851e-06,
      "loss": 0.1563,
      "step": 9787
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1355504886815047,
      "learning_rate": 2.178479385909402e-06,
      "loss": 0.1257,
      "step": 9788
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2908015201681449,
      "learning_rate": 2.1775231212309696e-06,
      "loss": 0.1408,
      "step": 9789
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2311623759404684,
      "learning_rate": 2.176567008051884e-06,
      "loss": 0.1815,
      "step": 9790
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2706656662120428,
      "learning_rate": 2.175611046423463e-06,
      "loss": 0.1367,
      "step": 9791
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4923360541063972,
      "learning_rate": 2.1746552363970204e-06,
      "loss": 0.1905,
      "step": 9792
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.363760267275215,
      "learning_rate": 2.1736995780238618e-06,
      "loss": 0.8033,
      "step": 9793
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4925938809631258,
      "learning_rate": 2.1727440713552785e-06,
      "loss": 0.1987,
      "step": 9794
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4424911442143913,
      "learning_rate": 2.171788716442564e-06,
      "loss": 0.196,
      "step": 9795
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5236799445345468,
      "learning_rate": 2.170833513336994e-06,
      "loss": 0.2111,
      "step": 9796
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2513693692354264,
      "learning_rate": 2.1698784620898433e-06,
      "loss": 0.1525,
      "step": 9797
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.199117341283263,
      "learning_rate": 2.1689235627523713e-06,
      "loss": 0.1627,
      "step": 9798
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2359381822172406,
      "learning_rate": 2.1679688153758373e-06,
      "loss": 0.1302,
      "step": 9799
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3557157055208306,
      "learning_rate": 2.1670142200114837e-06,
      "loss": 0.1824,
      "step": 9800
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1336371203771571,
      "learning_rate": 2.166059776710554e-06,
      "loss": 0.185,
      "step": 9801
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4009504690460342,
      "learning_rate": 2.1651054855242754e-06,
      "loss": 0.1655,
      "step": 9802
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1914045534611415,
      "learning_rate": 2.1641513465038714e-06,
      "loss": 0.1293,
      "step": 9803
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3707046480957197,
      "learning_rate": 2.1631973597005574e-06,
      "loss": 0.1887,
      "step": 9804
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2457669192725758,
      "learning_rate": 2.162243525165541e-06,
      "loss": 0.1487,
      "step": 9805
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2884176430997798,
      "learning_rate": 2.1612898429500162e-06,
      "loss": 0.153,
      "step": 9806
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2165153869360554,
      "learning_rate": 2.1603363131051767e-06,
      "loss": 0.1519,
      "step": 9807
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.196684562667685,
      "learning_rate": 2.159382935682201e-06,
      "loss": 0.1554,
      "step": 9808
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.035757256589133,
      "learning_rate": 2.158429710732266e-06,
      "loss": 0.6991,
      "step": 9809
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.386838360759269,
      "learning_rate": 2.157476638306533e-06,
      "loss": 0.1738,
      "step": 9810
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4564464976376614,
      "learning_rate": 2.1565237184561614e-06,
      "loss": 0.1973,
      "step": 9811
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.371689022823875,
      "learning_rate": 2.1555709512323004e-06,
      "loss": 0.5556,
      "step": 9812
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6310505582621118,
      "learning_rate": 2.1546183366860917e-06,
      "loss": 0.2048,
      "step": 9813
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.271839509677533,
      "learning_rate": 2.1536658748686652e-06,
      "loss": 0.1893,
      "step": 9814
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2292143440068781,
      "learning_rate": 2.1527135658311482e-06,
      "loss": 0.1775,
      "step": 9815
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.399743386058442,
      "learning_rate": 2.1517614096246537e-06,
      "loss": 0.214,
      "step": 9816
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.108654972731843,
      "learning_rate": 2.1508094063002933e-06,
      "loss": 0.4864,
      "step": 9817
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3752551184363477,
      "learning_rate": 2.1498575559091626e-06,
      "loss": 0.1887,
      "step": 9818
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4917364082271154,
      "learning_rate": 2.1489058585023555e-06,
      "loss": 0.1832,
      "step": 9819
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3879291609473101,
      "learning_rate": 2.147954314130955e-06,
      "loss": 0.1881,
      "step": 9820
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3730433904924315,
      "learning_rate": 2.1470029228460383e-06,
      "loss": 0.1681,
      "step": 9821
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2250160081962989,
      "learning_rate": 2.1460516846986677e-06,
      "loss": 0.1732,
      "step": 9822
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3869059515505648,
      "learning_rate": 2.1451005997399055e-06,
      "loss": 0.1977,
      "step": 9823
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2458557758400128,
      "learning_rate": 2.144149668020803e-06,
      "loss": 0.1735,
      "step": 9824
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.269771610125317,
      "learning_rate": 2.1431988895923985e-06,
      "loss": 0.1826,
      "step": 9825
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.414956748198051,
      "learning_rate": 2.1422482645057297e-06,
      "loss": 0.1994,
      "step": 9826
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4139962881941133,
      "learning_rate": 2.1412977928118194e-06,
      "loss": 0.168,
      "step": 9827
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4817229597478334,
      "learning_rate": 2.1403474745616863e-06,
      "loss": 0.1578,
      "step": 9828
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3136500060251255,
      "learning_rate": 2.1393973098063403e-06,
      "loss": 0.1607,
      "step": 9829
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.376707416839989,
      "learning_rate": 2.1384472985967834e-06,
      "loss": 0.1611,
      "step": 9830
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4227680260801439,
      "learning_rate": 2.137497440984006e-06,
      "loss": 0.1892,
      "step": 9831
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3010916342822003,
      "learning_rate": 2.136547737018995e-06,
      "loss": 0.1818,
      "step": 9832
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1882299843679442,
      "learning_rate": 2.1355981867527242e-06,
      "loss": 0.1553,
      "step": 9833
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3918362860062254,
      "learning_rate": 2.134648790236165e-06,
      "loss": 0.1777,
      "step": 9834
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.177337817612947,
      "learning_rate": 2.133699547520273e-06,
      "loss": 0.1623,
      "step": 9835
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2146118372014378,
      "learning_rate": 2.1327504586560027e-06,
      "loss": 0.1656,
      "step": 9836
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.859983861030236,
      "learning_rate": 2.131801523694296e-06,
      "loss": 0.6272,
      "step": 9837
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.086489384171177,
      "learning_rate": 2.1308527426860913e-06,
      "loss": 0.7281,
      "step": 9838
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.358152726796936,
      "learning_rate": 2.129904115682311e-06,
      "loss": 0.1933,
      "step": 9839
    },
    {
      "epoch": 0.7,
      "grad_norm": 14.402683899193852,
      "learning_rate": 2.128955642733877e-06,
      "loss": 0.559,
      "step": 9840
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2552467660847846,
      "learning_rate": 2.1280073238916963e-06,
      "loss": 0.1353,
      "step": 9841
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4164923735629824,
      "learning_rate": 2.127059159206675e-06,
      "loss": 0.1872,
      "step": 9842
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3616961451551473,
      "learning_rate": 2.1261111487297016e-06,
      "loss": 0.1901,
      "step": 9843
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4764420279718353,
      "learning_rate": 2.125163292511665e-06,
      "loss": 0.2138,
      "step": 9844
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4524883720008324,
      "learning_rate": 2.124215590603442e-06,
      "loss": 0.1892,
      "step": 9845
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6297056938074792,
      "learning_rate": 2.1232680430559023e-06,
      "loss": 0.1628,
      "step": 9846
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3378764285105262,
      "learning_rate": 2.122320649919904e-06,
      "loss": 0.2065,
      "step": 9847
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.328894460021943,
      "learning_rate": 2.1213734112463023e-06,
      "loss": 0.1826,
      "step": 9848
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3626453592908885,
      "learning_rate": 2.1204263270859382e-06,
      "loss": 0.189,
      "step": 9849
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3305004126096318,
      "learning_rate": 2.119479397489651e-06,
      "loss": 0.1843,
      "step": 9850
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.092708529617387,
      "learning_rate": 2.1185326225082635e-06,
      "loss": 0.4826,
      "step": 9851
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.24977364229167,
      "learning_rate": 2.117586002192598e-06,
      "loss": 0.5784,
      "step": 9852
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1703383260185363,
      "learning_rate": 2.1166395365934644e-06,
      "loss": 0.175,
      "step": 9853
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4133920836005898,
      "learning_rate": 2.1156932257616675e-06,
      "loss": 0.181,
      "step": 9854
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5406109202978615,
      "learning_rate": 2.1147470697479977e-06,
      "loss": 0.2006,
      "step": 9855
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2641386257074096,
      "learning_rate": 2.113801068603242e-06,
      "loss": 0.1567,
      "step": 9856
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.199845695098481,
      "learning_rate": 2.1128552223781813e-06,
      "loss": 0.178,
      "step": 9857
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4346026560783243,
      "learning_rate": 2.11190953112358e-06,
      "loss": 0.1845,
      "step": 9858
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.435736365003967,
      "learning_rate": 2.1109639948902035e-06,
      "loss": 0.6163,
      "step": 9859
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.493198960772928,
      "learning_rate": 2.1100186137288005e-06,
      "loss": 0.6731,
      "step": 9860
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.558155063149377,
      "learning_rate": 2.1090733876901167e-06,
      "loss": 0.1645,
      "step": 9861
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2368114440515035,
      "learning_rate": 2.1081283168248877e-06,
      "loss": 0.171,
      "step": 9862
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.707143597347296,
      "learning_rate": 2.107183401183845e-06,
      "loss": 0.5725,
      "step": 9863
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4476322191348134,
      "learning_rate": 2.106238640817702e-06,
      "loss": 0.1465,
      "step": 9864
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3857148022505723,
      "learning_rate": 2.1052940357771744e-06,
      "loss": 0.1534,
      "step": 9865
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1512504628727052,
      "learning_rate": 2.1043495861129616e-06,
      "loss": 0.1542,
      "step": 9866
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3643399260164875,
      "learning_rate": 2.103405291875761e-06,
      "loss": 0.179,
      "step": 9867
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3144717942036308,
      "learning_rate": 2.102461153116254e-06,
      "loss": 0.1745,
      "step": 9868
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3178520653149153,
      "learning_rate": 2.1015171698851224e-06,
      "loss": 0.1922,
      "step": 9869
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2049150813341212,
      "learning_rate": 2.1005733422330327e-06,
      "loss": 0.1842,
      "step": 9870
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3238706714513198,
      "learning_rate": 2.09962967021065e-06,
      "loss": 0.1645,
      "step": 9871
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.052272989707731,
      "learning_rate": 2.0986861538686217e-06,
      "loss": 0.5756,
      "step": 9872
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1180484463234799,
      "learning_rate": 2.0977427932575962e-06,
      "loss": 0.1774,
      "step": 9873
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3877540826058457,
      "learning_rate": 2.0967995884282063e-06,
      "loss": 0.2093,
      "step": 9874
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1723618846940806,
      "learning_rate": 2.0958565394310815e-06,
      "loss": 0.1538,
      "step": 9875
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.711517928748774,
      "learning_rate": 2.094913646316839e-06,
      "loss": 0.7216,
      "step": 9876
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.44195770017684,
      "learning_rate": 2.0939709091360903e-06,
      "loss": 0.6406,
      "step": 9877
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2670440421966707,
      "learning_rate": 2.093028327939439e-06,
      "loss": 0.1575,
      "step": 9878
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7101381262791193,
      "learning_rate": 2.0920859027774786e-06,
      "loss": 0.2087,
      "step": 9879
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2059864574999795,
      "learning_rate": 2.0911436337007935e-06,
      "loss": 0.1519,
      "step": 9880
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2782775145517447,
      "learning_rate": 2.0902015207599634e-06,
      "loss": 0.1547,
      "step": 9881
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2985311050471064,
      "learning_rate": 2.0892595640055548e-06,
      "loss": 0.16,
      "step": 9882
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.447111342126725,
      "learning_rate": 2.0883177634881296e-06,
      "loss": 0.2169,
      "step": 9883
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3203283399814725,
      "learning_rate": 2.0873761192582383e-06,
      "loss": 0.1533,
      "step": 9884
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1696047562410419,
      "learning_rate": 2.086434631366425e-06,
      "loss": 0.1726,
      "step": 9885
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.325639447985463,
      "learning_rate": 2.0854932998632267e-06,
      "loss": 0.2013,
      "step": 9886
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3861177849146418,
      "learning_rate": 2.0845521247991683e-06,
      "loss": 0.2155,
      "step": 9887
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.909888771878877,
      "learning_rate": 2.0836111062247717e-06,
      "loss": 0.5115,
      "step": 9888
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4270173560107637,
      "learning_rate": 2.0826702441905426e-06,
      "loss": 0.1888,
      "step": 9889
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.63940189521589,
      "learning_rate": 2.081729538746987e-06,
      "loss": 0.4169,
      "step": 9890
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1639412240224096,
      "learning_rate": 2.0807889899445944e-06,
      "loss": 0.1279,
      "step": 9891
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4223063733194534,
      "learning_rate": 2.0798485978338527e-06,
      "loss": 0.1761,
      "step": 9892
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1386563403704233,
      "learning_rate": 2.0789083624652364e-06,
      "loss": 0.1352,
      "step": 9893
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3067980433377382,
      "learning_rate": 2.0779682838892153e-06,
      "loss": 0.1748,
      "step": 9894
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.438303718230723,
      "learning_rate": 2.077028362156244e-06,
      "loss": 0.6173,
      "step": 9895
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3225269162277258,
      "learning_rate": 2.0760885973167822e-06,
      "loss": 0.1789,
      "step": 9896
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2403730206078722,
      "learning_rate": 2.0751489894212667e-06,
      "loss": 0.1529,
      "step": 9897
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4962382933372795,
      "learning_rate": 2.074209538520135e-06,
      "loss": 0.177,
      "step": 9898
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6072497697934938,
      "learning_rate": 2.07327024466381e-06,
      "loss": 0.2006,
      "step": 9899
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1065627984714854,
      "learning_rate": 2.072331107902713e-06,
      "loss": 0.1608,
      "step": 9900
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4269701400323827,
      "learning_rate": 2.0713921282872492e-06,
      "loss": 0.1861,
      "step": 9901
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2948558793292697,
      "learning_rate": 2.070453305867823e-06,
      "loss": 0.1614,
      "step": 9902
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3884971562191146,
      "learning_rate": 2.069514640694823e-06,
      "loss": 0.1466,
      "step": 9903
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3290346527325099,
      "learning_rate": 2.0685761328186347e-06,
      "loss": 0.1626,
      "step": 9904
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.346545724483843,
      "learning_rate": 2.0676377822896344e-06,
      "loss": 0.1606,
      "step": 9905
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3760334978048923,
      "learning_rate": 2.06669958915819e-06,
      "loss": 0.1768,
      "step": 9906
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3879009732809675,
      "learning_rate": 2.0657615534746556e-06,
      "loss": 0.1708,
      "step": 9907
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3111919220892199,
      "learning_rate": 2.0648236752893864e-06,
      "loss": 0.1564,
      "step": 9908
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4068572242027364,
      "learning_rate": 2.0638859546527197e-06,
      "loss": 0.1577,
      "step": 9909
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3082797608506729,
      "learning_rate": 2.0629483916149917e-06,
      "loss": 0.159,
      "step": 9910
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3662036495798913,
      "learning_rate": 2.0620109862265243e-06,
      "loss": 0.181,
      "step": 9911
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4756062145129323,
      "learning_rate": 2.061073738537635e-06,
      "loss": 0.165,
      "step": 9912
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4654128865018698,
      "learning_rate": 2.060136648598632e-06,
      "loss": 0.2105,
      "step": 9913
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3417563355846003,
      "learning_rate": 2.0591997164598156e-06,
      "loss": 0.1395,
      "step": 9914
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.37793674815471,
      "learning_rate": 2.0582629421714732e-06,
      "loss": 0.5017,
      "step": 9915
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4523592614749674,
      "learning_rate": 2.0573263257838916e-06,
      "loss": 0.2006,
      "step": 9916
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.667190535424932,
      "learning_rate": 2.0563898673473404e-06,
      "loss": 0.5101,
      "step": 9917
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4519457092220538,
      "learning_rate": 2.0554535669120884e-06,
      "loss": 0.1938,
      "step": 9918
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3950910565768653,
      "learning_rate": 2.054517424528389e-06,
      "loss": 0.1709,
      "step": 9919
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3277775889218504,
      "learning_rate": 2.0535814402464922e-06,
      "loss": 0.1455,
      "step": 9920
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.292053353997159,
      "learning_rate": 2.0526456141166383e-06,
      "loss": 0.2162,
      "step": 9921
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.278502844299361,
      "learning_rate": 2.0517099461890595e-06,
      "loss": 0.1805,
      "step": 9922
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2475328412925255,
      "learning_rate": 2.0507744365139787e-06,
      "loss": 0.1343,
      "step": 9923
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2698321236183767,
      "learning_rate": 2.049839085141608e-06,
      "loss": 0.141,
      "step": 9924
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3500628775322452,
      "learning_rate": 2.0489038921221567e-06,
      "loss": 0.1625,
      "step": 9925
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3754737956966643,
      "learning_rate": 2.047968857505818e-06,
      "loss": 0.1704,
      "step": 9926
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1611271725569305,
      "learning_rate": 2.0470339813427857e-06,
      "loss": 0.1801,
      "step": 9927
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4097202962325988,
      "learning_rate": 2.046099263683236e-06,
      "loss": 0.1768,
      "step": 9928
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.30380353723763,
      "learning_rate": 2.045164704577342e-06,
      "loss": 0.5791,
      "step": 9929
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2612467596651904,
      "learning_rate": 2.0442303040752687e-06,
      "loss": 0.1638,
      "step": 9930
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3172675483473377,
      "learning_rate": 2.0432960622271707e-06,
      "loss": 0.1713,
      "step": 9931
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2776405878156374,
      "learning_rate": 2.0423619790831926e-06,
      "loss": 0.1655,
      "step": 9932
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4325661991318008,
      "learning_rate": 2.041428054693475e-06,
      "loss": 0.1958,
      "step": 9933
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3905024627115765,
      "learning_rate": 2.0404942891081442e-06,
      "loss": 0.1588,
      "step": 9934
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3750492736772684,
      "learning_rate": 2.0395606823773238e-06,
      "loss": 0.1925,
      "step": 9935
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2091818285473133,
      "learning_rate": 2.038627234551123e-06,
      "loss": 0.1733,
      "step": 9936
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2428074399341464,
      "learning_rate": 2.0376939456796473e-06,
      "loss": 0.1837,
      "step": 9937
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2315659125835232,
      "learning_rate": 2.0367608158129924e-06,
      "loss": 0.1678,
      "step": 9938
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3487241133819814,
      "learning_rate": 2.0358278450012466e-06,
      "loss": 0.1976,
      "step": 9939
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2408577807948782,
      "learning_rate": 2.034895033294483e-06,
      "loss": 0.1895,
      "step": 9940
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.294963169940782,
      "learning_rate": 2.033962380742777e-06,
      "loss": 0.1765,
      "step": 9941
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.177799872035875,
      "learning_rate": 2.033029887396185e-06,
      "loss": 0.1402,
      "step": 9942
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3166406745042563,
      "learning_rate": 2.0320975533047636e-06,
      "loss": 0.1815,
      "step": 9943
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3162591720066166,
      "learning_rate": 2.0311653785185524e-06,
      "loss": 0.1309,
      "step": 9944
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3865209658904454,
      "learning_rate": 2.0302333630875896e-06,
      "loss": 0.2024,
      "step": 9945
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4070486787694327,
      "learning_rate": 2.0293015070619014e-06,
      "loss": 0.1492,
      "step": 9946
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3413435702578953,
      "learning_rate": 2.0283698104915077e-06,
      "loss": 0.1734,
      "step": 9947
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2040091017762542,
      "learning_rate": 2.027438273426416e-06,
      "loss": 0.1314,
      "step": 9948
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3255688061352577,
      "learning_rate": 2.0265068959166295e-06,
      "loss": 0.1637,
      "step": 9949
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.577348153990678,
      "learning_rate": 2.025575678012138e-06,
      "loss": 0.5839,
      "step": 9950
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3592618493514435,
      "learning_rate": 2.024644619762928e-06,
      "loss": 0.1713,
      "step": 9951
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4135629960848357,
      "learning_rate": 2.0237137212189757e-06,
      "loss": 0.2316,
      "step": 9952
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.660838090743961,
      "learning_rate": 2.0227829824302447e-06,
      "loss": 0.5356,
      "step": 9953
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5376454522037042,
      "learning_rate": 2.021852403446695e-06,
      "loss": 0.1626,
      "step": 9954
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.220637733301534,
      "learning_rate": 2.0209219843182774e-06,
      "loss": 0.1588,
      "step": 9955
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.231993916316084,
      "learning_rate": 2.0199917250949343e-06,
      "loss": 0.1846,
      "step": 9956
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.253543716851911,
      "learning_rate": 2.0190616258265944e-06,
      "loss": 0.1699,
      "step": 9957
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2599842700727308,
      "learning_rate": 2.0181316865631856e-06,
      "loss": 0.1527,
      "step": 9958
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2121234553443256,
      "learning_rate": 2.0172019073546194e-06,
      "loss": 0.1428,
      "step": 9959
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.604587799419003,
      "learning_rate": 2.0162722882508072e-06,
      "loss": 0.4892,
      "step": 9960
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.321906946177831,
      "learning_rate": 2.015342829301643e-06,
      "loss": 0.1524,
      "step": 9961
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3314032643648415,
      "learning_rate": 2.0144135305570176e-06,
      "loss": 0.1813,
      "step": 9962
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.391198732418022,
      "learning_rate": 2.0134843920668135e-06,
      "loss": 0.1856,
      "step": 9963
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2551270398026033,
      "learning_rate": 2.012555413880904e-06,
      "loss": 0.1462,
      "step": 9964
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.637962012578639,
      "learning_rate": 2.0116265960491498e-06,
      "loss": 0.1778,
      "step": 9965
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4563355136386023,
      "learning_rate": 2.0106979386214092e-06,
      "loss": 0.1746,
      "step": 9966
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5458919543662657,
      "learning_rate": 2.0097694416475256e-06,
      "loss": 0.1896,
      "step": 9967
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3830086251858307,
      "learning_rate": 2.008841105177341e-06,
      "loss": 0.2009,
      "step": 9968
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.320823688855238,
      "learning_rate": 2.0079129292606804e-06,
      "loss": 0.2028,
      "step": 9969
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3374183490147276,
      "learning_rate": 2.0069849139473668e-06,
      "loss": 0.1859,
      "step": 9970
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2062755336057875,
      "learning_rate": 2.006057059287213e-06,
      "loss": 0.1643,
      "step": 9971
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.517620449608032,
      "learning_rate": 2.005129365330023e-06,
      "loss": 0.2083,
      "step": 9972
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.500973331345831,
      "learning_rate": 2.0042018321255897e-06,
      "loss": 0.5806,
      "step": 9973
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5311301816314118,
      "learning_rate": 2.0032744597237013e-06,
      "loss": 0.1849,
      "step": 9974
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.288556996109179,
      "learning_rate": 2.0023472481741336e-06,
      "loss": 0.6284,
      "step": 9975
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.034290350631877,
      "learning_rate": 2.001420197526658e-06,
      "loss": 0.5797,
      "step": 9976
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.830220074581334,
      "learning_rate": 2.0004933078310314e-06,
      "loss": 0.4121,
      "step": 9977
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.275727810170509,
      "learning_rate": 1.999566579137008e-06,
      "loss": 0.1713,
      "step": 9978
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5029710576558715,
      "learning_rate": 1.99864001149433e-06,
      "loss": 0.2002,
      "step": 9979
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.295772751036187,
      "learning_rate": 1.9977136049527348e-06,
      "loss": 0.1541,
      "step": 9980
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4987024650526921,
      "learning_rate": 1.9967873595619438e-06,
      "loss": 0.1784,
      "step": 9981
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3964645802324291,
      "learning_rate": 1.9958612753716777e-06,
      "loss": 0.1934,
      "step": 9982
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.787899154429594,
      "learning_rate": 1.994935352431642e-06,
      "loss": 0.6112,
      "step": 9983
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3635541754658314,
      "learning_rate": 1.9940095907915385e-06,
      "loss": 0.1792,
      "step": 9984
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1980851306409523,
      "learning_rate": 1.993083990501059e-06,
      "loss": 0.1544,
      "step": 9985
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3459564980767784,
      "learning_rate": 1.992158551609884e-06,
      "loss": 0.2406,
      "step": 9986
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5303919745846444,
      "learning_rate": 1.9912332741676883e-06,
      "loss": 0.2167,
      "step": 9987
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.555531857659759,
      "learning_rate": 1.9903081582241368e-06,
      "loss": 0.5453,
      "step": 9988
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4371404332673614,
      "learning_rate": 1.9893832038288884e-06,
      "loss": 0.187,
      "step": 9989
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4627928288988505,
      "learning_rate": 1.9884584110315872e-06,
      "loss": 0.1917,
      "step": 9990
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.279006588744145,
      "learning_rate": 1.9875337798818768e-06,
      "loss": 0.1713,
      "step": 9991
    },
    {
      "epoch": 0.71,
      "grad_norm": 9.294157588002328,
      "learning_rate": 1.9866093104293825e-06,
      "loss": 0.627,
      "step": 9992
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1187815813718136,
      "learning_rate": 1.985685002723731e-06,
      "loss": 0.1534,
      "step": 9993
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.344608284221575,
      "learning_rate": 1.9847608568145322e-06,
      "loss": 0.1708,
      "step": 9994
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4110959257349243,
      "learning_rate": 1.9838368727513922e-06,
      "loss": 0.1923,
      "step": 9995
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4722265504927314,
      "learning_rate": 1.9829130505839058e-06,
      "loss": 0.1786,
      "step": 9996
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.169097784340746,
      "learning_rate": 1.98198939036166e-06,
      "loss": 0.5308,
      "step": 9997
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3565470534271693,
      "learning_rate": 1.9810658921342335e-06,
      "loss": 0.1539,
      "step": 9998
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2084401665284408,
      "learning_rate": 1.9801425559511983e-06,
      "loss": 0.1362,
      "step": 9999
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.308036139495296,
      "learning_rate": 1.9792193818621118e-06,
      "loss": 0.178,
      "step": 10000
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3002703300010954,
      "learning_rate": 1.97829636991653e-06,
      "loss": 0.181,
      "step": 10001
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3506514018375053,
      "learning_rate": 1.977373520163992e-06,
      "loss": 0.1588,
      "step": 10002
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.258429056924509,
      "learning_rate": 1.9764508326540373e-06,
      "loss": 0.1675,
      "step": 10003
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3616523183499503,
      "learning_rate": 1.975528307436188e-06,
      "loss": 0.1957,
      "step": 10004
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.515419103999382,
      "learning_rate": 1.9746059445599642e-06,
      "loss": 0.2022,
      "step": 10005
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3622111500585605,
      "learning_rate": 1.9736837440748736e-06,
      "loss": 0.2195,
      "step": 10006
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3516561393619462,
      "learning_rate": 1.9727617060304184e-06,
      "loss": 0.1945,
      "step": 10007
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4203374848910484,
      "learning_rate": 1.9718398304760865e-06,
      "loss": 0.1546,
      "step": 10008
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.319376882198309,
      "learning_rate": 1.970918117461364e-06,
      "loss": 0.2016,
      "step": 10009
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2041856623861882,
      "learning_rate": 1.969996567035722e-06,
      "loss": 0.1732,
      "step": 10010
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2950772399705848,
      "learning_rate": 1.9690751792486272e-06,
      "loss": 0.1745,
      "step": 10011
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.299377524873502,
      "learning_rate": 1.968153954149535e-06,
      "loss": 0.1824,
      "step": 10012
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2051540026024092,
      "learning_rate": 1.967232891787893e-06,
      "loss": 0.1451,
      "step": 10013
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3265857607816367,
      "learning_rate": 1.9663119922131406e-06,
      "loss": 0.1888,
      "step": 10014
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3483994333627227,
      "learning_rate": 1.965391255474709e-06,
      "loss": 0.195,
      "step": 10015
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4647502373067502,
      "learning_rate": 1.9644706816220198e-06,
      "loss": 0.1469,
      "step": 10016
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2521323132240383,
      "learning_rate": 1.963550270704484e-06,
      "loss": 0.1608,
      "step": 10017
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2955625198668186,
      "learning_rate": 1.962630022771508e-06,
      "loss": 0.1975,
      "step": 10018
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.574089501170558,
      "learning_rate": 1.9617099378724835e-06,
      "loss": 0.4704,
      "step": 10019
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3241325547782847,
      "learning_rate": 1.960790016056801e-06,
      "loss": 0.1613,
      "step": 10020
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2642620553053199,
      "learning_rate": 1.9598702573738343e-06,
      "loss": 0.1469,
      "step": 10021
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3660707676889527,
      "learning_rate": 1.958950661872955e-06,
      "loss": 0.1619,
      "step": 10022
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2421609850287287,
      "learning_rate": 1.958031229603522e-06,
      "loss": 0.1747,
      "step": 10023
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2945827617360746,
      "learning_rate": 1.9571119606148896e-06,
      "loss": 0.1774,
      "step": 10024
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.174257667911944,
      "learning_rate": 1.956192854956397e-06,
      "loss": 0.5525,
      "step": 10025
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.35846071902805,
      "learning_rate": 1.9552739126773807e-06,
      "loss": 0.1821,
      "step": 10026
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3905322218653868,
      "learning_rate": 1.9543551338271633e-06,
      "loss": 0.1413,
      "step": 10027
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3412091760095721,
      "learning_rate": 1.953436518455064e-06,
      "loss": 0.1479,
      "step": 10028
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.002538100300138,
      "learning_rate": 1.9525180666103877e-06,
      "loss": 0.469,
      "step": 10029
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.229061561538334,
      "learning_rate": 1.9515997783424346e-06,
      "loss": 0.1552,
      "step": 10030
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1380014851653208,
      "learning_rate": 1.950681653700494e-06,
      "loss": 0.1382,
      "step": 10031
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2582997770057984,
      "learning_rate": 1.9497636927338502e-06,
      "loss": 0.1662,
      "step": 10032
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3257997003881854,
      "learning_rate": 1.948845895491772e-06,
      "loss": 0.1454,
      "step": 10033
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.353889464227676,
      "learning_rate": 1.947928262023526e-06,
      "loss": 0.1798,
      "step": 10034
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3721545573066014,
      "learning_rate": 1.9470107923783636e-06,
      "loss": 0.1905,
      "step": 10035
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.333359598590068,
      "learning_rate": 1.946093486605535e-06,
      "loss": 0.172,
      "step": 10036
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2988439125189775,
      "learning_rate": 1.945176344754274e-06,
      "loss": 0.1612,
      "step": 10037
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.381404634904194,
      "learning_rate": 1.944259366873811e-06,
      "loss": 0.1757,
      "step": 10038
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3110011432744135,
      "learning_rate": 1.9433425530133643e-06,
      "loss": 0.1436,
      "step": 10039
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4265984553693554,
      "learning_rate": 1.9424259032221482e-06,
      "loss": 0.1945,
      "step": 10040
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4711115379390938,
      "learning_rate": 1.941509417549361e-06,
      "loss": 0.1783,
      "step": 10041
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.371635070062476,
      "learning_rate": 1.9405930960441987e-06,
      "loss": 0.2102,
      "step": 10042
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2427339482459354,
      "learning_rate": 1.9396769387558433e-06,
      "loss": 0.1684,
      "step": 10043
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3658652040803707,
      "learning_rate": 1.9387609457334734e-06,
      "loss": 0.1768,
      "step": 10044
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.37624817996891,
      "learning_rate": 1.9378451170262525e-06,
      "loss": 0.4902,
      "step": 10045
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2122479629862002,
      "learning_rate": 1.9369294526833406e-06,
      "loss": 0.1843,
      "step": 10046
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2193864298399844,
      "learning_rate": 1.936013952753887e-06,
      "loss": 0.1483,
      "step": 10047
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2996167821346771,
      "learning_rate": 1.9350986172870314e-06,
      "loss": 0.1823,
      "step": 10048
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3273127957719435,
      "learning_rate": 1.9341834463319077e-06,
      "loss": 0.1754,
      "step": 10049
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3914272874714133,
      "learning_rate": 1.9332684399376343e-06,
      "loss": 0.1589,
      "step": 10050
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2809711411306326,
      "learning_rate": 1.9323535981533303e-06,
      "loss": 0.1812,
      "step": 10051
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2215052659139345,
      "learning_rate": 1.931438921028096e-06,
      "loss": 0.1534,
      "step": 10052
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3799929602418441,
      "learning_rate": 1.9305244086110314e-06,
      "loss": 0.1851,
      "step": 10053
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4136260567362546,
      "learning_rate": 1.9296100609512203e-06,
      "loss": 0.1576,
      "step": 10054
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.428626801621541,
      "learning_rate": 1.9286958780977433e-06,
      "loss": 0.1944,
      "step": 10055
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5161120801102852,
      "learning_rate": 1.92778186009967e-06,
      "loss": 0.1945,
      "step": 10056
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2799630540918274,
      "learning_rate": 1.9268680070060624e-06,
      "loss": 0.1671,
      "step": 10057
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.598641089843731,
      "learning_rate": 1.92595431886597e-06,
      "loss": 0.169,
      "step": 10058
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1255589237568593,
      "learning_rate": 1.925040795728439e-06,
      "loss": 0.1561,
      "step": 10059
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3564080449031184,
      "learning_rate": 1.9241274376425e-06,
      "loss": 0.1946,
      "step": 10060
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3266289835649838,
      "learning_rate": 1.923214244657182e-06,
      "loss": 0.1802,
      "step": 10061
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2245294203361756,
      "learning_rate": 1.9223012168214987e-06,
      "loss": 0.1624,
      "step": 10062
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3565029340392736,
      "learning_rate": 1.9213883541844586e-06,
      "loss": 0.1463,
      "step": 10063
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4228776587705079,
      "learning_rate": 1.920475656795061e-06,
      "loss": 0.1809,
      "step": 10064
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.473325571278964,
      "learning_rate": 1.9195631247022977e-06,
      "loss": 0.1778,
      "step": 10065
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.224706257875385,
      "learning_rate": 1.918650757955146e-06,
      "loss": 0.1319,
      "step": 10066
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3103477211728565,
      "learning_rate": 1.9177385566025817e-06,
      "loss": 0.2033,
      "step": 10067
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2267250095222508,
      "learning_rate": 1.9168265206935655e-06,
      "loss": 0.1439,
      "step": 10068
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3295437827514043,
      "learning_rate": 1.9159146502770536e-06,
      "loss": 0.1582,
      "step": 10069
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3746214633304938,
      "learning_rate": 1.91500294540199e-06,
      "loss": 0.1777,
      "step": 10070
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.275820945195789,
      "learning_rate": 1.9140914061173115e-06,
      "loss": 0.1917,
      "step": 10071
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.225934904213671,
      "learning_rate": 1.9131800324719474e-06,
      "loss": 0.5913,
      "step": 10072
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1598667659136688,
      "learning_rate": 1.9122688245148168e-06,
      "loss": 0.1885,
      "step": 10073
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.37660095914232,
      "learning_rate": 1.9113577822948276e-06,
      "loss": 0.1605,
      "step": 10074
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3510985525150097,
      "learning_rate": 1.910446905860884e-06,
      "loss": 0.1836,
      "step": 10075
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.458284947029599,
      "learning_rate": 1.909536195261874e-06,
      "loss": 0.181,
      "step": 10076
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3453052074973277,
      "learning_rate": 1.9086256505466856e-06,
      "loss": 0.156,
      "step": 10077
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.293418025993272,
      "learning_rate": 1.907715271764189e-06,
      "loss": 0.1715,
      "step": 10078
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4331204823445411,
      "learning_rate": 1.9068050589632514e-06,
      "loss": 0.2183,
      "step": 10079
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2970533702774103,
      "learning_rate": 1.90589501219273e-06,
      "loss": 0.1485,
      "step": 10080
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4101586248907372,
      "learning_rate": 1.9049851315014716e-06,
      "loss": 0.1953,
      "step": 10081
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.448633265910556,
      "learning_rate": 1.9040754169383174e-06,
      "loss": 0.1752,
      "step": 10082
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3671429913082793,
      "learning_rate": 1.9031658685520936e-06,
      "loss": 0.1934,
      "step": 10083
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.195816997470254,
      "learning_rate": 1.902256486391625e-06,
      "loss": 0.1597,
      "step": 10084
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2902389125960165,
      "learning_rate": 1.9013472705057196e-06,
      "loss": 0.159,
      "step": 10085
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4564688015064238,
      "learning_rate": 1.9004382209431843e-06,
      "loss": 0.2074,
      "step": 10086
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.186170901211127,
      "learning_rate": 1.8995293377528096e-06,
      "loss": 0.1682,
      "step": 10087
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.262405906817386,
      "learning_rate": 1.8986206209833823e-06,
      "loss": 0.1334,
      "step": 10088
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1288114109466896,
      "learning_rate": 1.8977120706836788e-06,
      "loss": 0.1705,
      "step": 10089
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4114182890723526,
      "learning_rate": 1.8968036869024687e-06,
      "loss": 0.166,
      "step": 10090
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3364538906088588,
      "learning_rate": 1.8958954696885068e-06,
      "loss": 0.1829,
      "step": 10091
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2493150823516865,
      "learning_rate": 1.8949874190905453e-06,
      "loss": 0.1431,
      "step": 10092
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3950656544124114,
      "learning_rate": 1.8940795351573222e-06,
      "loss": 0.1957,
      "step": 10093
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4991913687742395,
      "learning_rate": 1.8931718179375724e-06,
      "loss": 0.2117,
      "step": 10094
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3383928984238893,
      "learning_rate": 1.8922642674800146e-06,
      "loss": 0.1675,
      "step": 10095
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2874503051690889,
      "learning_rate": 1.8913568838333663e-06,
      "loss": 0.1795,
      "step": 10096
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4104368280808777,
      "learning_rate": 1.8904496670463285e-06,
      "loss": 0.2047,
      "step": 10097
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.546129141814203,
      "learning_rate": 1.8895426171675996e-06,
      "loss": 0.3746,
      "step": 10098
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3119353317967886,
      "learning_rate": 1.8886357342458655e-06,
      "loss": 0.1825,
      "step": 10099
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4225633111052087,
      "learning_rate": 1.8877290183298058e-06,
      "loss": 0.1913,
      "step": 10100
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3444380409394994,
      "learning_rate": 1.886822469468087e-06,
      "loss": 0.1703,
      "step": 10101
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.408869452331511,
      "learning_rate": 1.8859160877093713e-06,
      "loss": 0.1577,
      "step": 10102
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3767288644558326,
      "learning_rate": 1.8850098731023064e-06,
      "loss": 0.1988,
      "step": 10103
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.62319917502275,
      "learning_rate": 1.8841038256955385e-06,
      "loss": 0.1714,
      "step": 10104
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1138993934121302,
      "learning_rate": 1.883197945537697e-06,
      "loss": 0.1558,
      "step": 10105
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.317572076631177,
      "learning_rate": 1.8822922326774067e-06,
      "loss": 0.1655,
      "step": 10106
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2694553512991862,
      "learning_rate": 1.8813866871632836e-06,
      "loss": 0.1539,
      "step": 10107
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.685160417849182,
      "learning_rate": 1.8804813090439355e-06,
      "loss": 0.5128,
      "step": 10108
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1326918715153882,
      "learning_rate": 1.8795760983679556e-06,
      "loss": 0.1602,
      "step": 10109
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.348760452413663,
      "learning_rate": 1.8786710551839354e-06,
      "loss": 0.578,
      "step": 10110
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3409790320427564,
      "learning_rate": 1.8777661795404511e-06,
      "loss": 0.1728,
      "step": 10111
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5593283660153445,
      "learning_rate": 1.8768614714860744e-06,
      "loss": 0.1908,
      "step": 10112
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3625331474009448,
      "learning_rate": 1.8759569310693682e-06,
      "loss": 0.1762,
      "step": 10113
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1796226418325513,
      "learning_rate": 1.8750525583388806e-06,
      "loss": 0.1437,
      "step": 10114
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4031245734410154,
      "learning_rate": 1.8741483533431575e-06,
      "loss": 0.1767,
      "step": 10115
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3855889811739275,
      "learning_rate": 1.8732443161307323e-06,
      "loss": 0.181,
      "step": 10116
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4070907333663598,
      "learning_rate": 1.8723404467501316e-06,
      "loss": 0.2023,
      "step": 10117
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1809365915931191,
      "learning_rate": 1.8714367452498688e-06,
      "loss": 0.1865,
      "step": 10118
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2647378106793603,
      "learning_rate": 1.8705332116784541e-06,
      "loss": 0.1453,
      "step": 10119
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4861448590844877,
      "learning_rate": 1.869629846084382e-06,
      "loss": 0.2352,
      "step": 10120
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.308040265671357,
      "learning_rate": 1.868726648516145e-06,
      "loss": 0.1626,
      "step": 10121
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.356088441059376,
      "learning_rate": 1.8678236190222204e-06,
      "loss": 0.1749,
      "step": 10122
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.338235114393291,
      "learning_rate": 1.86692075765108e-06,
      "loss": 0.2018,
      "step": 10123
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.519494095366238,
      "learning_rate": 1.8660180644511866e-06,
      "loss": 0.1716,
      "step": 10124
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5659262640062315,
      "learning_rate": 1.865115539470994e-06,
      "loss": 0.2107,
      "step": 10125
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4422404911992923,
      "learning_rate": 1.8642131827589439e-06,
      "loss": 0.1924,
      "step": 10126
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.299048714053764,
      "learning_rate": 1.8633109943634742e-06,
      "loss": 0.1812,
      "step": 10127
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.132101581496522,
      "learning_rate": 1.8624089743330064e-06,
      "loss": 0.1539,
      "step": 10128
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2850524368248895,
      "learning_rate": 1.8615071227159625e-06,
      "loss": 0.1531,
      "step": 10129
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.958026559012103,
      "learning_rate": 1.8606054395607453e-06,
      "loss": 0.6237,
      "step": 10130
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3975829041797732,
      "learning_rate": 1.8597039249157561e-06,
      "loss": 0.171,
      "step": 10131
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1390250326226639,
      "learning_rate": 1.8588025788293845e-06,
      "loss": 0.1767,
      "step": 10132
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4798546512367434,
      "learning_rate": 1.857901401350013e-06,
      "loss": 0.2272,
      "step": 10133
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5335107269031858,
      "learning_rate": 1.8570003925260095e-06,
      "loss": 0.2172,
      "step": 10134
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.837374540889827,
      "learning_rate": 1.8560995524057402e-06,
      "loss": 0.5395,
      "step": 10135
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3124938832903987,
      "learning_rate": 1.8551988810375554e-06,
      "loss": 0.1732,
      "step": 10136
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3556871507220654,
      "learning_rate": 1.8542983784698026e-06,
      "loss": 0.1452,
      "step": 10137
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2348658665079344,
      "learning_rate": 1.8533980447508138e-06,
      "loss": 0.1554,
      "step": 10138
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.334170734017181,
      "learning_rate": 1.8524978799289172e-06,
      "loss": 0.1478,
      "step": 10139
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3286826535945606,
      "learning_rate": 1.8515978840524302e-06,
      "loss": 0.1476,
      "step": 10140
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3020618273552271,
      "learning_rate": 1.8506980571696626e-06,
      "loss": 0.1841,
      "step": 10141
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.307529982436838,
      "learning_rate": 1.8497983993289097e-06,
      "loss": 0.176,
      "step": 10142
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.431564165757675,
      "learning_rate": 1.8488989105784638e-06,
      "loss": 0.5799,
      "step": 10143
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.215077135855464,
      "learning_rate": 1.8479995909666077e-06,
      "loss": 0.1453,
      "step": 10144
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4488265682353514,
      "learning_rate": 1.8471004405416094e-06,
      "loss": 0.1952,
      "step": 10145
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.958197397085143,
      "learning_rate": 1.8462014593517352e-06,
      "loss": 0.4985,
      "step": 10146
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.281644466615377,
      "learning_rate": 1.8453026474452358e-06,
      "loss": 0.5803,
      "step": 10147
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4590389105898813,
      "learning_rate": 1.844404004870357e-06,
      "loss": 0.1759,
      "step": 10148
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.22702914860479,
      "learning_rate": 1.8435055316753354e-06,
      "loss": 0.1676,
      "step": 10149
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1970796784903657,
      "learning_rate": 1.842607227908398e-06,
      "loss": 0.1422,
      "step": 10150
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2685443051548926,
      "learning_rate": 1.841709093617759e-06,
      "loss": 0.1712,
      "step": 10151
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.179842531562926,
      "learning_rate": 1.8408111288516306e-06,
      "loss": 0.6358,
      "step": 10152
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3342834562183434,
      "learning_rate": 1.8399133336582087e-06,
      "loss": 0.1837,
      "step": 10153
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.320163091674366,
      "learning_rate": 1.8390157080856858e-06,
      "loss": 0.1897,
      "step": 10154
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2221635757058384,
      "learning_rate": 1.8381182521822406e-06,
      "loss": 0.1526,
      "step": 10155
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1959551568915936,
      "learning_rate": 1.8372209659960454e-06,
      "loss": 0.1647,
      "step": 10156
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.234257269349216,
      "learning_rate": 1.8363238495752638e-06,
      "loss": 0.1568,
      "step": 10157
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5872728174744868,
      "learning_rate": 1.8354269029680515e-06,
      "loss": 0.1989,
      "step": 10158
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.304962824728764,
      "learning_rate": 1.834530126222549e-06,
      "loss": 0.1979,
      "step": 10159
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6567830139336046,
      "learning_rate": 1.8336335193868955e-06,
      "loss": 0.1862,
      "step": 10160
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2959701347169124,
      "learning_rate": 1.8327370825092134e-06,
      "loss": 0.1549,
      "step": 10161
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3733814263508257,
      "learning_rate": 1.8318408156376239e-06,
      "loss": 0.1722,
      "step": 10162
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4055393765905368,
      "learning_rate": 1.8309447188202306e-06,
      "loss": 0.1734,
      "step": 10163
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.413165966634706,
      "learning_rate": 1.8300487921051352e-06,
      "loss": 0.1848,
      "step": 10164
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3056398235970395,
      "learning_rate": 1.8291530355404275e-06,
      "loss": 0.1628,
      "step": 10165
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.432863663775086,
      "learning_rate": 1.828257449174189e-06,
      "loss": 0.165,
      "step": 10166
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2300735644118348,
      "learning_rate": 1.8273620330544882e-06,
      "loss": 0.1602,
      "step": 10167
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3721071519293426,
      "learning_rate": 1.8264667872293913e-06,
      "loss": 0.1648,
      "step": 10168
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.562851287250568,
      "learning_rate": 1.8255717117469474e-06,
      "loss": 0.176,
      "step": 10169
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3343953322580107,
      "learning_rate": 1.8246768066552046e-06,
      "loss": 0.1549,
      "step": 10170
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.189925218306639,
      "learning_rate": 1.8237820720021938e-06,
      "loss": 0.1564,
      "step": 10171
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1885149890326123,
      "learning_rate": 1.8228875078359432e-06,
      "loss": 0.1554,
      "step": 10172
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.974332899028695,
      "learning_rate": 1.8219931142044694e-06,
      "loss": 0.5856,
      "step": 10173
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5225721656793263,
      "learning_rate": 1.8210988911557813e-06,
      "loss": 0.1968,
      "step": 10174
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3441867967910932,
      "learning_rate": 1.820204838737874e-06,
      "loss": 0.163,
      "step": 10175
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.677084440811978,
      "learning_rate": 1.8193109569987376e-06,
      "loss": 0.554,
      "step": 10176
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3507115870748871,
      "learning_rate": 1.8184172459863553e-06,
      "loss": 0.197,
      "step": 10177
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1846633423280752,
      "learning_rate": 1.8175237057486938e-06,
      "loss": 0.171,
      "step": 10178
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.429224370159057,
      "learning_rate": 1.8166303363337179e-06,
      "loss": 0.1381,
      "step": 10179
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.798969255098325,
      "learning_rate": 1.8157371377893769e-06,
      "loss": 0.5402,
      "step": 10180
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2308471350396217,
      "learning_rate": 1.814844110163616e-06,
      "loss": 0.1574,
      "step": 10181
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.274521083760135,
      "learning_rate": 1.8139512535043696e-06,
      "loss": 0.1383,
      "step": 10182
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3055894021065348,
      "learning_rate": 1.8130585678595635e-06,
      "loss": 0.1549,
      "step": 10183
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.731425241716922,
      "learning_rate": 1.8121660532771113e-06,
      "loss": 0.6181,
      "step": 10184
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.920393490046399,
      "learning_rate": 1.811273709804922e-06,
      "loss": 0.6822,
      "step": 10185
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.5731593412723806,
      "learning_rate": 1.8103815374908906e-06,
      "loss": 0.5495,
      "step": 10186
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3462488704074438,
      "learning_rate": 1.809489536382908e-06,
      "loss": 0.1576,
      "step": 10187
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.913074934248999,
      "learning_rate": 1.8085977065288502e-06,
      "loss": 0.5428,
      "step": 10188
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.243961251542498,
      "learning_rate": 1.8077060479765908e-06,
      "loss": 0.1733,
      "step": 10189
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1952351373615455,
      "learning_rate": 1.8068145607739845e-06,
      "loss": 0.1589,
      "step": 10190
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2572704555585423,
      "learning_rate": 1.8059232449688903e-06,
      "loss": 0.1781,
      "step": 10191
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.140822276025719,
      "learning_rate": 1.8050321006091453e-06,
      "loss": 0.6907,
      "step": 10192
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4172125173299168,
      "learning_rate": 1.8041411277425858e-06,
      "loss": 0.1829,
      "step": 10193
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2121386475466875,
      "learning_rate": 1.8032503264170326e-06,
      "loss": 0.162,
      "step": 10194
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4813177471663996,
      "learning_rate": 1.802359696680303e-06,
      "loss": 0.1564,
      "step": 10195
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2338853544864892,
      "learning_rate": 1.8014692385802002e-06,
      "loss": 0.12,
      "step": 10196
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6424419343561885,
      "learning_rate": 1.8005789521645233e-06,
      "loss": 0.1956,
      "step": 10197
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2682472111428265,
      "learning_rate": 1.7996888374810556e-06,
      "loss": 0.1623,
      "step": 10198
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.26763806701586,
      "learning_rate": 1.798798894577577e-06,
      "loss": 0.6783,
      "step": 10199
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.802546050912952,
      "learning_rate": 1.7979091235018564e-06,
      "loss": 0.685,
      "step": 10200
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2986642922285885,
      "learning_rate": 1.797019524301654e-06,
      "loss": 0.1834,
      "step": 10201
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2055159528192068,
      "learning_rate": 1.7961300970247175e-06,
      "loss": 0.1515,
      "step": 10202
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3422415577262934,
      "learning_rate": 1.7952408417187905e-06,
      "loss": 0.1693,
      "step": 10203
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.386323278877843,
      "learning_rate": 1.7943517584316017e-06,
      "loss": 0.5714,
      "step": 10204
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.270325730975223,
      "learning_rate": 1.7934628472108768e-06,
      "loss": 0.1443,
      "step": 10205
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2933545340037202,
      "learning_rate": 1.7925741081043258e-06,
      "loss": 0.1603,
      "step": 10206
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4005850176619432,
      "learning_rate": 1.791685541159654e-06,
      "loss": 0.1674,
      "step": 10207
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2570882752983858,
      "learning_rate": 1.7907971464245565e-06,
      "loss": 0.1758,
      "step": 10208
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3387111047330447,
      "learning_rate": 1.7899089239467189e-06,
      "loss": 0.1663,
      "step": 10209
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3072043788948535,
      "learning_rate": 1.789020873773819e-06,
      "loss": 0.1898,
      "step": 10210
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.2616081088417435,
      "learning_rate": 1.7881329959535198e-06,
      "loss": 0.6354,
      "step": 10211
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6794633571536743,
      "learning_rate": 1.7872452905334836e-06,
      "loss": 0.2101,
      "step": 10212
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5223137599746444,
      "learning_rate": 1.7863577575613545e-06,
      "loss": 0.1795,
      "step": 10213
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4134285695549014,
      "learning_rate": 1.785470397084776e-06,
      "loss": 0.1967,
      "step": 10214
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3932791744452067,
      "learning_rate": 1.7845832091513737e-06,
      "loss": 0.2277,
      "step": 10215
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3050702155085232,
      "learning_rate": 1.783696193808771e-06,
      "loss": 0.1538,
      "step": 10216
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4841508095381029,
      "learning_rate": 1.7828093511045785e-06,
      "loss": 0.2247,
      "step": 10217
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4203334846316993,
      "learning_rate": 1.781922681086401e-06,
      "loss": 0.1835,
      "step": 10218
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3346839146909562,
      "learning_rate": 1.781036183801827e-06,
      "loss": 0.1846,
      "step": 10219
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.022095595227784,
      "learning_rate": 1.7801498592984445e-06,
      "loss": 0.5531,
      "step": 10220
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.280632275722684,
      "learning_rate": 1.779263707623824e-06,
      "loss": 0.1779,
      "step": 10221
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4004560787761884,
      "learning_rate": 1.7783777288255339e-06,
      "loss": 0.1925,
      "step": 10222
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3471469258233124,
      "learning_rate": 1.7774919229511268e-06,
      "loss": 0.1818,
      "step": 10223
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2006745154325504,
      "learning_rate": 1.7766062900481513e-06,
      "loss": 0.1669,
      "step": 10224
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7076920140567544,
      "learning_rate": 1.7757208301641437e-06,
      "loss": 0.135,
      "step": 10225
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3702522174339256,
      "learning_rate": 1.7748355433466347e-06,
      "loss": 0.1951,
      "step": 10226
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3612757438023597,
      "learning_rate": 1.7739504296431387e-06,
      "loss": 0.2028,
      "step": 10227
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2369234177538446,
      "learning_rate": 1.7730654891011695e-06,
      "loss": 0.1586,
      "step": 10228
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2500904942204571,
      "learning_rate": 1.7721807217682229e-06,
      "loss": 0.1514,
      "step": 10229
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2483830189244156,
      "learning_rate": 1.7712961276917933e-06,
      "loss": 0.2105,
      "step": 10230
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4024952363947432,
      "learning_rate": 1.7704117069193593e-06,
      "loss": 0.1788,
      "step": 10231
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5028497593640122,
      "learning_rate": 1.7695274594983941e-06,
      "loss": 0.2083,
      "step": 10232
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2171607507142124,
      "learning_rate": 1.7686433854763608e-06,
      "loss": 0.153,
      "step": 10233
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3672689977834445,
      "learning_rate": 1.7677594849007152e-06,
      "loss": 0.1743,
      "step": 10234
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.459238715373597,
      "learning_rate": 1.7668757578188973e-06,
      "loss": 0.1956,
      "step": 10235
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2664668718024144,
      "learning_rate": 1.7659922042783463e-06,
      "loss": 0.2008,
      "step": 10236
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.317961079422185,
      "learning_rate": 1.7651088243264836e-06,
      "loss": 0.1668,
      "step": 10237
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3724015700051369,
      "learning_rate": 1.7642256180107293e-06,
      "loss": 0.2199,
      "step": 10238
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2756975090785245,
      "learning_rate": 1.7633425853784875e-06,
      "loss": 0.1761,
      "step": 10239
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4800122003649414,
      "learning_rate": 1.762459726477157e-06,
      "loss": 0.1714,
      "step": 10240
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6054926673612844,
      "learning_rate": 1.761577041354126e-06,
      "loss": 0.1955,
      "step": 10241
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1309432146348073,
      "learning_rate": 1.760694530056774e-06,
      "loss": 0.1327,
      "step": 10242
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3520163506884577,
      "learning_rate": 1.759812192632473e-06,
      "loss": 0.1822,
      "step": 10243
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.235679869079916,
      "learning_rate": 1.7589300291285783e-06,
      "loss": 0.1644,
      "step": 10244
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1850099330109805,
      "learning_rate": 1.7580480395924454e-06,
      "loss": 0.166,
      "step": 10245
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.609668807186098,
      "learning_rate": 1.7571662240714127e-06,
      "loss": 0.1914,
      "step": 10246
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.327687833021984,
      "learning_rate": 1.7562845826128156e-06,
      "loss": 0.1855,
      "step": 10247
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.295100459348844,
      "learning_rate": 1.755403115263974e-06,
      "loss": 0.1584,
      "step": 10248
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.555737462437905,
      "learning_rate": 1.7545218220722026e-06,
      "loss": 0.5094,
      "step": 10249
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4959608456843099,
      "learning_rate": 1.7536407030848068e-06,
      "loss": 0.1806,
      "step": 10250
    },
    {
      "epoch": 0.73,
      "grad_norm": 10.645441765357326,
      "learning_rate": 1.7527597583490825e-06,
      "loss": 0.5271,
      "step": 10251
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6885021420052595,
      "learning_rate": 1.7518789879123117e-06,
      "loss": 0.1845,
      "step": 10252
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3448838005997752,
      "learning_rate": 1.7509983918217742e-06,
      "loss": 0.1801,
      "step": 10253
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.617342288056361,
      "learning_rate": 1.7501179701247344e-06,
      "loss": 0.502,
      "step": 10254
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3615924050057526,
      "learning_rate": 1.7492377228684527e-06,
      "loss": 0.1553,
      "step": 10255
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3772740404287978,
      "learning_rate": 1.748357650100173e-06,
      "loss": 0.1886,
      "step": 10256
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4321862567055832,
      "learning_rate": 1.7474777518671371e-06,
      "loss": 0.1886,
      "step": 10257
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.964062069815653,
      "learning_rate": 1.7465980282165734e-06,
      "loss": 0.5338,
      "step": 10258
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1828580178038595,
      "learning_rate": 1.745718479195705e-06,
      "loss": 0.1659,
      "step": 10259
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3839481715607915,
      "learning_rate": 1.7448391048517378e-06,
      "loss": 0.1728,
      "step": 10260
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3523519249224243,
      "learning_rate": 1.7439599052318768e-06,
      "loss": 0.1995,
      "step": 10261
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3163517500543693,
      "learning_rate": 1.743080880383311e-06,
      "loss": 0.1745,
      "step": 10262
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3080790255940928,
      "learning_rate": 1.7422020303532267e-06,
      "loss": 0.1713,
      "step": 10263
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3830825654508476,
      "learning_rate": 1.7413233551887925e-06,
      "loss": 0.1707,
      "step": 10264
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1891848071236455,
      "learning_rate": 1.740444854937175e-06,
      "loss": 0.1425,
      "step": 10265
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.423678319557384,
      "learning_rate": 1.7395665296455283e-06,
      "loss": 0.3815,
      "step": 10266
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.327568538456378,
      "learning_rate": 1.7386883793609983e-06,
      "loss": 0.175,
      "step": 10267
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4586768676139563,
      "learning_rate": 1.7378104041307187e-06,
      "loss": 0.1935,
      "step": 10268
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3120126249001447,
      "learning_rate": 1.7369326040018175e-06,
      "loss": 0.2049,
      "step": 10269
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.362670309884819,
      "learning_rate": 1.7360549790214092e-06,
      "loss": 0.1769,
      "step": 10270
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4106546786951808,
      "learning_rate": 1.735177529236603e-06,
      "loss": 0.1479,
      "step": 10271
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1786536575392577,
      "learning_rate": 1.7343002546944981e-06,
      "loss": 0.1667,
      "step": 10272
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5386550224797546,
      "learning_rate": 1.7334231554421798e-06,
      "loss": 0.1896,
      "step": 10273
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.383199666625297,
      "learning_rate": 1.7325462315267294e-06,
      "loss": 0.179,
      "step": 10274
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3630335475403168,
      "learning_rate": 1.7316694829952164e-06,
      "loss": 0.1631,
      "step": 10275
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.068085931777952,
      "learning_rate": 1.7307929098947023e-06,
      "loss": 0.5857,
      "step": 10276
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3700467327454295,
      "learning_rate": 1.7299165122722355e-06,
      "loss": 0.1553,
      "step": 10277
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.356427422847231,
      "learning_rate": 1.729040290174861e-06,
      "loss": 0.5547,
      "step": 10278
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.443538100790687,
      "learning_rate": 1.7281642436496065e-06,
      "loss": 0.2025,
      "step": 10279
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3940254264241456,
      "learning_rate": 1.7272883727434996e-06,
      "loss": 0.1807,
      "step": 10280
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.367920886966645,
      "learning_rate": 1.7264126775035488e-06,
      "loss": 0.1897,
      "step": 10281
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1746517986915292,
      "learning_rate": 1.7255371579767605e-06,
      "loss": 0.1418,
      "step": 10282
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.155317162675417,
      "learning_rate": 1.7246618142101285e-06,
      "loss": 0.1505,
      "step": 10283
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1749151156217477,
      "learning_rate": 1.7237866462506398e-06,
      "loss": 0.1626,
      "step": 10284
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3900181844576949,
      "learning_rate": 1.7229116541452668e-06,
      "loss": 0.1701,
      "step": 10285
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.218431596613596,
      "learning_rate": 1.7220368379409785e-06,
      "loss": 0.1238,
      "step": 10286
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3487715981494197,
      "learning_rate": 1.7211621976847286e-06,
      "loss": 0.1929,
      "step": 10287
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1902952843571097,
      "learning_rate": 1.7202877334234675e-06,
      "loss": 0.1649,
      "step": 10288
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3341946151862414,
      "learning_rate": 1.7194134452041294e-06,
      "loss": 0.1798,
      "step": 10289
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3274603843341255,
      "learning_rate": 1.7185393330736461e-06,
      "loss": 0.1809,
      "step": 10290
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1435499605767787,
      "learning_rate": 1.7176653970789314e-06,
      "loss": 0.1289,
      "step": 10291
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3726085536415662,
      "learning_rate": 1.7167916372669018e-06,
      "loss": 0.1851,
      "step": 10292
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4476079097881336,
      "learning_rate": 1.7159180536844523e-06,
      "loss": 0.1861,
      "step": 10293
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4678839805969595,
      "learning_rate": 1.7150446463784753e-06,
      "loss": 0.1574,
      "step": 10294
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4512821041850994,
      "learning_rate": 1.7141714153958505e-06,
      "loss": 0.2025,
      "step": 10295
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2341635450104858,
      "learning_rate": 1.7132983607834514e-06,
      "loss": 0.1466,
      "step": 10296
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4207675079157591,
      "learning_rate": 1.7124254825881375e-06,
      "loss": 0.188,
      "step": 10297
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.336059133559886,
      "learning_rate": 1.7115527808567644e-06,
      "loss": 0.1707,
      "step": 10298
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1717707869613851,
      "learning_rate": 1.7106802556361724e-06,
      "loss": 0.1781,
      "step": 10299
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3610610741309637,
      "learning_rate": 1.709807906973196e-06,
      "loss": 0.1915,
      "step": 10300
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.311107582636797,
      "learning_rate": 1.7089357349146601e-06,
      "loss": 0.1548,
      "step": 10301
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2550265407593348,
      "learning_rate": 1.7080637395073813e-06,
      "loss": 0.1864,
      "step": 10302
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.608291989965493,
      "learning_rate": 1.7071919207981607e-06,
      "loss": 0.2147,
      "step": 10303
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2457981051884262,
      "learning_rate": 1.7063202788337963e-06,
      "loss": 0.187,
      "step": 10304
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6505542947461438,
      "learning_rate": 1.705448813661076e-06,
      "loss": 0.199,
      "step": 10305
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.402698715295302,
      "learning_rate": 1.7045775253267727e-06,
      "loss": 0.2253,
      "step": 10306
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3944299251041152,
      "learning_rate": 1.7037064138776576e-06,
      "loss": 0.1697,
      "step": 10307
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.320449494207843,
      "learning_rate": 1.7028354793604857e-06,
      "loss": 0.1978,
      "step": 10308
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.2040076986967385,
      "learning_rate": 1.7019647218220059e-06,
      "loss": 0.4548,
      "step": 10309
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0962414813264134,
      "learning_rate": 1.701094141308957e-06,
      "loss": 0.1384,
      "step": 10310
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4099362471718222,
      "learning_rate": 1.7002237378680708e-06,
      "loss": 0.1734,
      "step": 10311
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.208247842365986,
      "learning_rate": 1.6993535115460635e-06,
      "loss": 0.1575,
      "step": 10312
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3832252743388607,
      "learning_rate": 1.6984834623896483e-06,
      "loss": 0.1902,
      "step": 10313
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3786921913391152,
      "learning_rate": 1.6976135904455233e-06,
      "loss": 0.1759,
      "step": 10314
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4218006978677453,
      "learning_rate": 1.6967438957603826e-06,
      "loss": 0.1885,
      "step": 10315
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2412303311716457,
      "learning_rate": 1.6958743783809052e-06,
      "loss": 0.1555,
      "step": 10316
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3721081873400627,
      "learning_rate": 1.6950050383537637e-06,
      "loss": 0.189,
      "step": 10317
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2036364596083167,
      "learning_rate": 1.694135875725622e-06,
      "loss": 0.1299,
      "step": 10318
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.555898266235163,
      "learning_rate": 1.6932668905431348e-06,
      "loss": 0.2032,
      "step": 10319
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3598769649801314,
      "learning_rate": 1.6923980828529424e-06,
      "loss": 0.1924,
      "step": 10320
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2599647475452873,
      "learning_rate": 1.6915294527016818e-06,
      "loss": 0.1445,
      "step": 10321
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.44036931483922,
      "learning_rate": 1.6906610001359752e-06,
      "loss": 0.1777,
      "step": 10322
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4131806100990976,
      "learning_rate": 1.68979272520244e-06,
      "loss": 0.1599,
      "step": 10323
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3798080389938432,
      "learning_rate": 1.688924627947679e-06,
      "loss": 0.2041,
      "step": 10324
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.253565151013493,
      "learning_rate": 1.6880567084182897e-06,
      "loss": 0.167,
      "step": 10325
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.329094527169436,
      "learning_rate": 1.6871889666608588e-06,
      "loss": 0.1661,
      "step": 10326
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3676643715970154,
      "learning_rate": 1.686321402721965e-06,
      "loss": 0.1678,
      "step": 10327
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.270827932603055,
      "learning_rate": 1.6854540166481715e-06,
      "loss": 0.1737,
      "step": 10328
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3367757327490406,
      "learning_rate": 1.6845868084860406e-06,
      "loss": 0.2021,
      "step": 10329
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4222947180171932,
      "learning_rate": 1.6837197782821162e-06,
      "loss": 0.1991,
      "step": 10330
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3507261696911916,
      "learning_rate": 1.6828529260829412e-06,
      "loss": 0.1653,
      "step": 10331
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3268775317838808,
      "learning_rate": 1.681986251935041e-06,
      "loss": 0.2294,
      "step": 10332
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2657716807797372,
      "learning_rate": 1.6811197558849373e-06,
      "loss": 0.1466,
      "step": 10333
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2538562372520918,
      "learning_rate": 1.6802534379791396e-06,
      "loss": 0.162,
      "step": 10334
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2921727358750732,
      "learning_rate": 1.679387298264149e-06,
      "loss": 0.1629,
      "step": 10335
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3598057348711288,
      "learning_rate": 1.6785213367864572e-06,
      "loss": 0.1929,
      "step": 10336
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4359303726512576,
      "learning_rate": 1.6776555535925432e-06,
      "loss": 0.1891,
      "step": 10337
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3769709236279324,
      "learning_rate": 1.676789948728882e-06,
      "loss": 0.1732,
      "step": 10338
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.2005371134326,
      "learning_rate": 1.6759245222419324e-06,
      "loss": 0.4455,
      "step": 10339
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3280111772057075,
      "learning_rate": 1.6750592741781496e-06,
      "loss": 0.1836,
      "step": 10340
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1821331067811789,
      "learning_rate": 1.6741942045839743e-06,
      "loss": 0.1546,
      "step": 10341
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2079398834993706,
      "learning_rate": 1.6733293135058416e-06,
      "loss": 0.1673,
      "step": 10342
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4146461877872094,
      "learning_rate": 1.6724646009901752e-06,
      "loss": 0.1928,
      "step": 10343
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.211824029428611,
      "learning_rate": 1.6716000670833903e-06,
      "loss": 0.1573,
      "step": 10344
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4332298482835029,
      "learning_rate": 1.6707357118318901e-06,
      "loss": 0.184,
      "step": 10345
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.473332589404938,
      "learning_rate": 1.6698715352820716e-06,
      "loss": 0.2126,
      "step": 10346
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5675267670883708,
      "learning_rate": 1.6690075374803178e-06,
      "loss": 0.1753,
      "step": 10347
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5720496604507306,
      "learning_rate": 1.6681437184730077e-06,
      "loss": 0.1912,
      "step": 10348
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3617973574152942,
      "learning_rate": 1.667280078306504e-06,
      "loss": 0.149,
      "step": 10349
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2774610620335403,
      "learning_rate": 1.6664166170271657e-06,
      "loss": 0.152,
      "step": 10350
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.489654349738053,
      "learning_rate": 1.6655533346813397e-06,
      "loss": 0.5743,
      "step": 10351
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6098959885164814,
      "learning_rate": 1.6646902313153651e-06,
      "loss": 0.2221,
      "step": 10352
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2932019864784865,
      "learning_rate": 1.6638273069755672e-06,
      "loss": 0.2279,
      "step": 10353
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.820790118026856,
      "learning_rate": 1.6629645617082668e-06,
      "loss": 0.6432,
      "step": 10354
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4448411440971078,
      "learning_rate": 1.66210199555977e-06,
      "loss": 0.1692,
      "step": 10355
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.362823644086111,
      "learning_rate": 1.6612396085763794e-06,
      "loss": 0.2045,
      "step": 10356
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.274987769968265,
      "learning_rate": 1.66037740080438e-06,
      "loss": 0.1654,
      "step": 10357
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3074232325559954,
      "learning_rate": 1.6595153722900542e-06,
      "loss": 0.1772,
      "step": 10358
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.994467196821839,
      "learning_rate": 1.6586535230796725e-06,
      "loss": 0.5328,
      "step": 10359
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3008356056869474,
      "learning_rate": 1.657791853219497e-06,
      "loss": 0.1399,
      "step": 10360
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4040576030812788,
      "learning_rate": 1.6569303627557749e-06,
      "loss": 0.195,
      "step": 10361
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3179984547509473,
      "learning_rate": 1.656069051734751e-06,
      "loss": 0.1665,
      "step": 10362
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.33828010953948,
      "learning_rate": 1.655207920202655e-06,
      "loss": 0.1868,
      "step": 10363
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5160046960303568,
      "learning_rate": 1.6543469682057105e-06,
      "loss": 0.185,
      "step": 10364
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.638459050556768,
      "learning_rate": 1.6534861957901277e-06,
      "loss": 0.74,
      "step": 10365
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2348844798402445,
      "learning_rate": 1.6526256030021109e-06,
      "loss": 0.1411,
      "step": 10366
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.374223300756354,
      "learning_rate": 1.6517651898878534e-06,
      "loss": 0.209,
      "step": 10367
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.261322820189383,
      "learning_rate": 1.6509049564935387e-06,
      "loss": 0.1642,
      "step": 10368
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.642022633800516,
      "learning_rate": 1.6500449028653425e-06,
      "loss": 0.4681,
      "step": 10369
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.258272432285171,
      "learning_rate": 1.6491850290494255e-06,
      "loss": 0.705,
      "step": 10370
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.424837576465882,
      "learning_rate": 1.6483253350919453e-06,
      "loss": 0.1635,
      "step": 10371
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.793462889141143,
      "learning_rate": 1.6474658210390443e-06,
      "loss": 0.5877,
      "step": 10372
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4163291508486673,
      "learning_rate": 1.6466064869368609e-06,
      "loss": 0.173,
      "step": 10373
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4869653565432819,
      "learning_rate": 1.6457473328315171e-06,
      "loss": 0.1665,
      "step": 10374
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.330769568187334,
      "learning_rate": 1.6448883587691311e-06,
      "loss": 0.1768,
      "step": 10375
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2075519557305774,
      "learning_rate": 1.6440295647958081e-06,
      "loss": 0.175,
      "step": 10376
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.599522185607976,
      "learning_rate": 1.6431709509576477e-06,
      "loss": 0.5564,
      "step": 10377
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2939894682175557,
      "learning_rate": 1.6423125173007332e-06,
      "loss": 0.2022,
      "step": 10378
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2238600316114605,
      "learning_rate": 1.6414542638711445e-06,
      "loss": 0.1737,
      "step": 10379
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2107266696746273,
      "learning_rate": 1.640596190714947e-06,
      "loss": 0.1615,
      "step": 10380
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3345972625107967,
      "learning_rate": 1.639738297878201e-06,
      "loss": 0.177,
      "step": 10381
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.981025231046975,
      "learning_rate": 1.6388805854069522e-06,
      "loss": 0.7038,
      "step": 10382
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4495144098321755,
      "learning_rate": 1.6380230533472402e-06,
      "loss": 0.1616,
      "step": 10383
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1676309135795562,
      "learning_rate": 1.6371657017450948e-06,
      "loss": 0.1523,
      "step": 10384
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.23665370651643,
      "learning_rate": 1.636308530646536e-06,
      "loss": 0.4602,
      "step": 10385
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.075335500028317,
      "learning_rate": 1.6354515400975702e-06,
      "loss": 0.6194,
      "step": 10386
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.264501645165069,
      "learning_rate": 1.6345947301442011e-06,
      "loss": 0.1777,
      "step": 10387
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3274838100671293,
      "learning_rate": 1.633738100832415e-06,
      "loss": 0.1549,
      "step": 10388
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.229808024579589,
      "learning_rate": 1.6328816522081958e-06,
      "loss": 0.1597,
      "step": 10389
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2499662371302345,
      "learning_rate": 1.632025384317511e-06,
      "loss": 0.1423,
      "step": 10390
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.275561075677841,
      "learning_rate": 1.6311692972063254e-06,
      "loss": 0.1485,
      "step": 10391
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3483622186384292,
      "learning_rate": 1.6303133909205859e-06,
      "loss": 0.158,
      "step": 10392
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.173664842272985,
      "learning_rate": 1.6294576655062372e-06,
      "loss": 0.536,
      "step": 10393
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.140890022534197,
      "learning_rate": 1.6286021210092101e-06,
      "loss": 0.1635,
      "step": 10394
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.48083233380804,
      "learning_rate": 1.627746757475429e-06,
      "loss": 0.2016,
      "step": 10395
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4528764647578005,
      "learning_rate": 1.6268915749508036e-06,
      "loss": 0.1957,
      "step": 10396
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4743110767665601,
      "learning_rate": 1.626036573481239e-06,
      "loss": 0.2009,
      "step": 10397
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2925835214429204,
      "learning_rate": 1.6251817531126252e-06,
      "loss": 0.14,
      "step": 10398
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2749657033878894,
      "learning_rate": 1.6243271138908479e-06,
      "loss": 0.1618,
      "step": 10399
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4880029236402061,
      "learning_rate": 1.623472655861782e-06,
      "loss": 0.2059,
      "step": 10400
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4890311812433514,
      "learning_rate": 1.6226183790712874e-06,
      "loss": 0.2206,
      "step": 10401
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.407354880385227,
      "learning_rate": 1.6217642835652215e-06,
      "loss": 0.1858,
      "step": 10402
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2731865191454874,
      "learning_rate": 1.6209103693894275e-06,
      "loss": 0.1547,
      "step": 10403
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1739939690333454,
      "learning_rate": 1.6200566365897414e-06,
      "loss": 0.1573,
      "step": 10404
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2864779234869765,
      "learning_rate": 1.619203085211986e-06,
      "loss": 0.1636,
      "step": 10405
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3716415650105631,
      "learning_rate": 1.61834971530198e-06,
      "loss": 0.2044,
      "step": 10406
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1904280828200353,
      "learning_rate": 1.6174965269055242e-06,
      "loss": 0.1591,
      "step": 10407
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1772613245249952,
      "learning_rate": 1.6166435200684193e-06,
      "loss": 0.1407,
      "step": 10408
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.381330590843818,
      "learning_rate": 1.6157906948364471e-06,
      "loss": 0.2043,
      "step": 10409
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2767343679928234,
      "learning_rate": 1.6149380512553854e-06,
      "loss": 0.1561,
      "step": 10410
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.272197621078242,
      "learning_rate": 1.6140855893710005e-06,
      "loss": 0.1446,
      "step": 10411
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3346564994831331,
      "learning_rate": 1.613233309229052e-06,
      "loss": 0.1729,
      "step": 10412
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2638352032189408,
      "learning_rate": 1.6123812108752828e-06,
      "loss": 0.1919,
      "step": 10413
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2074662517025514,
      "learning_rate": 1.6115292943554334e-06,
      "loss": 0.1455,
      "step": 10414
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.265967151498462,
      "learning_rate": 1.610677559715228e-06,
      "loss": 0.1365,
      "step": 10415
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.7277266145406465,
      "learning_rate": 1.6098260070003884e-06,
      "loss": 0.5701,
      "step": 10416
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3827494922084411,
      "learning_rate": 1.6089746362566182e-06,
      "loss": 0.1782,
      "step": 10417
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2560205681383758,
      "learning_rate": 1.6081234475296177e-06,
      "loss": 0.1853,
      "step": 10418
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1818173436187212,
      "learning_rate": 1.6072724408650752e-06,
      "loss": 0.1345,
      "step": 10419
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3522643277313509,
      "learning_rate": 1.6064216163086716e-06,
      "loss": 0.1495,
      "step": 10420
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6086104272864234,
      "learning_rate": 1.6055709739060715e-06,
      "loss": 0.2019,
      "step": 10421
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2061845364093557,
      "learning_rate": 1.604720513702938e-06,
      "loss": 0.1495,
      "step": 10422
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2344255580758596,
      "learning_rate": 1.6038702357449171e-06,
      "loss": 0.1545,
      "step": 10423
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.383317539515382,
      "learning_rate": 1.6030201400776513e-06,
      "loss": 0.1577,
      "step": 10424
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.422823544165772,
      "learning_rate": 1.602170226746767e-06,
      "loss": 0.5171,
      "step": 10425
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1964492058917868,
      "learning_rate": 1.6013204957978868e-06,
      "loss": 0.1762,
      "step": 10426
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2923217032177687,
      "learning_rate": 1.6004709472766195e-06,
      "loss": 0.1564,
      "step": 10427
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.388750144845829,
      "learning_rate": 1.5996215812285682e-06,
      "loss": 0.1675,
      "step": 10428
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.482674110094329,
      "learning_rate": 1.598772397699319e-06,
      "loss": 0.1701,
      "step": 10429
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2153599147872247,
      "learning_rate": 1.5979233967344575e-06,
      "loss": 0.1948,
      "step": 10430
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5015225829071586,
      "learning_rate": 1.5970745783795504e-06,
      "loss": 0.1913,
      "step": 10431
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.260102406718699,
      "learning_rate": 1.5962259426801613e-06,
      "loss": 0.1497,
      "step": 10432
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3919105355351218,
      "learning_rate": 1.5953774896818424e-06,
      "loss": 0.1771,
      "step": 10433
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.3817853805002125,
      "learning_rate": 1.5945292194301326e-06,
      "loss": 0.5207,
      "step": 10434
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.309524487955066,
      "learning_rate": 1.5936811319705648e-06,
      "loss": 0.1847,
      "step": 10435
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4364693382993523,
      "learning_rate": 1.5928332273486614e-06,
      "loss": 0.2253,
      "step": 10436
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4914262836960828,
      "learning_rate": 1.591985505609936e-06,
      "loss": 0.1748,
      "step": 10437
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3144699714126866,
      "learning_rate": 1.5911379667998877e-06,
      "loss": 0.1688,
      "step": 10438
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.863829595753437,
      "learning_rate": 1.590290610964012e-06,
      "loss": 0.4636,
      "step": 10439
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4717114721475364,
      "learning_rate": 1.589443438147789e-06,
      "loss": 0.1958,
      "step": 10440
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3174063206379354,
      "learning_rate": 1.5885964483966943e-06,
      "loss": 0.1531,
      "step": 10441
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3179528723446643,
      "learning_rate": 1.5877496417561876e-06,
      "loss": 0.1474,
      "step": 10442
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5442161100087044,
      "learning_rate": 1.5869030182717243e-06,
      "loss": 0.1988,
      "step": 10443
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.037153163028543,
      "learning_rate": 1.5860565779887478e-06,
      "loss": 0.6036,
      "step": 10444
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1890610143495717,
      "learning_rate": 1.585210320952692e-06,
      "loss": 0.153,
      "step": 10445
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2866077585034341,
      "learning_rate": 1.5843642472089794e-06,
      "loss": 0.1791,
      "step": 10446
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1838220178068828,
      "learning_rate": 1.5835183568030254e-06,
      "loss": 0.1586,
      "step": 10447
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2007826506998083,
      "learning_rate": 1.5826726497802314e-06,
      "loss": 0.1601,
      "step": 10448
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3296838973276808,
      "learning_rate": 1.5818271261859946e-06,
      "loss": 0.1498,
      "step": 10449
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4753703581185218,
      "learning_rate": 1.5809817860656967e-06,
      "loss": 0.1605,
      "step": 10450
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.542361186576464,
      "learning_rate": 1.5801366294647135e-06,
      "loss": 0.4413,
      "step": 10451
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2720675829811852,
      "learning_rate": 1.57929165642841e-06,
      "loss": 0.1525,
      "step": 10452
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.493673222914656,
      "learning_rate": 1.5784468670021418e-06,
      "loss": 0.1571,
      "step": 10453
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.923016990517658,
      "learning_rate": 1.5776022612312514e-06,
      "loss": 0.4507,
      "step": 10454
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4485115577611092,
      "learning_rate": 1.576757839161076e-06,
      "loss": 0.1601,
      "step": 10455
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3753279095893713,
      "learning_rate": 1.575913600836939e-06,
      "loss": 0.1764,
      "step": 10456
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3441596059211236,
      "learning_rate": 1.5750695463041581e-06,
      "loss": 0.1552,
      "step": 10457
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3578550107654128,
      "learning_rate": 1.574225675608036e-06,
      "loss": 0.1516,
      "step": 10458
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2782773816434025,
      "learning_rate": 1.5733819887938701e-06,
      "loss": 0.1706,
      "step": 10459
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.270020650996983,
      "learning_rate": 1.5725384859069454e-06,
      "loss": 0.151,
      "step": 10460
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4095682848122812,
      "learning_rate": 1.5716951669925396e-06,
      "loss": 0.1764,
      "step": 10461
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2710444153630065,
      "learning_rate": 1.5708520320959158e-06,
      "loss": 0.1885,
      "step": 10462
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5947326455532704,
      "learning_rate": 1.570009081262332e-06,
      "loss": 0.1948,
      "step": 10463
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4312996145327896,
      "learning_rate": 1.5691663145370351e-06,
      "loss": 0.213,
      "step": 10464
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2940615763119767,
      "learning_rate": 1.5683237319652594e-06,
      "loss": 0.1972,
      "step": 10465
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4541562881732073,
      "learning_rate": 1.567481333592234e-06,
      "loss": 0.1939,
      "step": 10466
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4593776333657507,
      "learning_rate": 1.566639119463172e-06,
      "loss": 0.1807,
      "step": 10467
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.2427895766204,
      "learning_rate": 1.5657970896232822e-06,
      "loss": 0.5081,
      "step": 10468
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2644237451651799,
      "learning_rate": 1.5649552441177612e-06,
      "loss": 0.1191,
      "step": 10469
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.37473850002062,
      "learning_rate": 1.5641135829917976e-06,
      "loss": 0.1958,
      "step": 10470
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4481417203664921,
      "learning_rate": 1.5632721062905653e-06,
      "loss": 0.2231,
      "step": 10471
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3610888233221126,
      "learning_rate": 1.5624308140592347e-06,
      "loss": 0.1284,
      "step": 10472
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2721491562571627,
      "learning_rate": 1.5615897063429591e-06,
      "loss": 0.1354,
      "step": 10473
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4442291585209646,
      "learning_rate": 1.56074878318689e-06,
      "loss": 0.1504,
      "step": 10474
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5515207498420296,
      "learning_rate": 1.5599080446361614e-06,
      "loss": 0.1822,
      "step": 10475
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3109183797055048,
      "learning_rate": 1.559067490735902e-06,
      "loss": 0.1728,
      "step": 10476
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2683373792990822,
      "learning_rate": 1.5582271215312294e-06,
      "loss": 0.154,
      "step": 10477
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.618002043239327,
      "learning_rate": 1.5573869370672534e-06,
      "loss": 0.6047,
      "step": 10478
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4837638178701218,
      "learning_rate": 1.556546937389068e-06,
      "loss": 0.1554,
      "step": 10479
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3536498608784664,
      "learning_rate": 1.5557071225417648e-06,
      "loss": 0.1473,
      "step": 10480
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4980320228970405,
      "learning_rate": 1.5548674925704178e-06,
      "loss": 0.1948,
      "step": 10481
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.439382566889818,
      "learning_rate": 1.5540280475200986e-06,
      "loss": 0.172,
      "step": 10482
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2071957810521379,
      "learning_rate": 1.5531887874358625e-06,
      "loss": 0.1409,
      "step": 10483
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3297630809915868,
      "learning_rate": 1.5523497123627584e-06,
      "loss": 0.1879,
      "step": 10484
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.760363322440483,
      "learning_rate": 1.5515108223458253e-06,
      "loss": 0.4804,
      "step": 10485
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3020760278978654,
      "learning_rate": 1.5506721174300926e-06,
      "loss": 0.1717,
      "step": 10486
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.050288118512901,
      "learning_rate": 1.5498335976605756e-06,
      "loss": 0.6893,
      "step": 10487
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2991265476024951,
      "learning_rate": 1.5489952630822858e-06,
      "loss": 0.1798,
      "step": 10488
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1791477022383148,
      "learning_rate": 1.5481571137402185e-06,
      "loss": 0.1594,
      "step": 10489
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2620432561465613,
      "learning_rate": 1.547319149679366e-06,
      "loss": 0.1463,
      "step": 10490
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.415792197423433,
      "learning_rate": 1.5464813709447023e-06,
      "loss": 0.1966,
      "step": 10491
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3312179585275428,
      "learning_rate": 1.5456437775812011e-06,
      "loss": 0.1569,
      "step": 10492
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.476917509438233,
      "learning_rate": 1.544806369633816e-06,
      "loss": 0.2036,
      "step": 10493
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.342948707354769,
      "learning_rate": 1.5439691471474989e-06,
      "loss": 0.1675,
      "step": 10494
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3633698369517413,
      "learning_rate": 1.5431321101671875e-06,
      "loss": 0.1877,
      "step": 10495
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.411179893935012,
      "learning_rate": 1.5422952587378114e-06,
      "loss": 0.1906,
      "step": 10496
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3983821057002455,
      "learning_rate": 1.5414585929042902e-06,
      "loss": 0.1592,
      "step": 10497
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5457220171447756,
      "learning_rate": 1.5406221127115306e-06,
      "loss": 0.1709,
      "step": 10498
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.59649189888967,
      "learning_rate": 1.539785818204434e-06,
      "loss": 0.188,
      "step": 10499
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3857419513300129,
      "learning_rate": 1.5389497094278861e-06,
      "loss": 0.1785,
      "step": 10500
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.453495273364983,
      "learning_rate": 1.5381137864267693e-06,
      "loss": 0.189,
      "step": 10501
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3588141733003223,
      "learning_rate": 1.53727804924595e-06,
      "loss": 0.202,
      "step": 10502
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2225155290765854,
      "learning_rate": 1.5364424979302877e-06,
      "loss": 0.1453,
      "step": 10503
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4257958323614714,
      "learning_rate": 1.535607132524633e-06,
      "loss": 0.1739,
      "step": 10504
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8002157627992004,
      "learning_rate": 1.5347719530738254e-06,
      "loss": 0.2025,
      "step": 10505
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1187456986922084,
      "learning_rate": 1.5339369596226905e-06,
      "loss": 0.1709,
      "step": 10506
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.092893006757683,
      "learning_rate": 1.5331021522160522e-06,
      "loss": 0.1246,
      "step": 10507
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3410714646742177,
      "learning_rate": 1.532267530898715e-06,
      "loss": 0.1605,
      "step": 10508
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.264595904732522,
      "learning_rate": 1.5314330957154817e-06,
      "loss": 0.1605,
      "step": 10509
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.262591230676526,
      "learning_rate": 1.5305988467111388e-06,
      "loss": 0.1793,
      "step": 10510
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4103210248021236,
      "learning_rate": 1.5297647839304663e-06,
      "loss": 0.172,
      "step": 10511
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3350509223522926,
      "learning_rate": 1.5289309074182334e-06,
      "loss": 0.1702,
      "step": 10512
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4154796423185076,
      "learning_rate": 1.5280972172192015e-06,
      "loss": 0.1882,
      "step": 10513
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.650132589344743,
      "learning_rate": 1.5272637133781159e-06,
      "loss": 0.6434,
      "step": 10514
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3873754900927415,
      "learning_rate": 1.5264303959397197e-06,
      "loss": 0.1687,
      "step": 10515
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.190551666666915,
      "learning_rate": 1.5255972649487388e-06,
      "loss": 0.1408,
      "step": 10516
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1721007522357072,
      "learning_rate": 1.5247643204498947e-06,
      "loss": 0.1942,
      "step": 10517
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3432918007970127,
      "learning_rate": 1.5239315624878942e-06,
      "loss": 0.1623,
      "step": 10518
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3498411248222175,
      "learning_rate": 1.5230989911074374e-06,
      "loss": 0.1597,
      "step": 10519
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3232695452513834,
      "learning_rate": 1.5222666063532138e-06,
      "loss": 0.185,
      "step": 10520
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3858168196914384,
      "learning_rate": 1.5214344082699045e-06,
      "loss": 0.1722,
      "step": 10521
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4316784861820746,
      "learning_rate": 1.5206023969021743e-06,
      "loss": 0.1655,
      "step": 10522
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2812976406998857,
      "learning_rate": 1.5197705722946866e-06,
      "loss": 0.1169,
      "step": 10523
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.812609794364607,
      "learning_rate": 1.518938934492087e-06,
      "loss": 0.5593,
      "step": 10524
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.549235289607002,
      "learning_rate": 1.5181074835390169e-06,
      "loss": 0.1642,
      "step": 10525
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.605472880724829,
      "learning_rate": 1.5172762194801028e-06,
      "loss": 0.6602,
      "step": 10526
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.831231461324541,
      "learning_rate": 1.5164451423599651e-06,
      "loss": 0.6236,
      "step": 10527
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5725697854023923,
      "learning_rate": 1.515614252223213e-06,
      "loss": 0.158,
      "step": 10528
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2503000998528866,
      "learning_rate": 1.5147835491144448e-06,
      "loss": 0.1592,
      "step": 10529
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.370360427676737,
      "learning_rate": 1.5139530330782515e-06,
      "loss": 0.1722,
      "step": 10530
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2886671493011164,
      "learning_rate": 1.5131227041592083e-06,
      "loss": 0.1418,
      "step": 10531
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3157957867042849,
      "learning_rate": 1.5122925624018875e-06,
      "loss": 0.1596,
      "step": 10532
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4251297502743585,
      "learning_rate": 1.5114626078508442e-06,
      "loss": 0.1993,
      "step": 10533
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1608913134696162,
      "learning_rate": 1.5106328405506304e-06,
      "loss": 0.1451,
      "step": 10534
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.46126676309415,
      "learning_rate": 1.5098032605457818e-06,
      "loss": 0.1524,
      "step": 10535
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5662206507480947,
      "learning_rate": 1.5089738678808285e-06,
      "loss": 0.1986,
      "step": 10536
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2158037939564321,
      "learning_rate": 1.5081446626002889e-06,
      "loss": 0.1318,
      "step": 10537
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6869950033723824,
      "learning_rate": 1.5073156447486725e-06,
      "loss": 0.194,
      "step": 10538
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1966204025155878,
      "learning_rate": 1.5064868143704752e-06,
      "loss": 0.1765,
      "step": 10539
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.45471694700186,
      "learning_rate": 1.5056581715101887e-06,
      "loss": 0.2049,
      "step": 10540
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3567814666740656,
      "learning_rate": 1.5048297162122872e-06,
      "loss": 0.1567,
      "step": 10541
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3143219045746624,
      "learning_rate": 1.5040014485212428e-06,
      "loss": 0.1848,
      "step": 10542
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4288827634511303,
      "learning_rate": 1.5031733684815103e-06,
      "loss": 0.1936,
      "step": 10543
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.330689146820718,
      "learning_rate": 1.5023454761375389e-06,
      "loss": 0.188,
      "step": 10544
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5496218307316194,
      "learning_rate": 1.501517771533767e-06,
      "loss": 0.1906,
      "step": 10545
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.712509488714706,
      "learning_rate": 1.500690254714624e-06,
      "loss": 0.6921,
      "step": 10546
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5104770457518657,
      "learning_rate": 1.4998629257245246e-06,
      "loss": 0.1742,
      "step": 10547
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.213206683677835,
      "learning_rate": 1.4990357846078795e-06,
      "loss": 0.1569,
      "step": 10548
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3278586200749611,
      "learning_rate": 1.498208831409083e-06,
      "loss": 0.1623,
      "step": 10549
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.374103606124417,
      "learning_rate": 1.4973820661725269e-06,
      "loss": 0.5857,
      "step": 10550
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3249415078196536,
      "learning_rate": 1.4965554889425843e-06,
      "loss": 0.1547,
      "step": 10551
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4391646263227567,
      "learning_rate": 1.4957290997636247e-06,
      "loss": 0.2013,
      "step": 10552
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2289012474443257,
      "learning_rate": 1.4949028986800053e-06,
      "loss": 0.1462,
      "step": 10553
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4348550957699242,
      "learning_rate": 1.4940768857360743e-06,
      "loss": 0.1647,
      "step": 10554
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.567119717207712,
      "learning_rate": 1.4932510609761668e-06,
      "loss": 0.2338,
      "step": 10555
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4374220379260532,
      "learning_rate": 1.4924254244446123e-06,
      "loss": 0.1661,
      "step": 10556
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2225683188446723,
      "learning_rate": 1.4915999761857246e-06,
      "loss": 0.1622,
      "step": 10557
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.207388201889036,
      "learning_rate": 1.4907747162438135e-06,
      "loss": 0.1722,
      "step": 10558
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3134720054135238,
      "learning_rate": 1.4899496446631728e-06,
      "loss": 0.1469,
      "step": 10559
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3548453884450626,
      "learning_rate": 1.4891247614880904e-06,
      "loss": 0.1413,
      "step": 10560
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.246687918094449,
      "learning_rate": 1.4883000667628428e-06,
      "loss": 0.197,
      "step": 10561
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5803243777354625,
      "learning_rate": 1.4874755605316966e-06,
      "loss": 0.1817,
      "step": 10562
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5371401992368383,
      "learning_rate": 1.486651242838909e-06,
      "loss": 0.1596,
      "step": 10563
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3902733442039896,
      "learning_rate": 1.4858271137287234e-06,
      "loss": 0.1818,
      "step": 10564
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3611509376914028,
      "learning_rate": 1.485003173245379e-06,
      "loss": 0.1703,
      "step": 10565
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3726066295824362,
      "learning_rate": 1.4841794214330984e-06,
      "loss": 0.1546,
      "step": 10566
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2509577715646216,
      "learning_rate": 1.4833558583361002e-06,
      "loss": 0.1719,
      "step": 10567
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3060200520189311,
      "learning_rate": 1.4825324839985872e-06,
      "loss": 0.1803,
      "step": 10568
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1927775072944458,
      "learning_rate": 1.4817092984647564e-06,
      "loss": 0.1313,
      "step": 10569
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.391061581669175,
      "learning_rate": 1.4808863017787928e-06,
      "loss": 0.1509,
      "step": 10570
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1670381189334535,
      "learning_rate": 1.4800634939848734e-06,
      "loss": 0.1616,
      "step": 10571
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3584365611697602,
      "learning_rate": 1.4792408751271603e-06,
      "loss": 0.1499,
      "step": 10572
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2825827495983173,
      "learning_rate": 1.4784184452498112e-06,
      "loss": 0.1795,
      "step": 10573
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.316344244296368,
      "learning_rate": 1.4775962043969678e-06,
      "loss": 0.4989,
      "step": 10574
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.370991733439597,
      "learning_rate": 1.4767741526127682e-06,
      "loss": 0.6052,
      "step": 10575
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.636132891095267,
      "learning_rate": 1.475952289941333e-06,
      "loss": 0.4593,
      "step": 10576
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3060210390871034,
      "learning_rate": 1.4751306164267792e-06,
      "loss": 0.1702,
      "step": 10577
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3664071629583912,
      "learning_rate": 1.4743091321132103e-06,
      "loss": 0.1965,
      "step": 10578
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1782094955392424,
      "learning_rate": 1.4734878370447214e-06,
      "loss": 0.122,
      "step": 10579
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2213035165067083,
      "learning_rate": 1.472666731265394e-06,
      "loss": 0.1713,
      "step": 10580
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4007137884037673,
      "learning_rate": 1.471845814819305e-06,
      "loss": 0.1357,
      "step": 10581
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.398630310099093,
      "learning_rate": 1.471025087750514e-06,
      "loss": 0.1759,
      "step": 10582
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.374612275955994,
      "learning_rate": 1.470204550103078e-06,
      "loss": 0.4612,
      "step": 10583
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3737434647516604,
      "learning_rate": 1.4693842019210374e-06,
      "loss": 0.1987,
      "step": 10584
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4919742294220693,
      "learning_rate": 1.4685640432484283e-06,
      "loss": 0.179,
      "step": 10585
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2144052809144879,
      "learning_rate": 1.4677440741292682e-06,
      "loss": 0.1604,
      "step": 10586
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.230132487344419,
      "learning_rate": 1.4669242946075762e-06,
      "loss": 0.1593,
      "step": 10587
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2755902067388574,
      "learning_rate": 1.4661047047273503e-06,
      "loss": 0.1782,
      "step": 10588
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.2034993819031525,
      "learning_rate": 1.4652853045325866e-06,
      "loss": 0.6184,
      "step": 10589
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2600635159021931,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.1534,
      "step": 10590
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3559647041955567,
      "learning_rate": 1.4636470733753534e-06,
      "loss": 0.198,
      "step": 10591
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4056354987457755,
      "learning_rate": 1.4628282425008217e-06,
      "loss": 0.178,
      "step": 10592
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3898302578419468,
      "learning_rate": 1.4620096014876156e-06,
      "loss": 0.1866,
      "step": 10593
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.382559226409949,
      "learning_rate": 1.4611911503796806e-06,
      "loss": 0.1899,
      "step": 10594
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2286515797178907,
      "learning_rate": 1.4603728892209418e-06,
      "loss": 0.1618,
      "step": 10595
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1764502219829687,
      "learning_rate": 1.4595548180553275e-06,
      "loss": 0.1398,
      "step": 10596
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.317049685017896,
      "learning_rate": 1.4587369369267435e-06,
      "loss": 0.188,
      "step": 10597
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2055789463948723,
      "learning_rate": 1.4579192458790937e-06,
      "loss": 0.1763,
      "step": 10598
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.401903892202455,
      "learning_rate": 1.4571017449562652e-06,
      "loss": 0.1449,
      "step": 10599
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1100904088419075,
      "learning_rate": 1.456284434202142e-06,
      "loss": 0.1224,
      "step": 10600
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2103193996723656,
      "learning_rate": 1.4554673136605896e-06,
      "loss": 0.1722,
      "step": 10601
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3886078014208094,
      "learning_rate": 1.454650383375472e-06,
      "loss": 0.1972,
      "step": 10602
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4483303317922087,
      "learning_rate": 1.4538336433906352e-06,
      "loss": 0.1956,
      "step": 10603
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5776078396865971,
      "learning_rate": 1.4530170937499206e-06,
      "loss": 0.2079,
      "step": 10604
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.532278756700247,
      "learning_rate": 1.4522007344971566e-06,
      "loss": 0.7007,
      "step": 10605
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.274398674317184,
      "learning_rate": 1.4513845656761644e-06,
      "loss": 0.1568,
      "step": 10606
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3929696665481452,
      "learning_rate": 1.450568587330749e-06,
      "loss": 0.2015,
      "step": 10607
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.05913217738532,
      "learning_rate": 1.4497527995047123e-06,
      "loss": 0.7864,
      "step": 10608
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2535824518215695,
      "learning_rate": 1.448937202241839e-06,
      "loss": 0.178,
      "step": 10609
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4345734708213747,
      "learning_rate": 1.4481217955859112e-06,
      "loss": 0.1528,
      "step": 10610
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4706450842209635,
      "learning_rate": 1.447306579580693e-06,
      "loss": 0.2059,
      "step": 10611
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2951598554853498,
      "learning_rate": 1.4464915542699432e-06,
      "loss": 0.157,
      "step": 10612
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3911767722779842,
      "learning_rate": 1.4456767196974098e-06,
      "loss": 0.1437,
      "step": 10613
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1517692208504924,
      "learning_rate": 1.4448620759068304e-06,
      "loss": 0.1106,
      "step": 10614
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2026303945974282,
      "learning_rate": 1.4440476229419299e-06,
      "loss": 0.1763,
      "step": 10615
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.274013434380245,
      "learning_rate": 1.4432333608464276e-06,
      "loss": 0.1538,
      "step": 10616
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1760165045887663,
      "learning_rate": 1.4424192896640265e-06,
      "loss": 0.169,
      "step": 10617
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4586743250582557,
      "learning_rate": 1.4416054094384257e-06,
      "loss": 0.1859,
      "step": 10618
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.245373465617723,
      "learning_rate": 1.4407917202133088e-06,
      "loss": 0.1449,
      "step": 10619
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4222419465369949,
      "learning_rate": 1.4399782220323515e-06,
      "loss": 0.1741,
      "step": 10620
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2488694661936237,
      "learning_rate": 1.4391649149392205e-06,
      "loss": 0.1859,
      "step": 10621
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3624180212075656,
      "learning_rate": 1.4383517989775708e-06,
      "loss": 0.1906,
      "step": 10622
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.267535233288876,
      "learning_rate": 1.4375388741910478e-06,
      "loss": 0.1584,
      "step": 10623
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.323832012700867,
      "learning_rate": 1.436726140623284e-06,
      "loss": 0.4138,
      "step": 10624
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2061636025153357,
      "learning_rate": 1.4359135983179062e-06,
      "loss": 0.1426,
      "step": 10625
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.455849778896278,
      "learning_rate": 1.4351012473185255e-06,
      "loss": 0.1575,
      "step": 10626
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6485764862807495,
      "learning_rate": 1.4342890876687487e-06,
      "loss": 0.2042,
      "step": 10627
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4016511792544684,
      "learning_rate": 1.4334771194121667e-06,
      "loss": 0.1816,
      "step": 10628
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3714802513009687,
      "learning_rate": 1.4326653425923637e-06,
      "loss": 0.1567,
      "step": 10629
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3169689483132603,
      "learning_rate": 1.4318537572529129e-06,
      "loss": 0.1821,
      "step": 10630
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3282215000613666,
      "learning_rate": 1.4310423634373781e-06,
      "loss": 0.1587,
      "step": 10631
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.22594977702295,
      "learning_rate": 1.4302311611893095e-06,
      "loss": 0.1319,
      "step": 10632
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.458736479077391,
      "learning_rate": 1.4294201505522516e-06,
      "loss": 0.1896,
      "step": 10633
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2752821605354525,
      "learning_rate": 1.428609331569733e-06,
      "loss": 0.1363,
      "step": 10634
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3078509392709274,
      "learning_rate": 1.4277987042852787e-06,
      "loss": 0.1518,
      "step": 10635
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3532817483186756,
      "learning_rate": 1.426988268742397e-06,
      "loss": 0.1921,
      "step": 10636
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3522630572229293,
      "learning_rate": 1.4261780249845902e-06,
      "loss": 0.1619,
      "step": 10637
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4884331423372616,
      "learning_rate": 1.4253679730553489e-06,
      "loss": 0.185,
      "step": 10638
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2350711450728686,
      "learning_rate": 1.424558112998155e-06,
      "loss": 0.1362,
      "step": 10639
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4189969768197646,
      "learning_rate": 1.4237484448564759e-06,
      "loss": 0.1591,
      "step": 10640
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.342769001404274,
      "learning_rate": 1.4229389686737737e-06,
      "loss": 0.1933,
      "step": 10641
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.690512209122188,
      "learning_rate": 1.4221296844934956e-06,
      "loss": 0.2197,
      "step": 10642
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3615159008118802,
      "learning_rate": 1.4213205923590833e-06,
      "loss": 0.1795,
      "step": 10643
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3460883783826403,
      "learning_rate": 1.4205116923139628e-06,
      "loss": 0.1586,
      "step": 10644
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4570477207600203,
      "learning_rate": 1.4197029844015541e-06,
      "loss": 0.1852,
      "step": 10645
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3259620844575377,
      "learning_rate": 1.4188944686652651e-06,
      "loss": 0.1681,
      "step": 10646
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3606126696906709,
      "learning_rate": 1.4180861451484962e-06,
      "loss": 0.1834,
      "step": 10647
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3751019903512172,
      "learning_rate": 1.4172780138946306e-06,
      "loss": 0.1626,
      "step": 10648
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3395099325671,
      "learning_rate": 1.41647007494705e-06,
      "loss": 0.165,
      "step": 10649
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4955340846064766,
      "learning_rate": 1.4156623283491171e-06,
      "loss": 0.1692,
      "step": 10650
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3156014401931164,
      "learning_rate": 1.4148547741441926e-06,
      "loss": 0.1725,
      "step": 10651
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4296605039360266,
      "learning_rate": 1.4140474123756186e-06,
      "loss": 0.1787,
      "step": 10652
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.6238803795863435,
      "learning_rate": 1.4132402430867343e-06,
      "loss": 0.6098,
      "step": 10653
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2443303865267252,
      "learning_rate": 1.4124332663208635e-06,
      "loss": 0.1592,
      "step": 10654
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2419548980365436,
      "learning_rate": 1.4116264821213221e-06,
      "loss": 0.1412,
      "step": 10655
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2839150273379119,
      "learning_rate": 1.4108198905314175e-06,
      "loss": 0.1334,
      "step": 10656
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.463591575678856,
      "learning_rate": 1.4100134915944403e-06,
      "loss": 0.2239,
      "step": 10657
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.255725157559926,
      "learning_rate": 1.409207285353678e-06,
      "loss": 0.6076,
      "step": 10658
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4751197645117728,
      "learning_rate": 1.408401271852402e-06,
      "loss": 0.1589,
      "step": 10659
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4230251284826614,
      "learning_rate": 1.4075954511338784e-06,
      "loss": 0.2015,
      "step": 10660
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.678180395079142,
      "learning_rate": 1.4067898232413574e-06,
      "loss": 0.5524,
      "step": 10661
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3069921696200468,
      "learning_rate": 1.405984388218084e-06,
      "loss": 0.18,
      "step": 10662
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.321398413399983,
      "learning_rate": 1.40517914610729e-06,
      "loss": 0.1776,
      "step": 10663
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3581919204301052,
      "learning_rate": 1.4043740969521996e-06,
      "loss": 0.1864,
      "step": 10664
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3715465049394573,
      "learning_rate": 1.4035692407960217e-06,
      "loss": 0.1939,
      "step": 10665
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3221011476093758,
      "learning_rate": 1.4027645776819605e-06,
      "loss": 0.1708,
      "step": 10666
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2498033559009347,
      "learning_rate": 1.4019601076532042e-06,
      "loss": 0.1749,
      "step": 10667
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1392541844568327,
      "learning_rate": 1.4011558307529366e-06,
      "loss": 0.1712,
      "step": 10668
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3490334017490764,
      "learning_rate": 1.400351747024325e-06,
      "loss": 0.1682,
      "step": 10669
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2581572199371105,
      "learning_rate": 1.3995478565105308e-06,
      "loss": 0.149,
      "step": 10670
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3912356378844166,
      "learning_rate": 1.3987441592547041e-06,
      "loss": 0.1837,
      "step": 10671
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.996208750013756,
      "learning_rate": 1.3979406552999852e-06,
      "loss": 0.5232,
      "step": 10672
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0483495864350354,
      "learning_rate": 1.3971373446895004e-06,
      "loss": 0.1332,
      "step": 10673
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2255004310564488,
      "learning_rate": 1.396334227466371e-06,
      "loss": 0.1024,
      "step": 10674
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.030652704900943,
      "learning_rate": 1.3955313036737022e-06,
      "loss": 0.7205,
      "step": 10675
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3234941592276754,
      "learning_rate": 1.3947285733545946e-06,
      "loss": 0.2008,
      "step": 10676
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2332184681005414,
      "learning_rate": 1.3939260365521328e-06,
      "loss": 0.1915,
      "step": 10677
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3201239693417748,
      "learning_rate": 1.393123693309395e-06,
      "loss": 0.1663,
      "step": 10678
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3791539649097073,
      "learning_rate": 1.3923215436694482e-06,
      "loss": 0.1367,
      "step": 10679
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2655939352799814,
      "learning_rate": 1.3915195876753495e-06,
      "loss": 0.1552,
      "step": 10680
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4942336063625583,
      "learning_rate": 1.3907178253701427e-06,
      "loss": 0.2201,
      "step": 10681
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3332296802201882,
      "learning_rate": 1.389916256796865e-06,
      "loss": 0.1563,
      "step": 10682
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.337693818909043,
      "learning_rate": 1.3891148819985395e-06,
      "loss": 0.4015,
      "step": 10683
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.49005665195222,
      "learning_rate": 1.388313701018183e-06,
      "loss": 0.1844,
      "step": 10684
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.953453429530292,
      "learning_rate": 1.3875127138987972e-06,
      "loss": 0.5978,
      "step": 10685
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3498551898418027,
      "learning_rate": 1.3867119206833773e-06,
      "loss": 0.1469,
      "step": 10686
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.559914919221235,
      "learning_rate": 1.3859113214149066e-06,
      "loss": 0.1717,
      "step": 10687
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4020726669669477,
      "learning_rate": 1.3851109161363584e-06,
      "loss": 0.1884,
      "step": 10688
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.136795463745805,
      "learning_rate": 1.3843107048906961e-06,
      "loss": 0.6171,
      "step": 10689
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.956703608838789,
      "learning_rate": 1.3835106877208698e-06,
      "loss": 0.453,
      "step": 10690
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3806030281590127,
      "learning_rate": 1.3827108646698234e-06,
      "loss": 0.1569,
      "step": 10691
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5446572164495918,
      "learning_rate": 1.3819112357804859e-06,
      "loss": 0.1968,
      "step": 10692
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3457324067460188,
      "learning_rate": 1.3811118010957803e-06,
      "loss": 0.161,
      "step": 10693
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3644741713327855,
      "learning_rate": 1.3803125606586149e-06,
      "loss": 0.1852,
      "step": 10694
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4531227434780998,
      "learning_rate": 1.3795135145118927e-06,
      "loss": 0.1499,
      "step": 10695
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3160971956684657,
      "learning_rate": 1.3787146626984999e-06,
      "loss": 0.1703,
      "step": 10696
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3509052832404558,
      "learning_rate": 1.3779160052613172e-06,
      "loss": 0.1431,
      "step": 10697
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2145492347667803,
      "learning_rate": 1.3771175422432137e-06,
      "loss": 0.161,
      "step": 10698
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3310841235838138,
      "learning_rate": 1.3763192736870496e-06,
      "loss": 0.181,
      "step": 10699
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.298008941856758,
      "learning_rate": 1.3755211996356687e-06,
      "loss": 0.1535,
      "step": 10700
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5573491298206854,
      "learning_rate": 1.374723320131912e-06,
      "loss": 0.1949,
      "step": 10701
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4205135782840004,
      "learning_rate": 1.3739256352186038e-06,
      "loss": 0.185,
      "step": 10702
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2957289293795626,
      "learning_rate": 1.373128144938563e-06,
      "loss": 0.1542,
      "step": 10703
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2427103455394968,
      "learning_rate": 1.3723308493345939e-06,
      "loss": 0.1424,
      "step": 10704
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2340094317129295,
      "learning_rate": 1.3715337484494917e-06,
      "loss": 0.1847,
      "step": 10705
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1743704986028893,
      "learning_rate": 1.3707368423260436e-06,
      "loss": 0.1663,
      "step": 10706
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1215933133294034,
      "learning_rate": 1.3699401310070249e-06,
      "loss": 0.1375,
      "step": 10707
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2938966493683683,
      "learning_rate": 1.3691436145351967e-06,
      "loss": 0.1907,
      "step": 10708
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.8701422556617615,
      "learning_rate": 1.3683472929533164e-06,
      "loss": 0.5278,
      "step": 10709
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4582423101502153,
      "learning_rate": 1.3675511663041235e-06,
      "loss": 0.2116,
      "step": 10710
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.051575536793547,
      "learning_rate": 1.3667552346303554e-06,
      "loss": 0.4116,
      "step": 10711
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2820486866287786,
      "learning_rate": 1.3659594979747304e-06,
      "loss": 0.1709,
      "step": 10712
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6009099370694422,
      "learning_rate": 1.3651639563799618e-06,
      "loss": 0.1589,
      "step": 10713
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.265859931306252,
      "learning_rate": 1.364368609888752e-06,
      "loss": 0.5951,
      "step": 10714
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.266262211605513,
      "learning_rate": 1.3635734585437933e-06,
      "loss": 0.1512,
      "step": 10715
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1850726031690055,
      "learning_rate": 1.3627785023877633e-06,
      "loss": 0.131,
      "step": 10716
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.190135136857808,
      "learning_rate": 1.361983741463334e-06,
      "loss": 0.4616,
      "step": 10717
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3671267463752237,
      "learning_rate": 1.3611891758131635e-06,
      "loss": 0.1679,
      "step": 10718
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2270367238849111,
      "learning_rate": 1.3603948054799016e-06,
      "loss": 0.1538,
      "step": 10719
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4798486614313762,
      "learning_rate": 1.3596006305061888e-06,
      "loss": 0.1802,
      "step": 10720
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4661646421872494,
      "learning_rate": 1.3588066509346498e-06,
      "loss": 0.2048,
      "step": 10721
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2889687781356651,
      "learning_rate": 1.3580128668079045e-06,
      "loss": 0.184,
      "step": 10722
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.985053012435174,
      "learning_rate": 1.3572192781685595e-06,
      "loss": 0.5419,
      "step": 10723
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.212364342184223,
      "learning_rate": 1.3564258850592131e-06,
      "loss": 0.1626,
      "step": 10724
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4136611149222236,
      "learning_rate": 1.3556326875224485e-06,
      "loss": 0.1638,
      "step": 10725
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3802477216946167,
      "learning_rate": 1.3548396856008445e-06,
      "loss": 0.1646,
      "step": 10726
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1029452967288487,
      "learning_rate": 1.3540468793369632e-06,
      "loss": 0.1415,
      "step": 10727
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4875400136410541,
      "learning_rate": 1.3532542687733624e-06,
      "loss": 0.1554,
      "step": 10728
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4135323257621524,
      "learning_rate": 1.3524618539525836e-06,
      "loss": 0.1946,
      "step": 10729
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4147146601308604,
      "learning_rate": 1.351669634917161e-06,
      "loss": 0.1788,
      "step": 10730
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.312406363220769,
      "learning_rate": 1.3508776117096184e-06,
      "loss": 0.1385,
      "step": 10731
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1981369969407047,
      "learning_rate": 1.3500857843724701e-06,
      "loss": 0.1407,
      "step": 10732
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3281657345711524,
      "learning_rate": 1.3492941529482145e-06,
      "loss": 0.1767,
      "step": 10733
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.007834809378267,
      "learning_rate": 1.3485027174793475e-06,
      "loss": 0.5169,
      "step": 10734
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2993655934141313,
      "learning_rate": 1.3477114780083462e-06,
      "loss": 0.1524,
      "step": 10735
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4490315968916014,
      "learning_rate": 1.3469204345776843e-06,
      "loss": 0.1752,
      "step": 10736
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2174064564698628,
      "learning_rate": 1.3461295872298197e-06,
      "loss": 0.1604,
      "step": 10737
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5923419265567251,
      "learning_rate": 1.3453389360072022e-06,
      "loss": 0.1876,
      "step": 10738
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3310326597531117,
      "learning_rate": 1.3445484809522714e-06,
      "loss": 0.1729,
      "step": 10739
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3161242769628008,
      "learning_rate": 1.3437582221074574e-06,
      "loss": 0.1407,
      "step": 10740
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3062713388224785,
      "learning_rate": 1.3429681595151755e-06,
      "loss": 0.1662,
      "step": 10741
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2212507405389408,
      "learning_rate": 1.342178293217835e-06,
      "loss": 0.1821,
      "step": 10742
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5155222247509201,
      "learning_rate": 1.341388623257831e-06,
      "loss": 0.1813,
      "step": 10743
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2930395148305247,
      "learning_rate": 1.3405991496775521e-06,
      "loss": 0.1843,
      "step": 10744
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.293207003362307,
      "learning_rate": 1.3398098725193715e-06,
      "loss": 0.1334,
      "step": 10745
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2403723215333968,
      "learning_rate": 1.3390207918256555e-06,
      "loss": 0.1543,
      "step": 10746
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2677875162831829,
      "learning_rate": 1.338231907638759e-06,
      "loss": 0.1838,
      "step": 10747
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2821863605813173,
      "learning_rate": 1.3374432200010278e-06,
      "loss": 0.1824,
      "step": 10748
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2432247652092665,
      "learning_rate": 1.3366547289547933e-06,
      "loss": 0.1468,
      "step": 10749
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.290581399909029,
      "learning_rate": 1.3358664345423783e-06,
      "loss": 0.1435,
      "step": 10750
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2478340865014668,
      "learning_rate": 1.3350783368060983e-06,
      "loss": 0.1537,
      "step": 10751
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3428385808940781,
      "learning_rate": 1.3342904357882514e-06,
      "loss": 0.1549,
      "step": 10752
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.759993489006135,
      "learning_rate": 1.333502731531132e-06,
      "loss": 0.6973,
      "step": 10753
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2335013416393081,
      "learning_rate": 1.3327152240770191e-06,
      "loss": 0.1641,
      "step": 10754
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2221054484779326,
      "learning_rate": 1.3319279134681829e-06,
      "loss": 0.1378,
      "step": 10755
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.197778683165155,
      "learning_rate": 1.331140799746884e-06,
      "loss": 0.6157,
      "step": 10756
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4217412452795737,
      "learning_rate": 1.3303538829553731e-06,
      "loss": 0.2072,
      "step": 10757
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2247622852655058,
      "learning_rate": 1.3295671631358852e-06,
      "loss": 0.1881,
      "step": 10758
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4264226701317393,
      "learning_rate": 1.328780640330652e-06,
      "loss": 0.1802,
      "step": 10759
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.909014684575279,
      "learning_rate": 1.3279943145818874e-06,
      "loss": 0.5543,
      "step": 10760
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.289576134396907,
      "learning_rate": 1.3272081859318014e-06,
      "loss": 0.1712,
      "step": 10761
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3161224002137772,
      "learning_rate": 1.3264222544225874e-06,
      "loss": 0.1726,
      "step": 10762
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4577136684418077,
      "learning_rate": 1.3256365200964321e-06,
      "loss": 0.1934,
      "step": 10763
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.453773862187874,
      "learning_rate": 1.324850982995511e-06,
      "loss": 0.4802,
      "step": 10764
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.202690816933377,
      "learning_rate": 1.3240656431619903e-06,
      "loss": 0.5093,
      "step": 10765
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3131469273857477,
      "learning_rate": 1.32328050063802e-06,
      "loss": 0.1896,
      "step": 10766
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.241394643224215,
      "learning_rate": 1.3224955554657475e-06,
      "loss": 0.163,
      "step": 10767
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2424292059084432,
      "learning_rate": 1.3217108076873026e-06,
      "loss": 0.1712,
      "step": 10768
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2235669730371863,
      "learning_rate": 1.3209262573448095e-06,
      "loss": 0.1124,
      "step": 10769
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3948656746150816,
      "learning_rate": 1.320141904480377e-06,
      "loss": 0.1842,
      "step": 10770
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3406788555407718,
      "learning_rate": 1.3193577491361082e-06,
      "loss": 0.1678,
      "step": 10771
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1699421257821243,
      "learning_rate": 1.318573791354093e-06,
      "loss": 0.1496,
      "step": 10772
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3627037525290404,
      "learning_rate": 1.3177900311764125e-06,
      "loss": 0.1815,
      "step": 10773
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3279741362380528,
      "learning_rate": 1.317006468645133e-06,
      "loss": 0.156,
      "step": 10774
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2818253928538257,
      "learning_rate": 1.316223103802316e-06,
      "loss": 0.1659,
      "step": 10775
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.212814788672113,
      "learning_rate": 1.3154399366900068e-06,
      "loss": 0.1547,
      "step": 10776
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3640529959116054,
      "learning_rate": 1.3146569673502447e-06,
      "loss": 0.1842,
      "step": 10777
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.233836640167664,
      "learning_rate": 1.3138741958250545e-06,
      "loss": 0.1375,
      "step": 10778
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.291666631101557,
      "learning_rate": 1.3130916221564528e-06,
      "loss": 0.1533,
      "step": 10779
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.42101048046062,
      "learning_rate": 1.3123092463864456e-06,
      "loss": 0.1917,
      "step": 10780
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2101729503749248,
      "learning_rate": 1.3115270685570292e-06,
      "loss": 0.1339,
      "step": 10781
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0872991727571533,
      "learning_rate": 1.3107450887101847e-06,
      "loss": 0.1341,
      "step": 10782
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2492778709917987,
      "learning_rate": 1.3099633068878875e-06,
      "loss": 0.1361,
      "step": 10783
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4926090007333466,
      "learning_rate": 1.3091817231321013e-06,
      "loss": 0.1931,
      "step": 10784
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.398321310675229,
      "learning_rate": 1.3084003374847765e-06,
      "loss": 0.2076,
      "step": 10785
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3068888023636955,
      "learning_rate": 1.3076191499878565e-06,
      "loss": 0.1791,
      "step": 10786
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.437667974651349,
      "learning_rate": 1.3068381606832702e-06,
      "loss": 0.1769,
      "step": 10787
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3661340203824657,
      "learning_rate": 1.3060573696129396e-06,
      "loss": 0.176,
      "step": 10788
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2177434541263317,
      "learning_rate": 1.3052767768187734e-06,
      "loss": 0.1511,
      "step": 10789
    },
    {
      "epoch": 0.77,
      "grad_norm": 12.481775094810473,
      "learning_rate": 1.3044963823426737e-06,
      "loss": 0.6787,
      "step": 10790
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3808560970726922,
      "learning_rate": 1.3037161862265252e-06,
      "loss": 0.2009,
      "step": 10791
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2720916362864194,
      "learning_rate": 1.3029361885122083e-06,
      "loss": 0.1844,
      "step": 10792
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.431056799062906,
      "learning_rate": 1.3021563892415879e-06,
      "loss": 0.1363,
      "step": 10793
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.674698553392927,
      "learning_rate": 1.3013767884565232e-06,
      "loss": 0.5085,
      "step": 10794
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7782155019521222,
      "learning_rate": 1.3005973861988574e-06,
      "loss": 0.2079,
      "step": 10795
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5463308552469406,
      "learning_rate": 1.2998181825104283e-06,
      "loss": 0.1791,
      "step": 10796
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3091103574733114,
      "learning_rate": 1.2990391774330573e-06,
      "loss": 0.1695,
      "step": 10797
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.353386788623521,
      "learning_rate": 1.2982603710085606e-06,
      "loss": 0.1644,
      "step": 10798
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4341121999679103,
      "learning_rate": 1.2974817632787406e-06,
      "loss": 0.1683,
      "step": 10799
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.297081967123346,
      "learning_rate": 1.2967033542853918e-06,
      "loss": 0.1585,
      "step": 10800
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3371651256172636,
      "learning_rate": 1.295925144070293e-06,
      "loss": 0.1878,
      "step": 10801
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4074227028619066,
      "learning_rate": 1.2951471326752184e-06,
      "loss": 0.2135,
      "step": 10802
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.443389796182147,
      "learning_rate": 1.294369320141926e-06,
      "loss": 0.167,
      "step": 10803
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4247181315889128,
      "learning_rate": 1.2935917065121684e-06,
      "loss": 0.2057,
      "step": 10804
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3429033729282496,
      "learning_rate": 1.292814291827681e-06,
      "loss": 0.1952,
      "step": 10805
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2981290072483687,
      "learning_rate": 1.2920370761301948e-06,
      "loss": 0.1744,
      "step": 10806
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3093861256219763,
      "learning_rate": 1.2912600594614276e-06,
      "loss": 0.1556,
      "step": 10807
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5418434318989682,
      "learning_rate": 1.290483241863088e-06,
      "loss": 0.1746,
      "step": 10808
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.599394965787028,
      "learning_rate": 1.289706623376869e-06,
      "loss": 0.1806,
      "step": 10809
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.1614939022579795,
      "learning_rate": 1.28893020404446e-06,
      "loss": 0.6115,
      "step": 10810
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1757367453635346,
      "learning_rate": 1.288153983907533e-06,
      "loss": 0.1286,
      "step": 10811
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6069544364596753,
      "learning_rate": 1.287377963007755e-06,
      "loss": 0.2062,
      "step": 10812
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3247923791551692,
      "learning_rate": 1.2866021413867768e-06,
      "loss": 0.159,
      "step": 10813
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2720482341489399,
      "learning_rate": 1.2858265190862434e-06,
      "loss": 0.1712,
      "step": 10814
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.543794340047884,
      "learning_rate": 1.2850510961477863e-06,
      "loss": 0.1826,
      "step": 10815
    },
    {
      "epoch": 0.77,
      "grad_norm": 10.422743007194265,
      "learning_rate": 1.2842758726130283e-06,
      "loss": 0.6393,
      "step": 10816
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.26988826346006,
      "learning_rate": 1.2835008485235811e-06,
      "loss": 0.1465,
      "step": 10817
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2215915476933914,
      "learning_rate": 1.2827260239210415e-06,
      "loss": 0.1514,
      "step": 10818
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.251385506681329,
      "learning_rate": 1.2819513988470027e-06,
      "loss": 0.5144,
      "step": 10819
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2703000025537514,
      "learning_rate": 1.2811769733430406e-06,
      "loss": 0.1469,
      "step": 10820
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.034792231837164,
      "learning_rate": 1.2804027474507259e-06,
      "loss": 0.5539,
      "step": 10821
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.373248559033873,
      "learning_rate": 1.2796287212116126e-06,
      "loss": 0.1991,
      "step": 10822
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3005342086617848,
      "learning_rate": 1.2788548946672496e-06,
      "loss": 0.1845,
      "step": 10823
    },
    {
      "epoch": 0.77,
      "grad_norm": 10.027266465669893,
      "learning_rate": 1.2780812678591724e-06,
      "loss": 0.5943,
      "step": 10824
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4741953311200033,
      "learning_rate": 1.277307840828908e-06,
      "loss": 0.1789,
      "step": 10825
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.195056417417213,
      "learning_rate": 1.276534613617968e-06,
      "loss": 0.6181,
      "step": 10826
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.209736903782424,
      "learning_rate": 1.2757615862678584e-06,
      "loss": 0.118,
      "step": 10827
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.316525220252832,
      "learning_rate": 1.27498875882007e-06,
      "loss": 0.1601,
      "step": 10828
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4922975863139818,
      "learning_rate": 1.2742161313160877e-06,
      "loss": 0.1983,
      "step": 10829
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.26735752829659,
      "learning_rate": 1.27344370379738e-06,
      "loss": 0.1792,
      "step": 10830
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2246445151973904,
      "learning_rate": 1.2726714763054104e-06,
      "loss": 0.1401,
      "step": 10831
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3422349553176591,
      "learning_rate": 1.2718994488816272e-06,
      "loss": 0.177,
      "step": 10832
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1797925820374757,
      "learning_rate": 1.2711276215674722e-06,
      "loss": 0.1416,
      "step": 10833
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.641156412222506,
      "learning_rate": 1.2703559944043714e-06,
      "loss": 0.3846,
      "step": 10834
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3225248956186373,
      "learning_rate": 1.2695845674337447e-06,
      "loss": 0.1421,
      "step": 10835
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2884608915474698,
      "learning_rate": 1.268813340696997e-06,
      "loss": 0.1572,
      "step": 10836
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.594848801598252,
      "learning_rate": 1.2680423142355274e-06,
      "loss": 0.733,
      "step": 10837
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2734102607225462,
      "learning_rate": 1.267271488090719e-06,
      "loss": 0.1652,
      "step": 10838
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4268329557703527,
      "learning_rate": 1.266500862303947e-06,
      "loss": 0.1796,
      "step": 10839
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2390693092215115,
      "learning_rate": 1.2657304369165768e-06,
      "loss": 0.1632,
      "step": 10840
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3174904981271964,
      "learning_rate": 1.2649602119699628e-06,
      "loss": 0.1996,
      "step": 10841
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.257970487634104,
      "learning_rate": 1.264190187505444e-06,
      "loss": 0.1792,
      "step": 10842
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3376028391109658,
      "learning_rate": 1.2634203635643566e-06,
      "loss": 0.2356,
      "step": 10843
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.779189018283161,
      "learning_rate": 1.2626507401880178e-06,
      "loss": 0.5841,
      "step": 10844
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.71826309729734,
      "learning_rate": 1.2618813174177408e-06,
      "loss": 0.1671,
      "step": 10845
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2294011185593903,
      "learning_rate": 1.2611120952948225e-06,
      "loss": 0.1719,
      "step": 10846
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3706348574669802,
      "learning_rate": 1.2603430738605532e-06,
      "loss": 0.1719,
      "step": 10847
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.562828388910144,
      "learning_rate": 1.2595742531562112e-06,
      "loss": 0.159,
      "step": 10848
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.165477515659987,
      "learning_rate": 1.2588056332230624e-06,
      "loss": 0.1452,
      "step": 10849
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3470305613664593,
      "learning_rate": 1.2580372141023666e-06,
      "loss": 0.1567,
      "step": 10850
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4298814770524364,
      "learning_rate": 1.2572689958353652e-06,
      "loss": 0.1645,
      "step": 10851
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1394094001619999,
      "learning_rate": 1.256500978463297e-06,
      "loss": 0.1454,
      "step": 10852
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4439355287298474,
      "learning_rate": 1.2557331620273822e-06,
      "loss": 0.1641,
      "step": 10853
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3018782675310325,
      "learning_rate": 1.2549655465688377e-06,
      "loss": 0.1571,
      "step": 10854
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2178248179763735,
      "learning_rate": 1.2541981321288627e-06,
      "loss": 0.1379,
      "step": 10855
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.248986747754545,
      "learning_rate": 1.253430918748651e-06,
      "loss": 0.1613,
      "step": 10856
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1738555423077708,
      "learning_rate": 1.252663906469383e-06,
      "loss": 0.1293,
      "step": 10857
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2671663491080725,
      "learning_rate": 1.2518970953322307e-06,
      "loss": 0.1754,
      "step": 10858
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.292379662548495,
      "learning_rate": 1.2511304853783507e-06,
      "loss": 0.1528,
      "step": 10859
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5040790519795335,
      "learning_rate": 1.250364076648894e-06,
      "loss": 0.2087,
      "step": 10860
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.626455555642887,
      "learning_rate": 1.2495978691849952e-06,
      "loss": 0.5038,
      "step": 10861
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2478144347957785,
      "learning_rate": 1.2488318630277857e-06,
      "loss": 0.1579,
      "step": 10862
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3496675002007257,
      "learning_rate": 1.2480660582183768e-06,
      "loss": 0.1635,
      "step": 10863
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.379713333188431,
      "learning_rate": 1.2473004547978767e-06,
      "loss": 0.1547,
      "step": 10864
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3782187428241317,
      "learning_rate": 1.2465350528073795e-06,
      "loss": 0.1948,
      "step": 10865
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3063237183790244,
      "learning_rate": 1.2457698522879708e-06,
      "loss": 0.2007,
      "step": 10866
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2587602041541603,
      "learning_rate": 1.2450048532807197e-06,
      "loss": 0.1636,
      "step": 10867
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.347123677399065,
      "learning_rate": 1.2442400558266915e-06,
      "loss": 0.1646,
      "step": 10868
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5743149012516464,
      "learning_rate": 1.2434754599669352e-06,
      "loss": 0.1786,
      "step": 10869
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3615329254520898,
      "learning_rate": 1.2427110657424934e-06,
      "loss": 0.1503,
      "step": 10870
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2555263759081152,
      "learning_rate": 1.241946873194394e-06,
      "loss": 0.1559,
      "step": 10871
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.387864910690997,
      "learning_rate": 1.2411828823636556e-06,
      "loss": 0.1784,
      "step": 10872
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4576703889354408,
      "learning_rate": 1.240419093291288e-06,
      "loss": 0.1873,
      "step": 10873
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.604596962934677,
      "learning_rate": 1.239655506018288e-06,
      "loss": 0.1913,
      "step": 10874
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3784569633327155,
      "learning_rate": 1.2388921205856402e-06,
      "loss": 0.1751,
      "step": 10875
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5264233469757673,
      "learning_rate": 1.238128937034323e-06,
      "loss": 0.137,
      "step": 10876
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2066734265868877,
      "learning_rate": 1.237365955405298e-06,
      "loss": 0.155,
      "step": 10877
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.316622575817147,
      "learning_rate": 1.2366031757395196e-06,
      "loss": 0.1661,
      "step": 10878
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.460600512919755,
      "learning_rate": 1.2358405980779336e-06,
      "loss": 0.1923,
      "step": 10879
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2185596180850247,
      "learning_rate": 1.2350782224614689e-06,
      "loss": 0.1684,
      "step": 10880
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.163605919937378,
      "learning_rate": 1.2343160489310474e-06,
      "loss": 0.2819,
      "step": 10881
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4604222302466239,
      "learning_rate": 1.2335540775275805e-06,
      "loss": 0.1576,
      "step": 10882
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.177485692583051,
      "learning_rate": 1.2327923082919686e-06,
      "loss": 0.5795,
      "step": 10883
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3418368450868547,
      "learning_rate": 1.2320307412650978e-06,
      "loss": 0.1696,
      "step": 10884
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.274821050036068,
      "learning_rate": 1.231269376487849e-06,
      "loss": 0.5121,
      "step": 10885
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2545933036410872,
      "learning_rate": 1.2305082140010866e-06,
      "loss": 0.1599,
      "step": 10886
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4646050978552463,
      "learning_rate": 1.2297472538456689e-06,
      "loss": 0.1458,
      "step": 10887
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4085044292407272,
      "learning_rate": 1.2289864960624388e-06,
      "loss": 0.1493,
      "step": 10888
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2075088285832922,
      "learning_rate": 1.2282259406922337e-06,
      "loss": 0.1613,
      "step": 10889
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.20321872976204,
      "learning_rate": 1.227465587775873e-06,
      "loss": 0.4779,
      "step": 10890
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.881576955854875,
      "learning_rate": 1.2267054373541743e-06,
      "loss": 0.5566,
      "step": 10891
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5304905032403282,
      "learning_rate": 1.2259454894679363e-06,
      "loss": 0.1637,
      "step": 10892
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3540299318797002,
      "learning_rate": 1.2251857441579518e-06,
      "loss": 0.1705,
      "step": 10893
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1621470967677188,
      "learning_rate": 1.2244262014649994e-06,
      "loss": 0.1147,
      "step": 10894
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3169760199568337,
      "learning_rate": 1.2236668614298497e-06,
      "loss": 0.1769,
      "step": 10895
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.09762923319316,
      "learning_rate": 1.222907724093259e-06,
      "loss": 0.5675,
      "step": 10896
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.360619120709428,
      "learning_rate": 1.2221487894959778e-06,
      "loss": 0.1857,
      "step": 10897
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1287142915909105,
      "learning_rate": 1.221390057678739e-06,
      "loss": 0.1079,
      "step": 10898
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4101895248522336,
      "learning_rate": 1.2206315286822707e-06,
      "loss": 0.1937,
      "step": 10899
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3269436171358282,
      "learning_rate": 1.2198732025472876e-06,
      "loss": 0.1714,
      "step": 10900
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1989913590412595,
      "learning_rate": 1.219115079314494e-06,
      "loss": 0.1535,
      "step": 10901
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1847195826871226,
      "learning_rate": 1.2183571590245813e-06,
      "loss": 0.1476,
      "step": 10902
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2460408635768618,
      "learning_rate": 1.2175994417182342e-06,
      "loss": 0.1632,
      "step": 10903
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.6970286332008815,
      "learning_rate": 1.216841927436121e-06,
      "loss": 0.4323,
      "step": 10904
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3083750472280593,
      "learning_rate": 1.2160846162189044e-06,
      "loss": 0.176,
      "step": 10905
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.211950787332152,
      "learning_rate": 1.215327508107232e-06,
      "loss": 0.1522,
      "step": 10906
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2951256769106785,
      "learning_rate": 1.2145706031417427e-06,
      "loss": 0.1834,
      "step": 10907
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5386536060807698,
      "learning_rate": 1.213813901363065e-06,
      "loss": 0.183,
      "step": 10908
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3964815659951433,
      "learning_rate": 1.2130574028118168e-06,
      "loss": 0.1764,
      "step": 10909
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.359976872244731,
      "learning_rate": 1.2123011075286012e-06,
      "loss": 0.1852,
      "step": 10910
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.335506216544898,
      "learning_rate": 1.2115450155540143e-06,
      "loss": 0.1479,
      "step": 10911
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.290646557587444,
      "learning_rate": 1.2107891269286415e-06,
      "loss": 0.1651,
      "step": 10912
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5120755787944296,
      "learning_rate": 1.210033441693053e-06,
      "loss": 0.1768,
      "step": 10913
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5747317261744758,
      "learning_rate": 1.2092779598878146e-06,
      "loss": 0.1834,
      "step": 10914
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.588502352649904,
      "learning_rate": 1.208522681553474e-06,
      "loss": 0.4306,
      "step": 10915
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2427522956813504,
      "learning_rate": 1.2077676067305726e-06,
      "loss": 0.1694,
      "step": 10916
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.409532477575555,
      "learning_rate": 1.2070127354596411e-06,
      "loss": 0.494,
      "step": 10917
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3406072820409283,
      "learning_rate": 1.2062580677811986e-06,
      "loss": 0.1599,
      "step": 10918
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3208482030664774,
      "learning_rate": 1.2055036037357493e-06,
      "loss": 0.1831,
      "step": 10919
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.547293881378643,
      "learning_rate": 1.2047493433637935e-06,
      "loss": 0.5855,
      "step": 10920
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4690520122105315,
      "learning_rate": 1.2039952867058136e-06,
      "loss": 0.167,
      "step": 10921
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3842997818030294,
      "learning_rate": 1.203241433802288e-06,
      "loss": 0.1748,
      "step": 10922
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.29114389934993,
      "learning_rate": 1.2024877846936767e-06,
      "loss": 0.1671,
      "step": 10923
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3196215655203563,
      "learning_rate": 1.201734339420434e-06,
      "loss": 0.1637,
      "step": 10924
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2819160789914978,
      "learning_rate": 1.200981098023003e-06,
      "loss": 0.1343,
      "step": 10925
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2842097934757721,
      "learning_rate": 1.2002280605418149e-06,
      "loss": 0.1468,
      "step": 10926
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.777021201743696,
      "learning_rate": 1.1994752270172877e-06,
      "loss": 0.4906,
      "step": 10927
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4700286300938474,
      "learning_rate": 1.1987225974898326e-06,
      "loss": 0.1488,
      "step": 10928
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.224188443943246,
      "learning_rate": 1.1979701719998454e-06,
      "loss": 0.1574,
      "step": 10929
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.338422247971015,
      "learning_rate": 1.1972179505877168e-06,
      "loss": 0.1685,
      "step": 10930
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2497246407815572,
      "learning_rate": 1.1964659332938189e-06,
      "loss": 0.1397,
      "step": 10931
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1449469990354073,
      "learning_rate": 1.1957141201585193e-06,
      "loss": 0.1216,
      "step": 10932
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3536989041127743,
      "learning_rate": 1.194962511222172e-06,
      "loss": 0.1502,
      "step": 10933
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3061859697305618,
      "learning_rate": 1.194211106525122e-06,
      "loss": 0.165,
      "step": 10934
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.35080872445764,
      "learning_rate": 1.193459906107699e-06,
      "loss": 0.1876,
      "step": 10935
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4377018365144416,
      "learning_rate": 1.1927089100102268e-06,
      "loss": 0.1638,
      "step": 10936
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2835555119446052,
      "learning_rate": 1.1919581182730134e-06,
      "loss": 0.1316,
      "step": 10937
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4375372783979672,
      "learning_rate": 1.1912075309363607e-06,
      "loss": 0.1915,
      "step": 10938
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2576594981208595,
      "learning_rate": 1.1904571480405552e-06,
      "loss": 0.1809,
      "step": 10939
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1909520427187312,
      "learning_rate": 1.1897069696258756e-06,
      "loss": 0.121,
      "step": 10940
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3492986099296709,
      "learning_rate": 1.1889569957325881e-06,
      "loss": 0.1651,
      "step": 10941
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.360638553551535,
      "learning_rate": 1.1882072264009486e-06,
      "loss": 0.157,
      "step": 10942
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.269837929133121,
      "learning_rate": 1.1874576616712036e-06,
      "loss": 0.5576,
      "step": 10943
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3857224228903668,
      "learning_rate": 1.1867083015835828e-06,
      "loss": 0.1545,
      "step": 10944
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.224328313978514,
      "learning_rate": 1.1859591461783131e-06,
      "loss": 0.151,
      "step": 10945
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4989331932872134,
      "learning_rate": 1.1852101954956019e-06,
      "loss": 0.2055,
      "step": 10946
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.036791669584762,
      "learning_rate": 1.184461449575654e-06,
      "loss": 0.6189,
      "step": 10947
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5474920700449022,
      "learning_rate": 1.1837129084586557e-06,
      "loss": 0.1768,
      "step": 10948
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.241620725124109,
      "learning_rate": 1.1829645721847872e-06,
      "loss": 0.1487,
      "step": 10949
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.8920319927479605,
      "learning_rate": 1.1822164407942156e-06,
      "loss": 0.5676,
      "step": 10950
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.426617961786165,
      "learning_rate": 1.1814685143271004e-06,
      "loss": 0.4035,
      "step": 10951
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.895290640409777,
      "learning_rate": 1.180720792823583e-06,
      "loss": 0.5889,
      "step": 10952
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0963924067989739,
      "learning_rate": 1.179973276323802e-06,
      "loss": 0.1496,
      "step": 10953
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.277924117986559,
      "learning_rate": 1.1792259648678782e-06,
      "loss": 0.1605,
      "step": 10954
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2597152640325617,
      "learning_rate": 1.1784788584959268e-06,
      "loss": 0.1645,
      "step": 10955
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4568895569386944,
      "learning_rate": 1.1777319572480468e-06,
      "loss": 0.172,
      "step": 10956
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.407177695676701,
      "learning_rate": 1.1769852611643296e-06,
      "loss": 0.1434,
      "step": 10957
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2817150566782245,
      "learning_rate": 1.176238770284856e-06,
      "loss": 0.1491,
      "step": 10958
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5240994665309255,
      "learning_rate": 1.1754924846496957e-06,
      "loss": 0.1746,
      "step": 10959
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.458375019058329,
      "learning_rate": 1.1747464042989037e-06,
      "loss": 0.1697,
      "step": 10960
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3178840320139094,
      "learning_rate": 1.1740005292725286e-06,
      "loss": 0.1732,
      "step": 10961
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4279350471352732,
      "learning_rate": 1.1732548596106041e-06,
      "loss": 0.1437,
      "step": 10962
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1519674419501997,
      "learning_rate": 1.1725093953531573e-06,
      "loss": 0.135,
      "step": 10963
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2798063383622331,
      "learning_rate": 1.1717641365401994e-06,
      "loss": 0.1342,
      "step": 10964
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2441209933602742,
      "learning_rate": 1.171019083211733e-06,
      "loss": 0.1738,
      "step": 10965
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5427944260905202,
      "learning_rate": 1.1702742354077511e-06,
      "loss": 0.1493,
      "step": 10966
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.33790118895171,
      "learning_rate": 1.1695295931682348e-06,
      "loss": 0.1411,
      "step": 10967
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1839384806564794,
      "learning_rate": 1.168785156533151e-06,
      "loss": 0.1471,
      "step": 10968
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3485268163845912,
      "learning_rate": 1.1680409255424606e-06,
      "loss": 0.1997,
      "step": 10969
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3813885208552825,
      "learning_rate": 1.1672969002361083e-06,
      "loss": 0.1696,
      "step": 10970
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.274772352696011,
      "learning_rate": 1.1665530806540331e-06,
      "loss": 0.1722,
      "step": 10971
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0982590317808154,
      "learning_rate": 1.165809466836158e-06,
      "loss": 0.1162,
      "step": 10972
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3796921745862174,
      "learning_rate": 1.1650660588223982e-06,
      "loss": 0.1721,
      "step": 10973
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6604738275298139,
      "learning_rate": 1.164322856652657e-06,
      "loss": 0.2042,
      "step": 10974
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3010538100470785,
      "learning_rate": 1.1635798603668264e-06,
      "loss": 0.1611,
      "step": 10975
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3016453362634859,
      "learning_rate": 1.1628370700047892e-06,
      "loss": 0.1516,
      "step": 10976
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4682939908293164,
      "learning_rate": 1.1620944856064126e-06,
      "loss": 0.1821,
      "step": 10977
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3005040209427214,
      "learning_rate": 1.161352107211558e-06,
      "loss": 0.1868,
      "step": 10978
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3997474192724004,
      "learning_rate": 1.1606099348600703e-06,
      "loss": 0.1641,
      "step": 10979
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.087893926523309,
      "learning_rate": 1.1598679685917901e-06,
      "loss": 0.5316,
      "step": 10980
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1928035700434374,
      "learning_rate": 1.15912620844654e-06,
      "loss": 0.1259,
      "step": 10981
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.380211240870162,
      "learning_rate": 1.1583846544641357e-06,
      "loss": 0.2123,
      "step": 10982
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1997649445909777,
      "learning_rate": 1.1576433066843818e-06,
      "loss": 0.1583,
      "step": 10983
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6457831230459785,
      "learning_rate": 1.1569021651470714e-06,
      "loss": 0.2054,
      "step": 10984
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5439557982592125,
      "learning_rate": 1.1561612298919834e-06,
      "loss": 0.2149,
      "step": 10985
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3450096149097845,
      "learning_rate": 1.1554205009588909e-06,
      "loss": 0.1614,
      "step": 10986
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.6552913173139245,
      "learning_rate": 1.1546799783875512e-06,
      "loss": 0.5728,
      "step": 10987
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4442767736012718,
      "learning_rate": 1.1539396622177152e-06,
      "loss": 0.1911,
      "step": 10988
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.292966122348718,
      "learning_rate": 1.1531995524891172e-06,
      "loss": 0.1881,
      "step": 10989
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4765604588833408,
      "learning_rate": 1.1524596492414857e-06,
      "loss": 0.1637,
      "step": 10990
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5081912897601593,
      "learning_rate": 1.1517199525145317e-06,
      "loss": 0.226,
      "step": 10991
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1510794035893048,
      "learning_rate": 1.150980462347966e-06,
      "loss": 0.1674,
      "step": 10992
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2851822223676088,
      "learning_rate": 1.1502411787814754e-06,
      "loss": 0.1578,
      "step": 10993
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3681006712494275,
      "learning_rate": 1.1495021018547459e-06,
      "loss": 0.1787,
      "step": 10994
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.039762524276647,
      "learning_rate": 1.1487632316074449e-06,
      "loss": 0.5763,
      "step": 10995
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.257111125667077,
      "learning_rate": 1.148024568079235e-06,
      "loss": 0.1777,
      "step": 10996
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1806158417239778,
      "learning_rate": 1.1472861113097616e-06,
      "loss": 0.1375,
      "step": 10997
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1166604448672575,
      "learning_rate": 1.1465478613386654e-06,
      "loss": 0.1201,
      "step": 10998
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2059095530530801,
      "learning_rate": 1.1458098182055693e-06,
      "loss": 0.1489,
      "step": 10999
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4598640185367813,
      "learning_rate": 1.1450719819500906e-06,
      "loss": 0.138,
      "step": 11000
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3073839265477671,
      "learning_rate": 1.1443343526118327e-06,
      "loss": 0.1769,
      "step": 11001
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2853183071958543,
      "learning_rate": 1.1435969302303907e-06,
      "loss": 0.1641,
      "step": 11002
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3743770635188666,
      "learning_rate": 1.1428597148453436e-06,
      "loss": 0.1906,
      "step": 11003
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4398085978963613,
      "learning_rate": 1.1421227064962641e-06,
      "loss": 0.2057,
      "step": 11004
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.406245580981124,
      "learning_rate": 1.1413859052227106e-06,
      "loss": 0.2087,
      "step": 11005
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2956258926391493,
      "learning_rate": 1.1406493110642313e-06,
      "loss": 0.1776,
      "step": 11006
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3252337636594103,
      "learning_rate": 1.1399129240603667e-06,
      "loss": 0.1653,
      "step": 11007
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2198248288399403,
      "learning_rate": 1.1391767442506395e-06,
      "loss": 0.1437,
      "step": 11008
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2558935264984894,
      "learning_rate": 1.1384407716745666e-06,
      "loss": 0.1528,
      "step": 11009
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.122582565894138,
      "learning_rate": 1.1377050063716517e-06,
      "loss": 0.4582,
      "step": 11010
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2766171705690863,
      "learning_rate": 1.1369694483813893e-06,
      "loss": 0.1483,
      "step": 11011
    },
    {
      "epoch": 0.79,
      "grad_norm": 16.524030106707816,
      "learning_rate": 1.1362340977432584e-06,
      "loss": 0.6287,
      "step": 11012
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5272315422516722,
      "learning_rate": 1.1354989544967331e-06,
      "loss": 0.1953,
      "step": 11013
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.719434985355951,
      "learning_rate": 1.134764018681269e-06,
      "loss": 0.4772,
      "step": 11014
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5824412095678229,
      "learning_rate": 1.1340292903363187e-06,
      "loss": 0.2163,
      "step": 11015
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2391756504547187,
      "learning_rate": 1.1332947695013158e-06,
      "loss": 0.151,
      "step": 11016
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3345650914400444,
      "learning_rate": 1.1325604562156877e-06,
      "loss": 0.1584,
      "step": 11017
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.244820217111886,
      "learning_rate": 1.1318263505188504e-06,
      "loss": 0.1525,
      "step": 11018
    },
    {
      "epoch": 0.79,
      "grad_norm": 12.945752358335145,
      "learning_rate": 1.1310924524502082e-06,
      "loss": 0.6471,
      "step": 11019
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.178713029249838,
      "learning_rate": 1.1303587620491513e-06,
      "loss": 0.6066,
      "step": 11020
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3204515922289943,
      "learning_rate": 1.1296252793550644e-06,
      "loss": 0.1691,
      "step": 11021
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3606917110714518,
      "learning_rate": 1.1288920044073148e-06,
      "loss": 0.1978,
      "step": 11022
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2969465212871056,
      "learning_rate": 1.1281589372452644e-06,
      "loss": 0.1532,
      "step": 11023
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1970237613391201,
      "learning_rate": 1.1274260779082597e-06,
      "loss": 0.1601,
      "step": 11024
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1617692053364916,
      "learning_rate": 1.1266934264356378e-06,
      "loss": 0.1424,
      "step": 11025
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3346372964204707,
      "learning_rate": 1.1259609828667246e-06,
      "loss": 0.1489,
      "step": 11026
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3926830046872551,
      "learning_rate": 1.125228747240837e-06,
      "loss": 0.1599,
      "step": 11027
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8564013714541625,
      "learning_rate": 1.1244967195972745e-06,
      "loss": 0.1906,
      "step": 11028
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3837937602297297,
      "learning_rate": 1.1237648999753336e-06,
      "loss": 0.1689,
      "step": 11029
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.198689690643117,
      "learning_rate": 1.1230332884142914e-06,
      "loss": 0.1326,
      "step": 11030
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2956544546503657,
      "learning_rate": 1.1223018849534217e-06,
      "loss": 0.2057,
      "step": 11031
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4489668151242507,
      "learning_rate": 1.12157068963198e-06,
      "loss": 0.1462,
      "step": 11032
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.201700960788802,
      "learning_rate": 1.120839702489216e-06,
      "loss": 0.1626,
      "step": 11033
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4109223765732992,
      "learning_rate": 1.1201089235643653e-06,
      "loss": 0.1758,
      "step": 11034
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.202113185419405,
      "learning_rate": 1.1193783528966552e-06,
      "loss": 0.6505,
      "step": 11035
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3690935597088953,
      "learning_rate": 1.1186479905252962e-06,
      "loss": 0.1842,
      "step": 11036
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1359073666882549,
      "learning_rate": 1.1179178364894949e-06,
      "loss": 0.1423,
      "step": 11037
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4926833009133986,
      "learning_rate": 1.1171878908284406e-06,
      "loss": 0.1831,
      "step": 11038
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2176937435874768,
      "learning_rate": 1.116458153581314e-06,
      "loss": 0.1428,
      "step": 11039
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3570133071606425,
      "learning_rate": 1.1157286247872873e-06,
      "loss": 0.174,
      "step": 11040
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3043834952177298,
      "learning_rate": 1.114999304485515e-06,
      "loss": 0.1889,
      "step": 11041
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4409855169941834,
      "learning_rate": 1.1142701927151456e-06,
      "loss": 0.2024,
      "step": 11042
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5580958977041677,
      "learning_rate": 1.1135412895153153e-06,
      "loss": 0.2257,
      "step": 11043
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5224752770619159,
      "learning_rate": 1.1128125949251494e-06,
      "loss": 0.1832,
      "step": 11044
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3456512075204285,
      "learning_rate": 1.1120841089837592e-06,
      "loss": 0.1542,
      "step": 11045
    },
    {
      "epoch": 0.79,
      "grad_norm": 69.65170279983215,
      "learning_rate": 1.1113558317302498e-06,
      "loss": 0.605,
      "step": 11046
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4040364388772975,
      "learning_rate": 1.1106277632037093e-06,
      "loss": 0.1803,
      "step": 11047
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.157671243520093,
      "learning_rate": 1.10989990344322e-06,
      "loss": 0.1836,
      "step": 11048
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.278594547886093,
      "learning_rate": 1.1091722524878478e-06,
      "loss": 0.1497,
      "step": 11049
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2050138216783188,
      "learning_rate": 1.108444810376652e-06,
      "loss": 0.1509,
      "step": 11050
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3429566855091466,
      "learning_rate": 1.1077175771486781e-06,
      "loss": 0.1546,
      "step": 11051
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.446702395408079,
      "learning_rate": 1.1069905528429631e-06,
      "loss": 0.1879,
      "step": 11052
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4333995907421395,
      "learning_rate": 1.1062637374985273e-06,
      "loss": 0.1997,
      "step": 11053
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3774328351601197,
      "learning_rate": 1.1055371311543872e-06,
      "loss": 0.198,
      "step": 11054
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.411625333503697,
      "learning_rate": 1.10481073384954e-06,
      "loss": 0.1795,
      "step": 11055
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4608589328073691,
      "learning_rate": 1.1040845456229793e-06,
      "loss": 0.1963,
      "step": 11056
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1713571222911876,
      "learning_rate": 1.1033585665136814e-06,
      "loss": 0.1318,
      "step": 11057
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.3446533914960135,
      "learning_rate": 1.102632796560615e-06,
      "loss": 0.6952,
      "step": 11058
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.22123189629881,
      "learning_rate": 1.1019072358027366e-06,
      "loss": 0.1452,
      "step": 11059
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4352019369823124,
      "learning_rate": 1.1011818842789928e-06,
      "loss": 0.1701,
      "step": 11060
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.387554273654965,
      "learning_rate": 1.1004567420283151e-06,
      "loss": 0.1616,
      "step": 11061
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2397862876273045,
      "learning_rate": 1.0997318090896291e-06,
      "loss": 0.1504,
      "step": 11062
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3753208188851656,
      "learning_rate": 1.0990070855018426e-06,
      "loss": 0.1811,
      "step": 11063
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.638540343584256,
      "learning_rate": 1.0982825713038596e-06,
      "loss": 0.6794,
      "step": 11064
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.323078266007812,
      "learning_rate": 1.0975582665345664e-06,
      "loss": 0.1787,
      "step": 11065
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.270777338925711,
      "learning_rate": 1.096834171232841e-06,
      "loss": 0.1659,
      "step": 11066
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.357225567824931,
      "learning_rate": 1.0961102854375515e-06,
      "loss": 0.2232,
      "step": 11067
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.48766266445047,
      "learning_rate": 1.0953866091875542e-06,
      "loss": 0.1977,
      "step": 11068
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4504211063509296,
      "learning_rate": 1.0946631425216897e-06,
      "loss": 0.1835,
      "step": 11069
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3918549185331648,
      "learning_rate": 1.0939398854787925e-06,
      "loss": 0.1959,
      "step": 11070
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3988013896047897,
      "learning_rate": 1.093216838097686e-06,
      "loss": 0.1873,
      "step": 11071
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4320590180105313,
      "learning_rate": 1.0924940004171775e-06,
      "loss": 0.1955,
      "step": 11072
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3737704532253179,
      "learning_rate": 1.091771372476068e-06,
      "loss": 0.1582,
      "step": 11073
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3656059511609862,
      "learning_rate": 1.0910489543131442e-06,
      "loss": 0.2115,
      "step": 11074
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3918198500801657,
      "learning_rate": 1.0903267459671823e-06,
      "loss": 0.1998,
      "step": 11075
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3087138198987212,
      "learning_rate": 1.0896047474769489e-06,
      "loss": 0.163,
      "step": 11076
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4585104126550326,
      "learning_rate": 1.088882958881199e-06,
      "loss": 0.1929,
      "step": 11077
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4827821821431055,
      "learning_rate": 1.0881613802186718e-06,
      "loss": 0.1761,
      "step": 11078
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.633022216853365,
      "learning_rate": 1.0874400115281025e-06,
      "loss": 0.5729,
      "step": 11079
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4014813406281752,
      "learning_rate": 1.0867188528482087e-06,
      "loss": 0.1515,
      "step": 11080
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.345593055023765,
      "learning_rate": 1.085997904217701e-06,
      "loss": 0.1966,
      "step": 11081
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4623345013369717,
      "learning_rate": 1.0852771656752753e-06,
      "loss": 0.1353,
      "step": 11082
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.181801533691612,
      "learning_rate": 1.0845566372596185e-06,
      "loss": 0.1493,
      "step": 11083
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5503009645819674,
      "learning_rate": 1.0838363190094065e-06,
      "loss": 0.1989,
      "step": 11084
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.380675386245187,
      "learning_rate": 1.0831162109633043e-06,
      "loss": 0.165,
      "step": 11085
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4117785828975489,
      "learning_rate": 1.0823963131599614e-06,
      "loss": 0.2174,
      "step": 11086
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2076115031588082,
      "learning_rate": 1.0816766256380218e-06,
      "loss": 0.1603,
      "step": 11087
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3552259275329406,
      "learning_rate": 1.080957148436113e-06,
      "loss": 0.1364,
      "step": 11088
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.321282782357995,
      "learning_rate": 1.0802378815928565e-06,
      "loss": 0.1669,
      "step": 11089
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1991648110365258,
      "learning_rate": 1.0795188251468569e-06,
      "loss": 0.1421,
      "step": 11090
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2779882611443147,
      "learning_rate": 1.0787999791367126e-06,
      "loss": 0.1689,
      "step": 11091
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2745288583651153,
      "learning_rate": 1.0780813436010057e-06,
      "loss": 0.1631,
      "step": 11092
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2890327840751468,
      "learning_rate": 1.077362918578312e-06,
      "loss": 0.2025,
      "step": 11093
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2252457671139332,
      "learning_rate": 1.076644704107192e-06,
      "loss": 0.1836,
      "step": 11094
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.128840446846303,
      "learning_rate": 1.0759267002261997e-06,
      "loss": 0.4893,
      "step": 11095
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3650289528342299,
      "learning_rate": 1.0752089069738713e-06,
      "loss": 0.1574,
      "step": 11096
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.405440071365288,
      "learning_rate": 1.0744913243887379e-06,
      "loss": 0.149,
      "step": 11097
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3214571934044328,
      "learning_rate": 1.0737739525093132e-06,
      "loss": 0.1634,
      "step": 11098
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.219696713589709,
      "learning_rate": 1.0730567913741063e-06,
      "loss": 0.1502,
      "step": 11099
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.478619192817674,
      "learning_rate": 1.0723398410216085e-06,
      "loss": 0.4887,
      "step": 11100
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.444602108478056,
      "learning_rate": 1.0716231014903038e-06,
      "loss": 0.162,
      "step": 11101
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3337560119010048,
      "learning_rate": 1.070906572818665e-06,
      "loss": 0.1863,
      "step": 11102
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3724028626408107,
      "learning_rate": 1.0701902550451516e-06,
      "loss": 0.1607,
      "step": 11103
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3776270531710393,
      "learning_rate": 1.0694741482082144e-06,
      "loss": 0.1654,
      "step": 11104
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2827369049576012,
      "learning_rate": 1.0687582523462887e-06,
      "loss": 0.1689,
      "step": 11105
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.843595525987163,
      "learning_rate": 1.0680425674978029e-06,
      "loss": 0.5041,
      "step": 11106
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.59237598848215,
      "learning_rate": 1.0673270937011703e-06,
      "loss": 0.5283,
      "step": 11107
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1169470440021918,
      "learning_rate": 1.066611830994797e-06,
      "loss": 0.1491,
      "step": 11108
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3723220396336342,
      "learning_rate": 1.0658967794170721e-06,
      "loss": 0.197,
      "step": 11109
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.459623452660205,
      "learning_rate": 1.0651819390063794e-06,
      "loss": 0.188,
      "step": 11110
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1663476028625093,
      "learning_rate": 1.0644673098010878e-06,
      "loss": 0.169,
      "step": 11111
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2179626154463086,
      "learning_rate": 1.0637528918395572e-06,
      "loss": 0.156,
      "step": 11112
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.292982170832504,
      "learning_rate": 1.0630386851601327e-06,
      "loss": 0.1779,
      "step": 11113
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.99203663331111,
      "learning_rate": 1.0623246898011514e-06,
      "loss": 0.6062,
      "step": 11114
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3551704713085768,
      "learning_rate": 1.0616109058009361e-06,
      "loss": 0.1375,
      "step": 11115
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3534551921940863,
      "learning_rate": 1.0608973331978028e-06,
      "loss": 0.169,
      "step": 11116
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0725766827610275,
      "learning_rate": 1.0601839720300495e-06,
      "loss": 0.1332,
      "step": 11117
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.28929409911985,
      "learning_rate": 1.059470822335969e-06,
      "loss": 0.1583,
      "step": 11118
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5052511454209667,
      "learning_rate": 1.0587578841538392e-06,
      "loss": 0.2082,
      "step": 11119
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2006494247886867,
      "learning_rate": 1.0580451575219304e-06,
      "loss": 0.1316,
      "step": 11120
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3878190942866944,
      "learning_rate": 1.0573326424784957e-06,
      "loss": 0.1848,
      "step": 11121
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3941424850751762,
      "learning_rate": 1.0566203390617823e-06,
      "loss": 0.1956,
      "step": 11122
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.477315206105067,
      "learning_rate": 1.0559082473100218e-06,
      "loss": 0.2041,
      "step": 11123
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3398934024864189,
      "learning_rate": 1.0551963672614385e-06,
      "loss": 0.1624,
      "step": 11124
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.101342489803704,
      "learning_rate": 1.0544846989542413e-06,
      "loss": 0.1433,
      "step": 11125
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5011513538676027,
      "learning_rate": 1.0537732424266311e-06,
      "loss": 0.2143,
      "step": 11126
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.230979814473399,
      "learning_rate": 1.0530619977167949e-06,
      "loss": 0.126,
      "step": 11127
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.435994810853229,
      "learning_rate": 1.0523509648629126e-06,
      "loss": 0.1897,
      "step": 11128
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1856642966476643,
      "learning_rate": 1.0516401439031453e-06,
      "loss": 0.158,
      "step": 11129
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4782973294687232,
      "learning_rate": 1.0509295348756503e-06,
      "loss": 0.2386,
      "step": 11130
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5197555220442451,
      "learning_rate": 1.0502191378185684e-06,
      "loss": 0.1847,
      "step": 11131
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3630595073131238,
      "learning_rate": 1.0495089527700325e-06,
      "loss": 0.1901,
      "step": 11132
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.015952216179803,
      "learning_rate": 1.0487989797681602e-06,
      "loss": 0.5251,
      "step": 11133
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.386678523195115,
      "learning_rate": 1.048089218851061e-06,
      "loss": 0.4171,
      "step": 11134
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2609467831012091,
      "learning_rate": 1.0473796700568328e-06,
      "loss": 0.15,
      "step": 11135
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3926719380068098,
      "learning_rate": 1.046670333423561e-06,
      "loss": 0.1668,
      "step": 11136
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2421880562324485,
      "learning_rate": 1.0459612089893217e-06,
      "loss": 0.1724,
      "step": 11137
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5600975396715222,
      "learning_rate": 1.0452522967921742e-06,
      "loss": 0.1893,
      "step": 11138
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5979140395794278,
      "learning_rate": 1.0445435968701733e-06,
      "loss": 0.1927,
      "step": 11139
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3501946618131464,
      "learning_rate": 1.043835109261357e-06,
      "loss": 0.175,
      "step": 11140
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2399340308141287,
      "learning_rate": 1.0431268340037565e-06,
      "loss": 0.1642,
      "step": 11141
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4959960914038462,
      "learning_rate": 1.0424187711353861e-06,
      "loss": 0.1607,
      "step": 11142
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5200543089606962,
      "learning_rate": 1.0417109206942533e-06,
      "loss": 0.198,
      "step": 11143
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3721799523865998,
      "learning_rate": 1.041003282718353e-06,
      "loss": 0.184,
      "step": 11144
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3275147752498082,
      "learning_rate": 1.0402958572456695e-06,
      "loss": 0.1808,
      "step": 11145
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4089808414788942,
      "learning_rate": 1.0395886443141718e-06,
      "loss": 0.1857,
      "step": 11146
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.74194148913727,
      "learning_rate": 1.0388816439618232e-06,
      "loss": 0.2455,
      "step": 11147
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2131626198331975,
      "learning_rate": 1.0381748562265704e-06,
      "loss": 0.1161,
      "step": 11148
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2857279175384386,
      "learning_rate": 1.0374682811463527e-06,
      "loss": 0.1653,
      "step": 11149
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3249149161539553,
      "learning_rate": 1.0367619187590939e-06,
      "loss": 0.1363,
      "step": 11150
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4291198725742187,
      "learning_rate": 1.0360557691027107e-06,
      "loss": 0.1885,
      "step": 11151
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.5946607914372644,
      "learning_rate": 1.0353498322151057e-06,
      "loss": 0.4811,
      "step": 11152
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2488817922310351,
      "learning_rate": 1.0346441081341719e-06,
      "loss": 0.1577,
      "step": 11153
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.184795198872465,
      "learning_rate": 1.0339385968977878e-06,
      "loss": 0.1585,
      "step": 11154
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4709592961050704,
      "learning_rate": 1.0332332985438248e-06,
      "loss": 0.2251,
      "step": 11155
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.166193031134425,
      "learning_rate": 1.0325282131101382e-06,
      "loss": 0.1486,
      "step": 11156
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4222976848670972,
      "learning_rate": 1.0318233406345763e-06,
      "loss": 0.1652,
      "step": 11157
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3450064486780235,
      "learning_rate": 1.031118681154971e-06,
      "loss": 0.1901,
      "step": 11158
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5101419180734172,
      "learning_rate": 1.0304142347091477e-06,
      "loss": 0.1747,
      "step": 11159
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.250474933122123,
      "learning_rate": 1.0297100013349181e-06,
      "loss": 0.1483,
      "step": 11160
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.347034506360916,
      "learning_rate": 1.0290059810700841e-06,
      "loss": 0.1706,
      "step": 11161
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3282864529891576,
      "learning_rate": 1.0283021739524313e-06,
      "loss": 0.1812,
      "step": 11162
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3180546110488653,
      "learning_rate": 1.0275985800197408e-06,
      "loss": 0.1522,
      "step": 11163
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.674186260902461,
      "learning_rate": 1.0268951993097753e-06,
      "loss": 0.5349,
      "step": 11164
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4010563696600737,
      "learning_rate": 1.0261920318602924e-06,
      "loss": 0.1947,
      "step": 11165
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2543933809436032,
      "learning_rate": 1.0254890777090331e-06,
      "loss": 0.1673,
      "step": 11166
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4006686664352774,
      "learning_rate": 1.0247863368937306e-06,
      "loss": 0.1905,
      "step": 11167
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.259052746900076,
      "learning_rate": 1.024083809452104e-06,
      "loss": 0.1598,
      "step": 11168
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3489387450139074,
      "learning_rate": 1.0233814954218634e-06,
      "loss": 0.182,
      "step": 11169
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3618575760577665,
      "learning_rate": 1.022679394840707e-06,
      "loss": 0.1602,
      "step": 11170
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2222195453749545,
      "learning_rate": 1.021977507746319e-06,
      "loss": 0.154,
      "step": 11171
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2776335246185315,
      "learning_rate": 1.0212758341763752e-06,
      "loss": 0.1625,
      "step": 11172
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3360590491440565,
      "learning_rate": 1.020574374168537e-06,
      "loss": 0.1679,
      "step": 11173
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.401117924397328,
      "learning_rate": 1.019873127760458e-06,
      "loss": 0.1708,
      "step": 11174
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0886023085883567,
      "learning_rate": 1.019172094989776e-06,
      "loss": 0.1457,
      "step": 11175
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.932938684432314,
      "learning_rate": 1.0184712758941212e-06,
      "loss": 0.6062,
      "step": 11176
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3562633741420573,
      "learning_rate": 1.0177706705111106e-06,
      "loss": 0.1602,
      "step": 11177
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.662255864120727,
      "learning_rate": 1.0170702788783509e-06,
      "loss": 0.5371,
      "step": 11178
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2162021078672893,
      "learning_rate": 1.0163701010334343e-06,
      "loss": 0.1283,
      "step": 11179
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4955097136095414,
      "learning_rate": 1.0156701370139454e-06,
      "loss": 0.1915,
      "step": 11180
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2831355204712909,
      "learning_rate": 1.0149703868574533e-06,
      "loss": 0.1446,
      "step": 11181
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.470990025490076,
      "learning_rate": 1.0142708506015208e-06,
      "loss": 0.158,
      "step": 11182
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3280912515408867,
      "learning_rate": 1.0135715282836933e-06,
      "loss": 0.1816,
      "step": 11183
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.370992289139157,
      "learning_rate": 1.0128724199415092e-06,
      "loss": 0.1427,
      "step": 11184
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.35704111104011,
      "learning_rate": 1.0121735256124925e-06,
      "loss": 0.5628,
      "step": 11185
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2724302697056273,
      "learning_rate": 1.0114748453341605e-06,
      "loss": 0.1295,
      "step": 11186
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3611937819286883,
      "learning_rate": 1.0107763791440118e-06,
      "loss": 0.1524,
      "step": 11187
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4750807001387294,
      "learning_rate": 1.0100781270795396e-06,
      "loss": 0.2131,
      "step": 11188
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4206421569999124,
      "learning_rate": 1.0093800891782214e-06,
      "loss": 0.181,
      "step": 11189
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2959495947946629,
      "learning_rate": 1.0086822654775274e-06,
      "loss": 0.1459,
      "step": 11190
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4925786311774873,
      "learning_rate": 1.0079846560149115e-06,
      "loss": 0.1727,
      "step": 11191
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3296792006644287,
      "learning_rate": 1.0072872608278206e-06,
      "loss": 0.1878,
      "step": 11192
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3846123085232542,
      "learning_rate": 1.0065900799536866e-06,
      "loss": 0.209,
      "step": 11193
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4740125596429086,
      "learning_rate": 1.0058931134299322e-06,
      "loss": 0.1693,
      "step": 11194
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.276414210041604,
      "learning_rate": 1.005196361293967e-06,
      "loss": 0.5279,
      "step": 11195
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2576006925643655,
      "learning_rate": 1.0044998235831927e-06,
      "loss": 0.1605,
      "step": 11196
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.749871280992744,
      "learning_rate": 1.0038035003349927e-06,
      "loss": 0.2544,
      "step": 11197
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1688265932300483,
      "learning_rate": 1.003107391586745e-06,
      "loss": 0.1449,
      "step": 11198
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3952724008586497,
      "learning_rate": 1.0024114973758154e-06,
      "loss": 0.1621,
      "step": 11199
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1960062898824,
      "learning_rate": 1.0017158177395531e-06,
      "loss": 0.1468,
      "step": 11200
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.429594431442411,
      "learning_rate": 1.0010203527153034e-06,
      "loss": 0.204,
      "step": 11201
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5348639876914003,
      "learning_rate": 1.000325102340392e-06,
      "loss": 0.2015,
      "step": 11202
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3688547526533104,
      "learning_rate": 9.996300666521397e-07,
      "loss": 0.205,
      "step": 11203
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4346566499001738,
      "learning_rate": 9.989352456878525e-07,
      "loss": 0.1699,
      "step": 11204
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3940865477919593,
      "learning_rate": 9.982406394848276e-07,
      "loss": 0.1598,
      "step": 11205
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3621843413009298,
      "learning_rate": 9.975462480803456e-07,
      "loss": 0.1691,
      "step": 11206
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1117209091608427,
      "learning_rate": 9.968520715116808e-07,
      "loss": 0.1443,
      "step": 11207
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3576463957783502,
      "learning_rate": 9.961581098160928e-07,
      "loss": 0.1776,
      "step": 11208
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.407979831960541,
      "learning_rate": 9.954643630308318e-07,
      "loss": 0.1924,
      "step": 11209
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.378835060706294,
      "learning_rate": 9.947708311931332e-07,
      "loss": 0.1883,
      "step": 11210
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.225115649224681,
      "learning_rate": 9.940775143402249e-07,
      "loss": 0.1461,
      "step": 11211
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.83665480730355,
      "learning_rate": 9.933844125093206e-07,
      "loss": 0.5009,
      "step": 11212
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.29434453618675,
      "learning_rate": 9.926915257376252e-07,
      "loss": 0.1538,
      "step": 11213
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3769176590798835,
      "learning_rate": 9.919988540623264e-07,
      "loss": 0.1624,
      "step": 11214
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2728529246022755,
      "learning_rate": 9.913063975206082e-07,
      "loss": 0.1814,
      "step": 11215
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3294890564974637,
      "learning_rate": 9.906141561496358e-07,
      "loss": 0.1757,
      "step": 11216
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.309861456377614,
      "learning_rate": 9.89922129986568e-07,
      "loss": 0.1602,
      "step": 11217
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5564195564796117,
      "learning_rate": 9.892303190685476e-07,
      "loss": 0.2072,
      "step": 11218
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3078303661955493,
      "learning_rate": 9.885387234327099e-07,
      "loss": 0.1883,
      "step": 11219
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3431925986799484,
      "learning_rate": 9.878473431161767e-07,
      "loss": 0.1252,
      "step": 11220
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4352237959940475,
      "learning_rate": 9.871561781560606e-07,
      "loss": 0.1584,
      "step": 11221
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3538200219329861,
      "learning_rate": 9.864652285894567e-07,
      "loss": 0.1889,
      "step": 11222
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4755556689624574,
      "learning_rate": 9.857744944534559e-07,
      "loss": 0.2228,
      "step": 11223
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4599485264010472,
      "learning_rate": 9.85083975785131e-07,
      "loss": 0.1875,
      "step": 11224
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.33836799638902,
      "learning_rate": 9.843936726215498e-07,
      "loss": 0.1946,
      "step": 11225
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.151640439349557,
      "learning_rate": 9.837035849997611e-07,
      "loss": 0.1372,
      "step": 11226
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2368020370743107,
      "learning_rate": 9.830137129568086e-07,
      "loss": 0.149,
      "step": 11227
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5137469651441258,
      "learning_rate": 9.82324056529721e-07,
      "loss": 0.2038,
      "step": 11228
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4635798524396362,
      "learning_rate": 9.816346157555184e-07,
      "loss": 0.1866,
      "step": 11229
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3918238980125193,
      "learning_rate": 9.80945390671204e-07,
      "loss": 0.1758,
      "step": 11230
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3695709245431835,
      "learning_rate": 9.802563813137738e-07,
      "loss": 0.1501,
      "step": 11231
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4978861349992698,
      "learning_rate": 9.795675877202132e-07,
      "loss": 0.2177,
      "step": 11232
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.548952476180618,
      "learning_rate": 9.78879009927491e-07,
      "loss": 0.346,
      "step": 11233
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2739197322761147,
      "learning_rate": 9.781906479725688e-07,
      "loss": 0.1628,
      "step": 11234
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4346648445233583,
      "learning_rate": 9.775025018923946e-07,
      "loss": 0.1784,
      "step": 11235
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1356241160994607,
      "learning_rate": 9.768145717239052e-07,
      "loss": 0.1552,
      "step": 11236
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4521472336792134,
      "learning_rate": 9.76126857504026e-07,
      "loss": 0.1795,
      "step": 11237
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.335878244633931,
      "learning_rate": 9.754393592696732e-07,
      "loss": 0.1864,
      "step": 11238
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.110333520921348,
      "learning_rate": 9.747520770577447e-07,
      "loss": 0.1605,
      "step": 11239
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1232923553732563,
      "learning_rate": 9.740650109051348e-07,
      "loss": 0.1665,
      "step": 11240
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.264057476381941,
      "learning_rate": 9.733781608487196e-07,
      "loss": 0.166,
      "step": 11241
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1511081259839442,
      "learning_rate": 9.726915269253695e-07,
      "loss": 0.1377,
      "step": 11242
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3396523826975002,
      "learning_rate": 9.720051091719369e-07,
      "loss": 0.1467,
      "step": 11243
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3083494937943594,
      "learning_rate": 9.713189076252676e-07,
      "loss": 0.1532,
      "step": 11244
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3724053453700986,
      "learning_rate": 9.706329223221945e-07,
      "loss": 0.1816,
      "step": 11245
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3597827551168278,
      "learning_rate": 9.699471532995397e-07,
      "loss": 0.1976,
      "step": 11246
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2291124590731324,
      "learning_rate": 9.692616005941098e-07,
      "loss": 0.1569,
      "step": 11247
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3197276477827402,
      "learning_rate": 9.685762642427048e-07,
      "loss": 0.1861,
      "step": 11248
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4975893590509837,
      "learning_rate": 9.678911442821087e-07,
      "loss": 0.1603,
      "step": 11249
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.402325256491505,
      "learning_rate": 9.67206240749099e-07,
      "loss": 0.1542,
      "step": 11250
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2314936044579967,
      "learning_rate": 9.66521553680435e-07,
      "loss": 0.1485,
      "step": 11251
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.362892631002434,
      "learning_rate": 9.658370831128704e-07,
      "loss": 0.1634,
      "step": 11252
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.579815020615234,
      "learning_rate": 9.65152829083144e-07,
      "loss": 0.5307,
      "step": 11253
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.305906845943606,
      "learning_rate": 9.644687916279854e-07,
      "loss": 0.147,
      "step": 11254
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3556620853025283,
      "learning_rate": 9.63784970784109e-07,
      "loss": 0.1523,
      "step": 11255
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4651132255342085,
      "learning_rate": 9.63101366588221e-07,
      "loss": 0.1841,
      "step": 11256
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.40862887362798,
      "learning_rate": 9.624179790770126e-07,
      "loss": 0.1672,
      "step": 11257
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.282911274062553,
      "learning_rate": 9.61734808287168e-07,
      "loss": 0.142,
      "step": 11258
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.352214193267794,
      "learning_rate": 9.610518542553549e-07,
      "loss": 0.1798,
      "step": 11259
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2273075770141646,
      "learning_rate": 9.603691170182316e-07,
      "loss": 0.1727,
      "step": 11260
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3260003560402982,
      "learning_rate": 9.596865966124463e-07,
      "loss": 0.1816,
      "step": 11261
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2908776147626249,
      "learning_rate": 9.590042930746323e-07,
      "loss": 0.1493,
      "step": 11262
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.331236886220097,
      "learning_rate": 9.58322206441416e-07,
      "loss": 0.2001,
      "step": 11263
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3262849837999742,
      "learning_rate": 9.576403367494054e-07,
      "loss": 0.1562,
      "step": 11264
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4402669358174165,
      "learning_rate": 9.56958684035203e-07,
      "loss": 0.1695,
      "step": 11265
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.382882190497153,
      "learning_rate": 9.562772483353949e-07,
      "loss": 0.1925,
      "step": 11266
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3225653988588857,
      "learning_rate": 9.555960296865614e-07,
      "loss": 0.1717,
      "step": 11267
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4205788326676936,
      "learning_rate": 9.549150281252633e-07,
      "loss": 0.2058,
      "step": 11268
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4707299829483176,
      "learning_rate": 9.542342436880564e-07,
      "loss": 0.1964,
      "step": 11269
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2425641157829774,
      "learning_rate": 9.535536764114827e-07,
      "loss": 0.1819,
      "step": 11270
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3950987545298927,
      "learning_rate": 9.528733263320727e-07,
      "loss": 0.1512,
      "step": 11271
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1862549024993225,
      "learning_rate": 9.521931934863432e-07,
      "loss": 0.1455,
      "step": 11272
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4972549067220295,
      "learning_rate": 9.515132779108033e-07,
      "loss": 0.1728,
      "step": 11273
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3870091575373642,
      "learning_rate": 9.508335796419454e-07,
      "loss": 0.198,
      "step": 11274
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3137581405691436,
      "learning_rate": 9.501540987162561e-07,
      "loss": 0.176,
      "step": 11275
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3449511853136782,
      "learning_rate": 9.494748351702038e-07,
      "loss": 0.1858,
      "step": 11276
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.4978959231228535,
      "learning_rate": 9.48795789040251e-07,
      "loss": 0.4166,
      "step": 11277
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2963616011920063,
      "learning_rate": 9.481169603628454e-07,
      "loss": 0.1482,
      "step": 11278
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1360199419642687,
      "learning_rate": 9.474383491744255e-07,
      "loss": 0.1395,
      "step": 11279
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3867549606824945,
      "learning_rate": 9.467599555114137e-07,
      "loss": 0.1863,
      "step": 11280
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4013565827116152,
      "learning_rate": 9.460817794102262e-07,
      "loss": 0.1515,
      "step": 11281
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.215970843830103,
      "learning_rate": 9.454038209072619e-07,
      "loss": 0.141,
      "step": 11282
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1888310603015435,
      "learning_rate": 9.447260800389135e-07,
      "loss": 0.1565,
      "step": 11283
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.44893397468075,
      "learning_rate": 9.440485568415575e-07,
      "loss": 0.5472,
      "step": 11284
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3624428742966304,
      "learning_rate": 9.433712513515624e-07,
      "loss": 0.1868,
      "step": 11285
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2468534220331826,
      "learning_rate": 9.426941636052805e-07,
      "loss": 0.1527,
      "step": 11286
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.308190581144608,
      "learning_rate": 9.420172936390592e-07,
      "loss": 0.1533,
      "step": 11287
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4393027433743304,
      "learning_rate": 9.413406414892268e-07,
      "loss": 0.1708,
      "step": 11288
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2942717252069884,
      "learning_rate": 9.406642071921062e-07,
      "loss": 0.2029,
      "step": 11289
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.526307482877949,
      "learning_rate": 9.39987990784002e-07,
      "loss": 0.1905,
      "step": 11290
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3918439999240453,
      "learning_rate": 9.39311992301215e-07,
      "loss": 0.1627,
      "step": 11291
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2534505998061605,
      "learning_rate": 9.386362117800262e-07,
      "loss": 0.1401,
      "step": 11292
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3368209071596542,
      "learning_rate": 9.379606492567122e-07,
      "loss": 0.1587,
      "step": 11293
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.27340979791407,
      "learning_rate": 9.372853047675318e-07,
      "loss": 0.203,
      "step": 11294
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.102416515403265,
      "learning_rate": 9.366101783487352e-07,
      "loss": 0.1283,
      "step": 11295
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4825521752301705,
      "learning_rate": 9.35935270036562e-07,
      "loss": 0.1869,
      "step": 11296
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3643023559245617,
      "learning_rate": 9.352605798672377e-07,
      "loss": 0.2059,
      "step": 11297
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4065346423057608,
      "learning_rate": 9.34586107876978e-07,
      "loss": 0.1707,
      "step": 11298
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4342465639184196,
      "learning_rate": 9.339118541019837e-07,
      "loss": 0.1836,
      "step": 11299
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2939700053869756,
      "learning_rate": 9.332378185784491e-07,
      "loss": 0.2016,
      "step": 11300
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4438399744559105,
      "learning_rate": 9.325640013425502e-07,
      "loss": 0.183,
      "step": 11301
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4061487188463309,
      "learning_rate": 9.318904024304587e-07,
      "loss": 0.1844,
      "step": 11302
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5700146340377061,
      "learning_rate": 9.312170218783268e-07,
      "loss": 0.251,
      "step": 11303
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2820168252746016,
      "learning_rate": 9.30543859722301e-07,
      "loss": 0.1501,
      "step": 11304
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.6487025389185614,
      "learning_rate": 9.298709159985131e-07,
      "loss": 0.591,
      "step": 11305
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4385275887805435,
      "learning_rate": 9.291981907430864e-07,
      "loss": 0.185,
      "step": 11306
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4779126918065233,
      "learning_rate": 9.285256839921275e-07,
      "loss": 0.2095,
      "step": 11307
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3975371211484617,
      "learning_rate": 9.278533957817354e-07,
      "loss": 0.1601,
      "step": 11308
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3515387290452283,
      "learning_rate": 9.271813261479945e-07,
      "loss": 0.1824,
      "step": 11309
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3197301239519237,
      "learning_rate": 9.265094751269804e-07,
      "loss": 0.1637,
      "step": 11310
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4255502567030556,
      "learning_rate": 9.258378427547532e-07,
      "loss": 0.1857,
      "step": 11311
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.734510015504786,
      "learning_rate": 9.251664290673651e-07,
      "loss": 0.5718,
      "step": 11312
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1925586206748733,
      "learning_rate": 9.244952341008545e-07,
      "loss": 0.1304,
      "step": 11313
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3335609167361322,
      "learning_rate": 9.238242578912505e-07,
      "loss": 0.1987,
      "step": 11314
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.239442218519999,
      "learning_rate": 9.231535004745645e-07,
      "loss": 0.1804,
      "step": 11315
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1431423655919553,
      "learning_rate": 9.224829618868037e-07,
      "loss": 0.1191,
      "step": 11316
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3757847730555146,
      "learning_rate": 9.218126421639572e-07,
      "loss": 0.1862,
      "step": 11317
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.275887073863003,
      "learning_rate": 9.211425413420072e-07,
      "loss": 0.5805,
      "step": 11318
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3749864353155,
      "learning_rate": 9.204726594569202e-07,
      "loss": 0.1439,
      "step": 11319
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3130767003706008,
      "learning_rate": 9.198029965446537e-07,
      "loss": 0.1541,
      "step": 11320
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4830886576452098,
      "learning_rate": 9.191335526411527e-07,
      "loss": 0.2086,
      "step": 11321
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.545185840944322,
      "learning_rate": 9.184643277823512e-07,
      "loss": 0.1933,
      "step": 11322
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.186782484786305,
      "learning_rate": 9.17795322004168e-07,
      "loss": 0.1504,
      "step": 11323
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4537303817645482,
      "learning_rate": 9.171265353425158e-07,
      "loss": 0.1636,
      "step": 11324
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.399512774088883,
      "learning_rate": 9.164579678332897e-07,
      "loss": 0.1431,
      "step": 11325
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3082206393433835,
      "learning_rate": 9.157896195123767e-07,
      "loss": 0.1426,
      "step": 11326
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.108821635545509,
      "learning_rate": 9.151214904156525e-07,
      "loss": 0.1407,
      "step": 11327
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.330332453553592,
      "learning_rate": 9.144535805789767e-07,
      "loss": 0.1744,
      "step": 11328
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.436775348103306,
      "learning_rate": 9.137858900382018e-07,
      "loss": 0.1825,
      "step": 11329
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3685097889302205,
      "learning_rate": 9.13118418829167e-07,
      "loss": 0.1814,
      "step": 11330
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.7935673906164675,
      "learning_rate": 9.124511669877001e-07,
      "loss": 0.5478,
      "step": 11331
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1650898738466133,
      "learning_rate": 9.117841345496142e-07,
      "loss": 0.1319,
      "step": 11332
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2585985478773416,
      "learning_rate": 9.111173215507164e-07,
      "loss": 0.1899,
      "step": 11333
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4357708877913766,
      "learning_rate": 9.104507280267944e-07,
      "loss": 0.154,
      "step": 11334
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3207506081304636,
      "learning_rate": 9.09784354013632e-07,
      "loss": 0.1913,
      "step": 11335
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5386178379533013,
      "learning_rate": 9.091181995469945e-07,
      "loss": 0.1521,
      "step": 11336
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4545662678168756,
      "learning_rate": 9.084522646626398e-07,
      "loss": 0.158,
      "step": 11337
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.75007569615903,
      "learning_rate": 9.077865493963129e-07,
      "loss": 0.5576,
      "step": 11338
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3659008162949513,
      "learning_rate": 9.071210537837477e-07,
      "loss": 0.1825,
      "step": 11339
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.067957807273403,
      "learning_rate": 9.064557778606631e-07,
      "loss": 0.4836,
      "step": 11340
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1863284559765537,
      "learning_rate": 9.057907216627703e-07,
      "loss": 0.1659,
      "step": 11341
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.329821303008681,
      "learning_rate": 9.051258852257655e-07,
      "loss": 0.1701,
      "step": 11342
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3625375559348458,
      "learning_rate": 9.04461268585336e-07,
      "loss": 0.1865,
      "step": 11343
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5490187725445175,
      "learning_rate": 9.037968717771539e-07,
      "loss": 0.2056,
      "step": 11344
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3031422435958897,
      "learning_rate": 9.031326948368824e-07,
      "loss": 0.1886,
      "step": 11345
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5742537878036162,
      "learning_rate": 9.024687378001712e-07,
      "loss": 0.1866,
      "step": 11346
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.197228067563145,
      "learning_rate": 9.018050007026613e-07,
      "loss": 0.1239,
      "step": 11347
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3414003794383405,
      "learning_rate": 9.011414835799764e-07,
      "loss": 0.1841,
      "step": 11348
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5361535567940363,
      "learning_rate": 9.004781864677342e-07,
      "loss": 0.2063,
      "step": 11349
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4741188672701016,
      "learning_rate": 8.99815109401535e-07,
      "loss": 0.1833,
      "step": 11350
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3197866264161884,
      "learning_rate": 8.991522524169726e-07,
      "loss": 0.1592,
      "step": 11351
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5314598928140957,
      "learning_rate": 8.984896155496242e-07,
      "loss": 0.211,
      "step": 11352
    },
    {
      "epoch": 0.81,
      "grad_norm": 10.295104709775284,
      "learning_rate": 8.97827198835059e-07,
      "loss": 0.5759,
      "step": 11353
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2320781253601287,
      "learning_rate": 8.971650023088329e-07,
      "loss": 0.1293,
      "step": 11354
    },
    {
      "epoch": 0.81,
      "grad_norm": 19.50510936719855,
      "learning_rate": 8.965030260064906e-07,
      "loss": 0.4975,
      "step": 11355
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.72883765476052,
      "learning_rate": 8.958412699635627e-07,
      "loss": 0.5195,
      "step": 11356
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.377114862755084,
      "learning_rate": 8.951797342155716e-07,
      "loss": 0.1801,
      "step": 11357
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2830103403679676,
      "learning_rate": 8.945184187980233e-07,
      "loss": 0.1922,
      "step": 11358
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2304614256782398,
      "learning_rate": 8.93857323746416e-07,
      "loss": 0.1532,
      "step": 11359
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3953649092550644,
      "learning_rate": 8.931964490962364e-07,
      "loss": 0.1463,
      "step": 11360
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2127193347416148,
      "learning_rate": 8.925357948829544e-07,
      "loss": 0.1451,
      "step": 11361
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3575276153551967,
      "learning_rate": 8.918753611420328e-07,
      "loss": 0.1868,
      "step": 11362
    },
    {
      "epoch": 0.81,
      "grad_norm": 12.052468043288988,
      "learning_rate": 8.912151479089215e-07,
      "loss": 0.5023,
      "step": 11363
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2873481785687018,
      "learning_rate": 8.905551552190589e-07,
      "loss": 0.1368,
      "step": 11364
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3328600272095767,
      "learning_rate": 8.898953831078678e-07,
      "loss": 0.1892,
      "step": 11365
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1123152633721056,
      "learning_rate": 8.892358316107658e-07,
      "loss": 0.1217,
      "step": 11366
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3636489877997258,
      "learning_rate": 8.885765007631514e-07,
      "loss": 0.1891,
      "step": 11367
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1376326174490177,
      "learning_rate": 8.879173906004184e-07,
      "loss": 0.1352,
      "step": 11368
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.395818120418276,
      "learning_rate": 8.872585011579416e-07,
      "loss": 0.1755,
      "step": 11369
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.281478242564968,
      "learning_rate": 8.8659983247109e-07,
      "loss": 0.1797,
      "step": 11370
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3361140416471964,
      "learning_rate": 8.859413845752174e-07,
      "loss": 0.1534,
      "step": 11371
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4126760701955117,
      "learning_rate": 8.852831575056681e-07,
      "loss": 0.1629,
      "step": 11372
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3519387220846326,
      "learning_rate": 8.846251512977711e-07,
      "loss": 0.1957,
      "step": 11373
    },
    {
      "epoch": 0.81,
      "grad_norm": 10.412770170307391,
      "learning_rate": 8.839673659868475e-07,
      "loss": 0.5328,
      "step": 11374
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7326474782113603,
      "learning_rate": 8.833098016082025e-07,
      "loss": 0.1764,
      "step": 11375
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3436427884261881,
      "learning_rate": 8.826524581971335e-07,
      "loss": 0.1684,
      "step": 11376
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.396479269852451,
      "learning_rate": 8.819953357889221e-07,
      "loss": 0.1467,
      "step": 11377
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5507408245191352,
      "learning_rate": 8.813384344188414e-07,
      "loss": 0.1943,
      "step": 11378
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3410763796504985,
      "learning_rate": 8.806817541221507e-07,
      "loss": 0.1861,
      "step": 11379
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3681216249588533,
      "learning_rate": 8.800252949340998e-07,
      "loss": 0.1628,
      "step": 11380
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2029895125939711,
      "learning_rate": 8.793690568899216e-07,
      "loss": 0.1412,
      "step": 11381
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3565488484412431,
      "learning_rate": 8.787130400248434e-07,
      "loss": 0.154,
      "step": 11382
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3403941249537674,
      "learning_rate": 8.780572443740754e-07,
      "loss": 0.1762,
      "step": 11383
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2628129983635457,
      "learning_rate": 8.774016699728194e-07,
      "loss": 0.1664,
      "step": 11384
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.45450465953174,
      "learning_rate": 8.767463168562628e-07,
      "loss": 0.1761,
      "step": 11385
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1698539011781106,
      "learning_rate": 8.760911850595844e-07,
      "loss": 0.1363,
      "step": 11386
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2393681878220268,
      "learning_rate": 8.754362746179445e-07,
      "loss": 0.1608,
      "step": 11387
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.786102518694291,
      "learning_rate": 8.747815855665026e-07,
      "loss": 0.1773,
      "step": 11388
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5191709697516351,
      "learning_rate": 8.741271179403954e-07,
      "loss": 0.2134,
      "step": 11389
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5989698275939244,
      "learning_rate": 8.734728717747531e-07,
      "loss": 0.2261,
      "step": 11390
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4320000167371147,
      "learning_rate": 8.728188471046944e-07,
      "loss": 0.2141,
      "step": 11391
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4061933115162406,
      "learning_rate": 8.721650439653223e-07,
      "loss": 0.182,
      "step": 11392
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3411206071685393,
      "learning_rate": 8.715114623917325e-07,
      "loss": 0.1829,
      "step": 11393
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4047648766427407,
      "learning_rate": 8.708581024190049e-07,
      "loss": 0.1807,
      "step": 11394
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3997925347838647,
      "learning_rate": 8.702049640822119e-07,
      "loss": 0.1723,
      "step": 11395
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.329303012777994,
      "learning_rate": 8.695520474164077e-07,
      "loss": 0.1983,
      "step": 11396
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2566335286900716,
      "learning_rate": 8.688993524566403e-07,
      "loss": 0.1417,
      "step": 11397
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3755171497958136,
      "learning_rate": 8.68246879237944e-07,
      "loss": 0.1527,
      "step": 11398
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.902837798432318,
      "learning_rate": 8.675946277953418e-07,
      "loss": 0.4532,
      "step": 11399
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2897325512104756,
      "learning_rate": 8.669425981638413e-07,
      "loss": 0.1325,
      "step": 11400
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3546675765651925,
      "learning_rate": 8.662907903784434e-07,
      "loss": 0.1756,
      "step": 11401
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.437645662695946,
      "learning_rate": 8.65639204474133e-07,
      "loss": 0.5031,
      "step": 11402
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3563981453144178,
      "learning_rate": 8.649878404858858e-07,
      "loss": 0.1851,
      "step": 11403
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3611351621977739,
      "learning_rate": 8.643366984486628e-07,
      "loss": 0.1683,
      "step": 11404
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5114863224056716,
      "learning_rate": 8.636857783974156e-07,
      "loss": 0.1811,
      "step": 11405
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3436726027838803,
      "learning_rate": 8.630350803670828e-07,
      "loss": 0.1947,
      "step": 11406
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.291571610785287,
      "learning_rate": 8.623846043925931e-07,
      "loss": 0.155,
      "step": 11407
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.472627137613963,
      "learning_rate": 8.617343505088593e-07,
      "loss": 0.1946,
      "step": 11408
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.36291889844303,
      "learning_rate": 8.610843187507856e-07,
      "loss": 0.1504,
      "step": 11409
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.355327121725816,
      "learning_rate": 8.604345091532618e-07,
      "loss": 0.1441,
      "step": 11410
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.854175717794461,
      "learning_rate": 8.597849217511689e-07,
      "loss": 0.5641,
      "step": 11411
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4962333643174324,
      "learning_rate": 8.591355565793724e-07,
      "loss": 0.1652,
      "step": 11412
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.331070670423646,
      "learning_rate": 8.584864136727278e-07,
      "loss": 0.1686,
      "step": 11413
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3814311996215893,
      "learning_rate": 8.578374930660793e-07,
      "loss": 0.1807,
      "step": 11414
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2960419642529422,
      "learning_rate": 8.571887947942598e-07,
      "loss": 0.1971,
      "step": 11415
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3386559675980279,
      "learning_rate": 8.565403188920862e-07,
      "loss": 0.1791,
      "step": 11416
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4456846623644037,
      "learning_rate": 8.558920653943681e-07,
      "loss": 0.1765,
      "step": 11417
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1894196749349266,
      "learning_rate": 8.552440343358987e-07,
      "loss": 0.1444,
      "step": 11418
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2940086993018718,
      "learning_rate": 8.545962257514651e-07,
      "loss": 0.1991,
      "step": 11419
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.191183798086964,
      "learning_rate": 8.539486396758357e-07,
      "loss": 0.1662,
      "step": 11420
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.433385408384829,
      "learning_rate": 8.53301276143772e-07,
      "loss": 0.1646,
      "step": 11421
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6347912932052613,
      "learning_rate": 8.52654135190022e-07,
      "loss": 0.1902,
      "step": 11422
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5317605837676436,
      "learning_rate": 8.520072168493215e-07,
      "loss": 0.2058,
      "step": 11423
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3151030158179964,
      "learning_rate": 8.513605211563958e-07,
      "loss": 0.167,
      "step": 11424
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3456674562440323,
      "learning_rate": 8.507140481459541e-07,
      "loss": 0.1543,
      "step": 11425
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2698463288012949,
      "learning_rate": 8.500677978526995e-07,
      "loss": 0.1284,
      "step": 11426
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4545367473438622,
      "learning_rate": 8.494217703113173e-07,
      "loss": 0.1603,
      "step": 11427
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3828906073421843,
      "learning_rate": 8.487759655564864e-07,
      "loss": 0.1912,
      "step": 11428
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3783875459732904,
      "learning_rate": 8.481303836228683e-07,
      "loss": 0.173,
      "step": 11429
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1927576599979075,
      "learning_rate": 8.474850245451172e-07,
      "loss": 0.1461,
      "step": 11430
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4555194343619704,
      "learning_rate": 8.468398883578726e-07,
      "loss": 0.2197,
      "step": 11431
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3264473940926609,
      "learning_rate": 8.461949750957643e-07,
      "loss": 0.1856,
      "step": 11432
    },
    {
      "epoch": 0.82,
      "grad_norm": 83.21352343716754,
      "learning_rate": 8.455502847934066e-07,
      "loss": 0.582,
      "step": 11433
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4432548050672056,
      "learning_rate": 8.449058174854058e-07,
      "loss": 0.1459,
      "step": 11434
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3589038430181848,
      "learning_rate": 8.442615732063525e-07,
      "loss": 0.1932,
      "step": 11435
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3136081518102787,
      "learning_rate": 8.436175519908291e-07,
      "loss": 0.1693,
      "step": 11436
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2946457571192662,
      "learning_rate": 8.429737538734017e-07,
      "loss": 0.1502,
      "step": 11437
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3921904235005222,
      "learning_rate": 8.423301788886285e-07,
      "loss": 0.1847,
      "step": 11438
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.227592518715334,
      "learning_rate": 8.416868270710538e-07,
      "loss": 0.1757,
      "step": 11439
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.417611348627177,
      "learning_rate": 8.410436984552112e-07,
      "loss": 0.1935,
      "step": 11440
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2878340989843002,
      "learning_rate": 8.404007930756186e-07,
      "loss": 0.1759,
      "step": 11441
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.416962663817964,
      "learning_rate": 8.397581109667879e-07,
      "loss": 0.5042,
      "step": 11442
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.219356507567168,
      "learning_rate": 8.391156521632126e-07,
      "loss": 0.4531,
      "step": 11443
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4151476458946368,
      "learning_rate": 8.384734166993802e-07,
      "loss": 0.1722,
      "step": 11444
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3489144496479875,
      "learning_rate": 8.378314046097607e-07,
      "loss": 0.1832,
      "step": 11445
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.261660195567686,
      "learning_rate": 8.371896159288162e-07,
      "loss": 0.1889,
      "step": 11446
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1899621055589107,
      "learning_rate": 8.365480506909945e-07,
      "loss": 0.1372,
      "step": 11447
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5048603569251995,
      "learning_rate": 8.359067089307349e-07,
      "loss": 0.1814,
      "step": 11448
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3390024617764553,
      "learning_rate": 8.35265590682458e-07,
      "loss": 0.1733,
      "step": 11449
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2799014281780703,
      "learning_rate": 8.346246959805804e-07,
      "loss": 0.1541,
      "step": 11450
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.232379939509497,
      "learning_rate": 8.339840248594999e-07,
      "loss": 0.1323,
      "step": 11451
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.540355309579994,
      "learning_rate": 8.33343577353607e-07,
      "loss": 0.6364,
      "step": 11452
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.532081170578114,
      "learning_rate": 8.327033534972767e-07,
      "loss": 0.1493,
      "step": 11453
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.687825726883838,
      "learning_rate": 8.320633533248745e-07,
      "loss": 0.2218,
      "step": 11454
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5075911252993697,
      "learning_rate": 8.314235768707529e-07,
      "loss": 0.2012,
      "step": 11455
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2120210753157823,
      "learning_rate": 8.307840241692533e-07,
      "loss": 0.1615,
      "step": 11456
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2334533982612899,
      "learning_rate": 8.301446952547049e-07,
      "loss": 0.1422,
      "step": 11457
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.150364470744097,
      "learning_rate": 8.295055901614218e-07,
      "loss": 0.1577,
      "step": 11458
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2932311573561976,
      "learning_rate": 8.288667089237118e-07,
      "loss": 0.1462,
      "step": 11459
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.481512813055017,
      "learning_rate": 8.282280515758639e-07,
      "loss": 0.4908,
      "step": 11460
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1685834832873647,
      "learning_rate": 8.275896181521625e-07,
      "loss": 0.129,
      "step": 11461
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2564310152375446,
      "learning_rate": 8.26951408686873e-07,
      "loss": 0.1377,
      "step": 11462
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3100266148914808,
      "learning_rate": 8.263134232142533e-07,
      "loss": 0.1606,
      "step": 11463
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.924961285467801,
      "learning_rate": 8.256756617685474e-07,
      "loss": 0.5518,
      "step": 11464
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3099353995253473,
      "learning_rate": 8.2503812438399e-07,
      "loss": 0.1477,
      "step": 11465
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.404820815623232,
      "learning_rate": 8.244008110947982e-07,
      "loss": 0.1657,
      "step": 11466
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.410008435631488,
      "learning_rate": 8.237637219351835e-07,
      "loss": 0.1636,
      "step": 11467
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3521792707814833,
      "learning_rate": 8.231268569393396e-07,
      "loss": 0.1794,
      "step": 11468
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.274283928033697,
      "learning_rate": 8.224902161414538e-07,
      "loss": 0.1284,
      "step": 11469
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4471274992141288,
      "learning_rate": 8.218537995756953e-07,
      "loss": 0.1982,
      "step": 11470
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2723078407743786,
      "learning_rate": 8.212176072762257e-07,
      "loss": 0.1623,
      "step": 11471
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.232020983422353,
      "learning_rate": 8.205816392771937e-07,
      "loss": 0.1886,
      "step": 11472
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3133249356227799,
      "learning_rate": 8.199458956127366e-07,
      "loss": 0.1552,
      "step": 11473
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3989406792886945,
      "learning_rate": 8.193103763169763e-07,
      "loss": 0.1972,
      "step": 11474
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.488434652451793,
      "learning_rate": 8.186750814240268e-07,
      "loss": 0.1814,
      "step": 11475
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2927563889891591,
      "learning_rate": 8.180400109679864e-07,
      "loss": 0.1434,
      "step": 11476
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.378222540491481,
      "learning_rate": 8.174051649829456e-07,
      "loss": 0.1669,
      "step": 11477
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4021134040163132,
      "learning_rate": 8.167705435029777e-07,
      "loss": 0.1628,
      "step": 11478
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2863336310940636,
      "learning_rate": 8.161361465621481e-07,
      "loss": 0.169,
      "step": 11479
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4294140842320746,
      "learning_rate": 8.15501974194508e-07,
      "loss": 0.1831,
      "step": 11480
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3911381286819142,
      "learning_rate": 8.148680264340997e-07,
      "loss": 0.1802,
      "step": 11481
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2343342185221675,
      "learning_rate": 8.142343033149475e-07,
      "loss": 0.1347,
      "step": 11482
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3439496996293885,
      "learning_rate": 8.136008048710697e-07,
      "loss": 0.1663,
      "step": 11483
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4269009778769877,
      "learning_rate": 8.129675311364682e-07,
      "loss": 0.1954,
      "step": 11484
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3086128953887586,
      "learning_rate": 8.123344821451362e-07,
      "loss": 0.1455,
      "step": 11485
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3170625609030797,
      "learning_rate": 8.117016579310516e-07,
      "loss": 0.1468,
      "step": 11486
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.022690193465774,
      "learning_rate": 8.110690585281827e-07,
      "loss": 0.5532,
      "step": 11487
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4332464655896164,
      "learning_rate": 8.104366839704864e-07,
      "loss": 0.1994,
      "step": 11488
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3252524370680863,
      "learning_rate": 8.09804534291902e-07,
      "loss": 0.1807,
      "step": 11489
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4290117914058154,
      "learning_rate": 8.091726095263658e-07,
      "loss": 0.1639,
      "step": 11490
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3559999589836476,
      "learning_rate": 8.085409097077935e-07,
      "loss": 0.154,
      "step": 11491
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2169854693587618,
      "learning_rate": 8.07909434870095e-07,
      "loss": 0.131,
      "step": 11492
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4983154037756823,
      "learning_rate": 8.072781850471617e-07,
      "loss": 0.166,
      "step": 11493
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.215137305195108,
      "learning_rate": 8.066471602728804e-07,
      "loss": 0.1309,
      "step": 11494
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2600736785724975,
      "learning_rate": 8.060163605811189e-07,
      "loss": 0.1619,
      "step": 11495
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4411261810835667,
      "learning_rate": 8.05385786005739e-07,
      "loss": 0.2178,
      "step": 11496
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4487999820197899,
      "learning_rate": 8.047554365805837e-07,
      "loss": 0.1956,
      "step": 11497
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5063790981367113,
      "learning_rate": 8.041253123394905e-07,
      "loss": 0.24,
      "step": 11498
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.466884247536365,
      "learning_rate": 8.034954133162809e-07,
      "loss": 0.19,
      "step": 11499
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3924120962871684,
      "learning_rate": 8.02865739544767e-07,
      "loss": 0.1866,
      "step": 11500
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3057473474268322,
      "learning_rate": 8.022362910587444e-07,
      "loss": 0.167,
      "step": 11501
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.41886909693053,
      "learning_rate": 8.016070678920023e-07,
      "loss": 0.1563,
      "step": 11502
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2455875720942915,
      "learning_rate": 8.009780700783126e-07,
      "loss": 0.14,
      "step": 11503
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2819670123112297,
      "learning_rate": 8.003492976514387e-07,
      "loss": 0.1556,
      "step": 11504
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2170296388803648,
      "learning_rate": 7.997207506451293e-07,
      "loss": 0.1504,
      "step": 11505
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4207155604639734,
      "learning_rate": 7.990924290931235e-07,
      "loss": 0.1839,
      "step": 11506
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6266247095023547,
      "learning_rate": 7.984643330291464e-07,
      "loss": 0.1973,
      "step": 11507
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.400452828264983,
      "learning_rate": 7.978364624869134e-07,
      "loss": 0.2009,
      "step": 11508
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1847663780948023,
      "learning_rate": 7.972088175001236e-07,
      "loss": 0.1569,
      "step": 11509
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3898018615701242,
      "learning_rate": 7.96581398102469e-07,
      "loss": 0.171,
      "step": 11510
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3423827400338622,
      "learning_rate": 7.959542043276242e-07,
      "loss": 0.1795,
      "step": 11511
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3889643900998607,
      "learning_rate": 7.953272362092573e-07,
      "loss": 0.1763,
      "step": 11512
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.158905902456587,
      "learning_rate": 7.947004937810188e-07,
      "loss": 0.4405,
      "step": 11513
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.298443666662164,
      "learning_rate": 7.94073977076551e-07,
      "loss": 0.1529,
      "step": 11514
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4650860255634925,
      "learning_rate": 7.934476861294831e-07,
      "loss": 0.1795,
      "step": 11515
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2192932766570668,
      "learning_rate": 7.92821620973433e-07,
      "loss": 0.148,
      "step": 11516
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.383336794134288,
      "learning_rate": 7.921957816420023e-07,
      "loss": 0.1807,
      "step": 11517
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.279964189088151,
      "learning_rate": 7.915701681687854e-07,
      "loss": 0.1879,
      "step": 11518
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1165254041883554,
      "learning_rate": 7.909447805873643e-07,
      "loss": 0.1454,
      "step": 11519
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3521343890275785,
      "learning_rate": 7.903196189313039e-07,
      "loss": 0.1488,
      "step": 11520
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.229123036957462,
      "learning_rate": 7.89694683234164e-07,
      "loss": 0.1589,
      "step": 11521
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4916578370647167,
      "learning_rate": 7.890699735294849e-07,
      "loss": 0.2255,
      "step": 11522
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3684949267094058,
      "learning_rate": 7.88445489850801e-07,
      "loss": 0.1718,
      "step": 11523
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5259462181532597,
      "learning_rate": 7.878212322316315e-07,
      "loss": 0.2103,
      "step": 11524
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2530013089782075,
      "learning_rate": 7.871972007054857e-07,
      "loss": 0.1651,
      "step": 11525
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2965918925973268,
      "learning_rate": 7.865733953058557e-07,
      "loss": 0.1707,
      "step": 11526
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3543305548312021,
      "learning_rate": 7.859498160662288e-07,
      "loss": 0.1622,
      "step": 11527
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.461249997521325,
      "learning_rate": 7.853264630200724e-07,
      "loss": 0.7479,
      "step": 11528
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2334353141045586,
      "learning_rate": 7.847033362008488e-07,
      "loss": 0.1688,
      "step": 11529
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5638901386218238,
      "learning_rate": 7.840804356420023e-07,
      "loss": 0.1906,
      "step": 11530
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3085516759229985,
      "learning_rate": 7.834577613769695e-07,
      "loss": 0.1449,
      "step": 11531
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3983067791455244,
      "learning_rate": 7.82835313439172e-07,
      "loss": 0.1496,
      "step": 11532
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3649651066379158,
      "learning_rate": 7.822130918620229e-07,
      "loss": 0.1668,
      "step": 11533
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2842317132142549,
      "learning_rate": 7.81591096678917e-07,
      "loss": 0.1615,
      "step": 11534
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4081704303421252,
      "learning_rate": 7.809693279232433e-07,
      "loss": 0.1598,
      "step": 11535
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3008653200642926,
      "learning_rate": 7.803477856283737e-07,
      "loss": 0.1661,
      "step": 11536
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6934171084701195,
      "learning_rate": 7.797264698276724e-07,
      "loss": 0.1928,
      "step": 11537
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2801505967998739,
      "learning_rate": 7.791053805544868e-07,
      "loss": 0.1525,
      "step": 11538
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4122828018364577,
      "learning_rate": 7.784845178421557e-07,
      "loss": 0.1849,
      "step": 11539
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.3762601828516114,
      "learning_rate": 7.778638817240042e-07,
      "loss": 0.4396,
      "step": 11540
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.293272809696471,
      "learning_rate": 7.772434722333472e-07,
      "loss": 0.4673,
      "step": 11541
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3378789806836056,
      "learning_rate": 7.766232894034836e-07,
      "loss": 0.1927,
      "step": 11542
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2215449572499835,
      "learning_rate": 7.760033332677042e-07,
      "loss": 0.193,
      "step": 11543
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.278849011623813,
      "learning_rate": 7.753836038592832e-07,
      "loss": 0.5433,
      "step": 11544
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2552225211964956,
      "learning_rate": 7.747641012114888e-07,
      "loss": 0.167,
      "step": 11545
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4000809638611953,
      "learning_rate": 7.741448253575696e-07,
      "loss": 0.1888,
      "step": 11546
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3722678543938278,
      "learning_rate": 7.73525776330768e-07,
      "loss": 0.1433,
      "step": 11547
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2435592500271755,
      "learning_rate": 7.72906954164312e-07,
      "loss": 0.1449,
      "step": 11548
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4152898672306937,
      "learning_rate": 7.722883588914182e-07,
      "loss": 0.1773,
      "step": 11549
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.381987639110904,
      "learning_rate": 7.71669990545289e-07,
      "loss": 0.1694,
      "step": 11550
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.42479961541026,
      "learning_rate": 7.710518491591157e-07,
      "loss": 0.1659,
      "step": 11551
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4462711227043956,
      "learning_rate": 7.7043393476608e-07,
      "loss": 0.1914,
      "step": 11552
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.34193221328926,
      "learning_rate": 7.69816247399346e-07,
      "loss": 0.1551,
      "step": 11553
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2681811340099547,
      "learning_rate": 7.691987870920714e-07,
      "loss": 0.1432,
      "step": 11554
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3866136376429892,
      "learning_rate": 7.685815538773972e-07,
      "loss": 0.1789,
      "step": 11555
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3636563466687117,
      "learning_rate": 7.67964547788454e-07,
      "loss": 0.1692,
      "step": 11556
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1881460364550938,
      "learning_rate": 7.673477688583614e-07,
      "loss": 0.1423,
      "step": 11557
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.326759707480179,
      "learning_rate": 7.667312171202262e-07,
      "loss": 0.1655,
      "step": 11558
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.253416671078902,
      "learning_rate": 7.661148926071399e-07,
      "loss": 0.1467,
      "step": 11559
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1965902275491853,
      "learning_rate": 7.654987953521875e-07,
      "loss": 0.1366,
      "step": 11560
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3389678297848768,
      "learning_rate": 7.648829253884354e-07,
      "loss": 0.201,
      "step": 11561
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3517352503918176,
      "learning_rate": 7.642672827489439e-07,
      "loss": 0.1834,
      "step": 11562
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.611422802926312,
      "learning_rate": 7.636518674667559e-07,
      "loss": 0.174,
      "step": 11563
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4253466716311418,
      "learning_rate": 7.63036679574905e-07,
      "loss": 0.1715,
      "step": 11564
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2376411307084338,
      "learning_rate": 7.624217191064132e-07,
      "loss": 0.1559,
      "step": 11565
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3446571342551363,
      "learning_rate": 7.618069860942895e-07,
      "loss": 0.1549,
      "step": 11566
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5191627813968365,
      "learning_rate": 7.611924805715281e-07,
      "loss": 0.1809,
      "step": 11567
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.437873724648136,
      "learning_rate": 7.605782025711162e-07,
      "loss": 0.1921,
      "step": 11568
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3331462797618217,
      "learning_rate": 7.59964152126022e-07,
      "loss": 0.1624,
      "step": 11569
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4792015457858434,
      "learning_rate": 7.593503292692089e-07,
      "loss": 0.1671,
      "step": 11570
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2777413872750198,
      "learning_rate": 7.587367340336221e-07,
      "loss": 0.1429,
      "step": 11571
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0861017482570525,
      "learning_rate": 7.581233664521975e-07,
      "loss": 0.1182,
      "step": 11572
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2408880862072895,
      "learning_rate": 7.575102265578582e-07,
      "loss": 0.1383,
      "step": 11573
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4634111106649268,
      "learning_rate": 7.568973143835167e-07,
      "loss": 0.1628,
      "step": 11574
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2217449727468324,
      "learning_rate": 7.562846299620697e-07,
      "loss": 0.1654,
      "step": 11575
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.249319849310265,
      "learning_rate": 7.55672173326405e-07,
      "loss": 0.1638,
      "step": 11576
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.291951851251199,
      "learning_rate": 7.550599445093954e-07,
      "loss": 0.1581,
      "step": 11577
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.257928676082074,
      "learning_rate": 7.54447943543905e-07,
      "loss": 0.1491,
      "step": 11578
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5349822952156822,
      "learning_rate": 7.538361704627806e-07,
      "loss": 0.1936,
      "step": 11579
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4003094983468192,
      "learning_rate": 7.532246252988617e-07,
      "loss": 0.1936,
      "step": 11580
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.284220739888052,
      "learning_rate": 7.526133080849735e-07,
      "loss": 0.1298,
      "step": 11581
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2832934872079642,
      "learning_rate": 7.520022188539289e-07,
      "loss": 0.1623,
      "step": 11582
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.512801460576542,
      "learning_rate": 7.513913576385301e-07,
      "loss": 0.525,
      "step": 11583
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2630420408284937,
      "learning_rate": 7.507807244715632e-07,
      "loss": 0.1568,
      "step": 11584
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3029361632584875,
      "learning_rate": 7.501703193858067e-07,
      "loss": 0.1817,
      "step": 11585
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.564370039556058,
      "learning_rate": 7.495601424140225e-07,
      "loss": 0.2096,
      "step": 11586
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3706090738354475,
      "learning_rate": 7.489501935889654e-07,
      "loss": 0.1609,
      "step": 11587
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1932657385144498,
      "learning_rate": 7.483404729433719e-07,
      "loss": 0.1306,
      "step": 11588
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.945988227655147,
      "learning_rate": 7.47730980509972e-07,
      "loss": 0.6897,
      "step": 11589
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4562936630667112,
      "learning_rate": 7.471217163214772e-07,
      "loss": 0.1649,
      "step": 11590
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2408819525654768,
      "learning_rate": 7.46512680410596e-07,
      "loss": 0.1516,
      "step": 11591
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.679494843254499,
      "learning_rate": 7.459038728100138e-07,
      "loss": 0.6117,
      "step": 11592
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5048268288352258,
      "learning_rate": 7.452952935524127e-07,
      "loss": 0.1623,
      "step": 11593
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3704584312833803,
      "learning_rate": 7.446869426704561e-07,
      "loss": 0.1882,
      "step": 11594
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.087557321193094,
      "learning_rate": 7.440788201968002e-07,
      "loss": 0.139,
      "step": 11595
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.302819372526447,
      "learning_rate": 7.434709261640838e-07,
      "loss": 0.1648,
      "step": 11596
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.199608237312297,
      "learning_rate": 7.428632606049391e-07,
      "loss": 0.1387,
      "step": 11597
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.330136663119566,
      "learning_rate": 7.422558235519806e-07,
      "loss": 0.17,
      "step": 11598
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3711883535313825,
      "learning_rate": 7.416486150378144e-07,
      "loss": 0.1619,
      "step": 11599
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4004435705121814,
      "learning_rate": 7.410416350950333e-07,
      "loss": 0.1789,
      "step": 11600
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2076941643903294,
      "learning_rate": 7.40434883756218e-07,
      "loss": 0.1443,
      "step": 11601
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2884450094197355,
      "learning_rate": 7.398283610539353e-07,
      "loss": 0.1622,
      "step": 11602
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4814363222316966,
      "learning_rate": 7.392220670207423e-07,
      "loss": 0.2145,
      "step": 11603
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2472243685304822,
      "learning_rate": 7.386160016891802e-07,
      "loss": 0.1816,
      "step": 11604
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3212437636954517,
      "learning_rate": 7.380101650917826e-07,
      "loss": 0.1584,
      "step": 11605
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5248995669252374,
      "learning_rate": 7.374045572610666e-07,
      "loss": 0.1774,
      "step": 11606
    },
    {
      "epoch": 0.83,
      "grad_norm": 9.708464933852696,
      "learning_rate": 7.367991782295392e-07,
      "loss": 0.6073,
      "step": 11607
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5007342902331788,
      "learning_rate": 7.361940280296953e-07,
      "loss": 0.176,
      "step": 11608
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3013840924045739,
      "learning_rate": 7.355891066940179e-07,
      "loss": 0.1711,
      "step": 11609
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.196484274712304,
      "learning_rate": 7.349844142549744e-07,
      "loss": 0.1679,
      "step": 11610
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4110754991383962,
      "learning_rate": 7.34379950745025e-07,
      "loss": 0.1842,
      "step": 11611
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3294933767706165,
      "learning_rate": 7.337757161966119e-07,
      "loss": 0.1533,
      "step": 11612
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3295565377974825,
      "learning_rate": 7.331717106421709e-07,
      "loss": 0.1866,
      "step": 11613
    },
    {
      "epoch": 0.83,
      "grad_norm": 30.98738037654596,
      "learning_rate": 7.3256793411412e-07,
      "loss": 0.5968,
      "step": 11614
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.393754771874616,
      "learning_rate": 7.319643866448689e-07,
      "loss": 0.2062,
      "step": 11615
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.455280162807254,
      "learning_rate": 7.313610682668132e-07,
      "loss": 0.1984,
      "step": 11616
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3775703371519914,
      "learning_rate": 7.307579790123376e-07,
      "loss": 0.1582,
      "step": 11617
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3831351045595033,
      "learning_rate": 7.301551189138134e-07,
      "loss": 0.1708,
      "step": 11618
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.339201170953808,
      "learning_rate": 7.295524880035987e-07,
      "loss": 0.2,
      "step": 11619
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3418064995382029,
      "learning_rate": 7.289500863140414e-07,
      "loss": 0.1836,
      "step": 11620
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3326598824030378,
      "learning_rate": 7.283479138774746e-07,
      "loss": 0.1667,
      "step": 11621
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3655307260463638,
      "learning_rate": 7.277459707262225e-07,
      "loss": 0.1581,
      "step": 11622
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4331386375969262,
      "learning_rate": 7.27144256892593e-07,
      "loss": 0.2059,
      "step": 11623
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2804462693553644,
      "learning_rate": 7.265427724088842e-07,
      "loss": 0.2018,
      "step": 11624
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2914246183466542,
      "learning_rate": 7.25941517307382e-07,
      "loss": 0.1333,
      "step": 11625
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5763593066306556,
      "learning_rate": 7.253404916203605e-07,
      "loss": 0.1827,
      "step": 11626
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.588771475788876,
      "learning_rate": 7.247396953800778e-07,
      "loss": 0.5215,
      "step": 11627
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6097079485031904,
      "learning_rate": 7.241391286187849e-07,
      "loss": 0.2276,
      "step": 11628
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3581610147676522,
      "learning_rate": 7.235387913687153e-07,
      "loss": 0.1778,
      "step": 11629
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.439366639484726,
      "learning_rate": 7.22938683662095e-07,
      "loss": 0.1459,
      "step": 11630
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.421754298246453,
      "learning_rate": 7.223388055311331e-07,
      "loss": 0.154,
      "step": 11631
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4037248082670815,
      "learning_rate": 7.217391570080302e-07,
      "loss": 0.1811,
      "step": 11632
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.458678308573073,
      "learning_rate": 7.211397381249724e-07,
      "loss": 0.4446,
      "step": 11633
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4752498990753813,
      "learning_rate": 7.205405489141354e-07,
      "loss": 0.1646,
      "step": 11634
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.388908443341257,
      "learning_rate": 7.199415894076795e-07,
      "loss": 0.1582,
      "step": 11635
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2895418286990967,
      "learning_rate": 7.193428596377561e-07,
      "loss": 0.1305,
      "step": 11636
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.01856345789147,
      "learning_rate": 7.187443596365013e-07,
      "loss": 0.4766,
      "step": 11637
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3089173390219897,
      "learning_rate": 7.181460894360415e-07,
      "loss": 0.1635,
      "step": 11638
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3765236494637514,
      "learning_rate": 7.175480490684872e-07,
      "loss": 0.1729,
      "step": 11639
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7509183075420982,
      "learning_rate": 7.16950238565941e-07,
      "loss": 0.2087,
      "step": 11640
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1854208294479573,
      "learning_rate": 7.163526579604902e-07,
      "loss": 0.1328,
      "step": 11641
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.450009021457237,
      "learning_rate": 7.157553072842116e-07,
      "loss": 0.1619,
      "step": 11642
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3579177639365734,
      "learning_rate": 7.151581865691671e-07,
      "loss": 0.1712,
      "step": 11643
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3798007825921677,
      "learning_rate": 7.14561295847409e-07,
      "loss": 0.1619,
      "step": 11644
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3706440834218203,
      "learning_rate": 7.139646351509744e-07,
      "loss": 0.2247,
      "step": 11645
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.489584081379273,
      "learning_rate": 7.133682045118906e-07,
      "loss": 0.2412,
      "step": 11646
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.303560384052622,
      "learning_rate": 7.127720039621727e-07,
      "loss": 0.1595,
      "step": 11647
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2322530008983776,
      "learning_rate": 7.121760335338201e-07,
      "loss": 0.1594,
      "step": 11648
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3701177571012602,
      "learning_rate": 7.115802932588239e-07,
      "loss": 0.183,
      "step": 11649
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3448546900148413,
      "learning_rate": 7.109847831691608e-07,
      "loss": 0.1647,
      "step": 11650
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.249504772027403,
      "learning_rate": 7.103895032967955e-07,
      "loss": 0.1523,
      "step": 11651
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.719241041882766,
      "learning_rate": 7.097944536736795e-07,
      "loss": 0.5115,
      "step": 11652
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.105762148973235,
      "learning_rate": 7.091996343317537e-07,
      "loss": 0.7344,
      "step": 11653
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3814628572246865,
      "learning_rate": 7.086050453029442e-07,
      "loss": 0.1923,
      "step": 11654
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.371654667850567,
      "learning_rate": 7.080106866191683e-07,
      "loss": 0.1568,
      "step": 11655
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4375467628545004,
      "learning_rate": 7.074165583123266e-07,
      "loss": 0.1862,
      "step": 11656
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.911442313692837,
      "learning_rate": 7.068226604143103e-07,
      "loss": 0.6756,
      "step": 11657
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.381787643863725,
      "learning_rate": 7.062289929569977e-07,
      "loss": 0.1875,
      "step": 11658
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3426209192917509,
      "learning_rate": 7.056355559722555e-07,
      "loss": 0.1663,
      "step": 11659
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4536859530141466,
      "learning_rate": 7.05042349491935e-07,
      "loss": 0.1737,
      "step": 11660
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.6978744719980305,
      "learning_rate": 7.044493735478797e-07,
      "loss": 0.472,
      "step": 11661
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2818340837972102,
      "learning_rate": 7.03856628171915e-07,
      "loss": 0.1673,
      "step": 11662
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4849495835658328,
      "learning_rate": 7.032641133958607e-07,
      "loss": 0.169,
      "step": 11663
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2451367941156322,
      "learning_rate": 7.026718292515166e-07,
      "loss": 0.1336,
      "step": 11664
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3027510813242822,
      "learning_rate": 7.020797757706771e-07,
      "loss": 0.1567,
      "step": 11665
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4526465700969229,
      "learning_rate": 7.014879529851204e-07,
      "loss": 0.1703,
      "step": 11666
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3267610438653357,
      "learning_rate": 7.008963609266145e-07,
      "loss": 0.1804,
      "step": 11667
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3754745718316033,
      "learning_rate": 7.003049996269112e-07,
      "loss": 0.1514,
      "step": 11668
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.606624661694305,
      "learning_rate": 6.997138691177546e-07,
      "loss": 0.6555,
      "step": 11669
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5229793942128156,
      "learning_rate": 6.991229694308726e-07,
      "loss": 0.1924,
      "step": 11670
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3983102002880965,
      "learning_rate": 6.985323005979844e-07,
      "loss": 0.1975,
      "step": 11671
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0278462527220869,
      "learning_rate": 6.979418626507922e-07,
      "loss": 0.1003,
      "step": 11672
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.273742050539887,
      "learning_rate": 6.973516556209898e-07,
      "loss": 0.1642,
      "step": 11673
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4212775476674828,
      "learning_rate": 6.967616795402565e-07,
      "loss": 0.1677,
      "step": 11674
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5959808353107412,
      "learning_rate": 6.961719344402617e-07,
      "loss": 0.1854,
      "step": 11675
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3091535942384105,
      "learning_rate": 6.955824203526585e-07,
      "loss": 0.1653,
      "step": 11676
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3134927942392767,
      "learning_rate": 6.949931373090912e-07,
      "loss": 0.1409,
      "step": 11677
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3652037895055094,
      "learning_rate": 6.944040853411882e-07,
      "loss": 0.1762,
      "step": 11678
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2766694701772208,
      "learning_rate": 6.93815264480569e-07,
      "loss": 0.1542,
      "step": 11679
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.335967256640693,
      "learning_rate": 6.932266747588395e-07,
      "loss": 0.1578,
      "step": 11680
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.481941753227402,
      "learning_rate": 6.926383162075917e-07,
      "loss": 0.4685,
      "step": 11681
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3858531760525818,
      "learning_rate": 6.920501888584063e-07,
      "loss": 0.1516,
      "step": 11682
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2770421713865137,
      "learning_rate": 6.914622927428527e-07,
      "loss": 0.1561,
      "step": 11683
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.329417357024852,
      "learning_rate": 6.908746278924872e-07,
      "loss": 0.6285,
      "step": 11684
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2641289409351533,
      "learning_rate": 6.902871943388512e-07,
      "loss": 0.1421,
      "step": 11685
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3261687506803588,
      "learning_rate": 6.896999921134784e-07,
      "loss": 0.1836,
      "step": 11686
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5977296438011017,
      "learning_rate": 6.891130212478847e-07,
      "loss": 0.2024,
      "step": 11687
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3346254901050132,
      "learning_rate": 6.885262817735789e-07,
      "loss": 0.1428,
      "step": 11688
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3739951000940909,
      "learning_rate": 6.879397737220528e-07,
      "loss": 0.1453,
      "step": 11689
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.347399895127108,
      "learning_rate": 6.873534971247897e-07,
      "loss": 0.1751,
      "step": 11690
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2898639620122447,
      "learning_rate": 6.867674520132567e-07,
      "loss": 0.1539,
      "step": 11691
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3188204756353004,
      "learning_rate": 6.861816384189107e-07,
      "loss": 0.1597,
      "step": 11692
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5334764211286867,
      "learning_rate": 6.855960563731972e-07,
      "loss": 0.2061,
      "step": 11693
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3828502166818224,
      "learning_rate": 6.850107059075478e-07,
      "loss": 0.189,
      "step": 11694
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4380249797393059,
      "learning_rate": 6.844255870533806e-07,
      "loss": 0.1833,
      "step": 11695
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3844340246367486,
      "learning_rate": 6.838406998421038e-07,
      "loss": 0.1819,
      "step": 11696
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6103480951829856,
      "learning_rate": 6.8325604430511e-07,
      "loss": 0.2018,
      "step": 11697
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3346237743896638,
      "learning_rate": 6.826716204737832e-07,
      "loss": 0.158,
      "step": 11698
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3466762863298858,
      "learning_rate": 6.820874283794909e-07,
      "loss": 0.1735,
      "step": 11699
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.558454454418486,
      "learning_rate": 6.815034680535915e-07,
      "loss": 0.442,
      "step": 11700
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2728376002677624,
      "learning_rate": 6.809197395274291e-07,
      "loss": 0.1763,
      "step": 11701
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.036437312375747,
      "learning_rate": 6.803362428323379e-07,
      "loss": 0.5571,
      "step": 11702
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2178136934159807,
      "learning_rate": 6.797529779996342e-07,
      "loss": 0.1582,
      "step": 11703
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2617567877666496,
      "learning_rate": 6.791699450606293e-07,
      "loss": 0.1724,
      "step": 11704
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.26442055427084,
      "learning_rate": 6.78587144046614e-07,
      "loss": 0.1921,
      "step": 11705
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.371830499006445,
      "learning_rate": 6.780045749888742e-07,
      "loss": 0.159,
      "step": 11706
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3599719811081583,
      "learning_rate": 6.774222379186774e-07,
      "loss": 0.1722,
      "step": 11707
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.349665489670711,
      "learning_rate": 6.768401328672825e-07,
      "loss": 0.5265,
      "step": 11708
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4471261458332878,
      "learning_rate": 6.762582598659334e-07,
      "loss": 0.1626,
      "step": 11709
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4713515872406844,
      "learning_rate": 6.756766189458642e-07,
      "loss": 0.1582,
      "step": 11710
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4358802364193017,
      "learning_rate": 6.750952101382957e-07,
      "loss": 0.2113,
      "step": 11711
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2107552455140782,
      "learning_rate": 6.745140334744327e-07,
      "loss": 0.1566,
      "step": 11712
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2892818200222695,
      "learning_rate": 6.739330889854739e-07,
      "loss": 0.1567,
      "step": 11713
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4724895935737239,
      "learning_rate": 6.73352376702599e-07,
      "loss": 0.166,
      "step": 11714
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.337973964978723,
      "learning_rate": 6.727718966569807e-07,
      "loss": 0.6202,
      "step": 11715
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1445821298792591,
      "learning_rate": 6.721916488797748e-07,
      "loss": 0.1605,
      "step": 11716
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2882209120008221,
      "learning_rate": 6.716116334021272e-07,
      "loss": 0.2137,
      "step": 11717
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.321049666032275,
      "learning_rate": 6.710318502551721e-07,
      "loss": 0.5727,
      "step": 11718
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.329302298124122,
      "learning_rate": 6.704522994700296e-07,
      "loss": 0.14,
      "step": 11719
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3388685316716111,
      "learning_rate": 6.698729810778065e-07,
      "loss": 0.1589,
      "step": 11720
    },
    {
      "epoch": 0.84,
      "grad_norm": 9.88308495341784,
      "learning_rate": 6.692938951096006e-07,
      "loss": 0.7317,
      "step": 11721
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4206784227754827,
      "learning_rate": 6.687150415964916e-07,
      "loss": 0.2065,
      "step": 11722
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2779763353784492,
      "learning_rate": 6.681364205695534e-07,
      "loss": 0.1706,
      "step": 11723
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3476026400011756,
      "learning_rate": 6.675580320598418e-07,
      "loss": 0.1743,
      "step": 11724
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4393112906738605,
      "learning_rate": 6.669798760984026e-07,
      "loss": 0.1803,
      "step": 11725
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4368406290964721,
      "learning_rate": 6.664019527162697e-07,
      "loss": 0.1482,
      "step": 11726
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1262472278577211,
      "learning_rate": 6.658242619444644e-07,
      "loss": 0.158,
      "step": 11727
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1935727083902938,
      "learning_rate": 6.652468038139931e-07,
      "loss": 0.1325,
      "step": 11728
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.464242741678958,
      "learning_rate": 6.646695783558532e-07,
      "loss": 0.1756,
      "step": 11729
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3428680940062958,
      "learning_rate": 6.640925856010261e-07,
      "loss": 0.1557,
      "step": 11730
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.137829852659806,
      "learning_rate": 6.635158255804841e-07,
      "loss": 0.5337,
      "step": 11731
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1385992638106823,
      "learning_rate": 6.629392983251842e-07,
      "loss": 0.1209,
      "step": 11732
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5452992999716484,
      "learning_rate": 6.623630038660722e-07,
      "loss": 0.18,
      "step": 11733
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7144292028908832,
      "learning_rate": 6.617869422340812e-07,
      "loss": 0.1693,
      "step": 11734
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.47553365268462,
      "learning_rate": 6.61211113460134e-07,
      "loss": 0.1634,
      "step": 11735
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2106834650915264,
      "learning_rate": 6.606355175751361e-07,
      "loss": 0.1737,
      "step": 11736
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4289696172642374,
      "learning_rate": 6.60060154609985e-07,
      "loss": 0.148,
      "step": 11737
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1332711007831808,
      "learning_rate": 6.59485024595562e-07,
      "loss": 0.1117,
      "step": 11738
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5044072373214703,
      "learning_rate": 6.589101275627402e-07,
      "loss": 0.1818,
      "step": 11739
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3228327735703826,
      "learning_rate": 6.583354635423755e-07,
      "loss": 0.1487,
      "step": 11740
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3225351863819077,
      "learning_rate": 6.577610325653144e-07,
      "loss": 0.1598,
      "step": 11741
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3295954408994202,
      "learning_rate": 6.5718683466239e-07,
      "loss": 0.1768,
      "step": 11742
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3735337339069924,
      "learning_rate": 6.566128698644237e-07,
      "loss": 0.1521,
      "step": 11743
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.778216073936575,
      "learning_rate": 6.560391382022241e-07,
      "loss": 0.6154,
      "step": 11744
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4864768870777096,
      "learning_rate": 6.55465639706585e-07,
      "loss": 0.1512,
      "step": 11745
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4979391761733065,
      "learning_rate": 6.548923744082913e-07,
      "loss": 0.1879,
      "step": 11746
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4561052640747052,
      "learning_rate": 6.543193423381117e-07,
      "loss": 0.1855,
      "step": 11747
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3658502460776525,
      "learning_rate": 6.537465435268065e-07,
      "loss": 0.1541,
      "step": 11748
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0913594619378433,
      "learning_rate": 6.531739780051194e-07,
      "loss": 0.1452,
      "step": 11749
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5195907251245484,
      "learning_rate": 6.526016458037837e-07,
      "loss": 0.1784,
      "step": 11750
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.146031419662332,
      "learning_rate": 6.520295469535204e-07,
      "loss": 0.6037,
      "step": 11751
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.206777256735477,
      "learning_rate": 6.514576814850382e-07,
      "loss": 0.4465,
      "step": 11752
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4179645020141463,
      "learning_rate": 6.508860494290309e-07,
      "loss": 0.207,
      "step": 11753
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.882637412159327,
      "learning_rate": 6.503146508161839e-07,
      "loss": 0.6348,
      "step": 11754
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2418436042677932,
      "learning_rate": 6.497434856771645e-07,
      "loss": 0.1226,
      "step": 11755
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4257577774940373,
      "learning_rate": 6.491725540426335e-07,
      "loss": 0.1765,
      "step": 11756
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4515318733357008,
      "learning_rate": 6.486018559432338e-07,
      "loss": 0.2055,
      "step": 11757
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3886170265344622,
      "learning_rate": 6.480313914095992e-07,
      "loss": 0.1504,
      "step": 11758
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3806203570120676,
      "learning_rate": 6.474611604723496e-07,
      "loss": 0.162,
      "step": 11759
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3017799672656005,
      "learning_rate": 6.46891163162095e-07,
      "loss": 0.1545,
      "step": 11760
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4955182263148958,
      "learning_rate": 6.463213995094275e-07,
      "loss": 0.1737,
      "step": 11761
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3447545243109928,
      "learning_rate": 6.457518695449316e-07,
      "loss": 0.1845,
      "step": 11762
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2131151573920949,
      "learning_rate": 6.451825732991762e-07,
      "loss": 0.1249,
      "step": 11763
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.279126757749715,
      "learning_rate": 6.446135108027208e-07,
      "loss": 0.1442,
      "step": 11764
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.238124636974609,
      "learning_rate": 6.440446820861073e-07,
      "loss": 0.1293,
      "step": 11765
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2678413577697123,
      "learning_rate": 6.434760871798707e-07,
      "loss": 0.1635,
      "step": 11766
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3596953656098216,
      "learning_rate": 6.4290772611453e-07,
      "loss": 0.1562,
      "step": 11767
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3308912794158525,
      "learning_rate": 6.423395989205938e-07,
      "loss": 0.1536,
      "step": 11768
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4022258646513146,
      "learning_rate": 6.417717056285544e-07,
      "loss": 0.1771,
      "step": 11769
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3232166314615519,
      "learning_rate": 6.412040462688973e-07,
      "loss": 0.1634,
      "step": 11770
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.411760500873118,
      "learning_rate": 6.406366208720887e-07,
      "loss": 0.1476,
      "step": 11771
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.257320733796225,
      "learning_rate": 6.400694294685889e-07,
      "loss": 0.1864,
      "step": 11772
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4213901649618883,
      "learning_rate": 6.395024720888398e-07,
      "loss": 0.1709,
      "step": 11773
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.474540886126251,
      "learning_rate": 6.389357487632747e-07,
      "loss": 0.551,
      "step": 11774
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3676068144540694,
      "learning_rate": 6.383692595223129e-07,
      "loss": 0.1468,
      "step": 11775
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3631988095529393,
      "learning_rate": 6.378030043963618e-07,
      "loss": 0.1951,
      "step": 11776
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3231786675784036,
      "learning_rate": 6.37236983415816e-07,
      "loss": 0.1551,
      "step": 11777
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.30640779582237,
      "learning_rate": 6.366711966110556e-07,
      "loss": 0.1621,
      "step": 11778
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3270666637225543,
      "learning_rate": 6.361056440124519e-07,
      "loss": 0.1676,
      "step": 11779
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3213712133721982,
      "learning_rate": 6.355403256503595e-07,
      "loss": 0.1721,
      "step": 11780
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.974781036389823,
      "learning_rate": 6.349752415551241e-07,
      "loss": 0.47,
      "step": 11781
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.979692414166524,
      "learning_rate": 6.34410391757076e-07,
      "loss": 0.5925,
      "step": 11782
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.39539996003602,
      "learning_rate": 6.338457762865336e-07,
      "loss": 0.1687,
      "step": 11783
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.281130222513427,
      "learning_rate": 6.332813951738048e-07,
      "loss": 0.1439,
      "step": 11784
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3886407096596323,
      "learning_rate": 6.327172484491834e-07,
      "loss": 0.1656,
      "step": 11785
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.347948408407155,
      "learning_rate": 6.321533361429489e-07,
      "loss": 0.1344,
      "step": 11786
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2493491195629831,
      "learning_rate": 6.315896582853714e-07,
      "loss": 0.131,
      "step": 11787
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3098562943588363,
      "learning_rate": 6.310262149067059e-07,
      "loss": 0.1596,
      "step": 11788
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1956999952490035,
      "learning_rate": 6.304630060371969e-07,
      "loss": 0.1739,
      "step": 11789
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4462340067272852,
      "learning_rate": 6.29900031707073e-07,
      "loss": 0.192,
      "step": 11790
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.231425066252052,
      "learning_rate": 6.293372919465557e-07,
      "loss": 0.1492,
      "step": 11791
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.645304568722303,
      "learning_rate": 6.287747867858473e-07,
      "loss": 0.1794,
      "step": 11792
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3705834856034897,
      "learning_rate": 6.282125162551428e-07,
      "loss": 0.1552,
      "step": 11793
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.294808462813357,
      "learning_rate": 6.276504803846218e-07,
      "loss": 0.1947,
      "step": 11794
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2945688230608996,
      "learning_rate": 6.27088679204454e-07,
      "loss": 0.1524,
      "step": 11795
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4997974318618315,
      "learning_rate": 6.26527112744792e-07,
      "loss": 0.183,
      "step": 11796
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2443360259166645,
      "learning_rate": 6.259657810357806e-07,
      "loss": 0.1475,
      "step": 11797
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3192035620163904,
      "learning_rate": 6.254046841075484e-07,
      "loss": 0.1762,
      "step": 11798
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.178712571124446,
      "learning_rate": 6.248438219902142e-07,
      "loss": 0.4251,
      "step": 11799
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.8792583565370435,
      "learning_rate": 6.242831947138806e-07,
      "loss": 0.559,
      "step": 11800
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4683693918250638,
      "learning_rate": 6.237228023086417e-07,
      "loss": 0.1632,
      "step": 11801
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1669918420301704,
      "learning_rate": 6.231626448045768e-07,
      "loss": 0.163,
      "step": 11802
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4678020617138623,
      "learning_rate": 6.226027222317537e-07,
      "loss": 0.1769,
      "step": 11803
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2121870294760644,
      "learning_rate": 6.22043034620225e-07,
      "loss": 0.1398,
      "step": 11804
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.606257562154117,
      "learning_rate": 6.214835820000348e-07,
      "loss": 0.6958,
      "step": 11805
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3514621994016,
      "learning_rate": 6.209243644012097e-07,
      "loss": 0.1626,
      "step": 11806
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5014267658838967,
      "learning_rate": 6.20365381853768e-07,
      "loss": 0.1681,
      "step": 11807
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3409474272230335,
      "learning_rate": 6.198066343877135e-07,
      "loss": 0.1615,
      "step": 11808
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.376892335599446,
      "learning_rate": 6.19248122033037e-07,
      "loss": 0.1503,
      "step": 11809
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.418224130834307,
      "learning_rate": 6.186898448197171e-07,
      "loss": 0.1827,
      "step": 11810
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.136715036733706,
      "learning_rate": 6.181318027777206e-07,
      "loss": 0.1487,
      "step": 11811
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3966931557076556,
      "learning_rate": 6.175739959370014e-07,
      "loss": 0.1885,
      "step": 11812
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2910033251439874,
      "learning_rate": 6.170164243274985e-07,
      "loss": 0.1969,
      "step": 11813
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4004184911286062,
      "learning_rate": 6.164590879791426e-07,
      "loss": 0.175,
      "step": 11814
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1986470757325758,
      "learning_rate": 6.159019869218469e-07,
      "loss": 0.1524,
      "step": 11815
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3452678226203771,
      "learning_rate": 6.153451211855166e-07,
      "loss": 0.1474,
      "step": 11816
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3621701275636258,
      "learning_rate": 6.147884908000395e-07,
      "loss": 0.1852,
      "step": 11817
    },
    {
      "epoch": 0.85,
      "grad_norm": 12.68012984037672,
      "learning_rate": 6.142320957952952e-07,
      "loss": 0.6479,
      "step": 11818
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.176518127917164,
      "learning_rate": 6.13675936201148e-07,
      "loss": 0.6383,
      "step": 11819
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4058398596729822,
      "learning_rate": 6.131200120474512e-07,
      "loss": 0.2144,
      "step": 11820
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4397957447801177,
      "learning_rate": 6.125643233640433e-07,
      "loss": 0.1505,
      "step": 11821
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3713555901524892,
      "learning_rate": 6.120088701807536e-07,
      "loss": 0.2013,
      "step": 11822
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2689658787351874,
      "learning_rate": 6.114536525273934e-07,
      "loss": 0.1697,
      "step": 11823
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.84563822572434,
      "learning_rate": 6.108986704337683e-07,
      "loss": 0.6613,
      "step": 11824
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2877933397532546,
      "learning_rate": 6.103439239296638e-07,
      "loss": 0.1528,
      "step": 11825
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5016383123625436,
      "learning_rate": 6.097894130448584e-07,
      "loss": 0.1843,
      "step": 11826
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3230757480386068,
      "learning_rate": 6.092351378091161e-07,
      "loss": 0.1588,
      "step": 11827
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.404649833155314,
      "learning_rate": 6.086810982521891e-07,
      "loss": 0.1943,
      "step": 11828
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5628874625177338,
      "learning_rate": 6.081272944038141e-07,
      "loss": 0.1943,
      "step": 11829
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2319735528470148,
      "learning_rate": 6.075737262937187e-07,
      "loss": 0.1379,
      "step": 11830
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3462818246507613,
      "learning_rate": 6.070203939516145e-07,
      "loss": 0.1864,
      "step": 11831
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2711814688721346,
      "learning_rate": 6.064672974072044e-07,
      "loss": 0.1337,
      "step": 11832
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4983709367505482,
      "learning_rate": 6.059144366901737e-07,
      "loss": 0.1966,
      "step": 11833
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3814432226186442,
      "learning_rate": 6.053618118301996e-07,
      "loss": 0.1765,
      "step": 11834
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.234577445202359,
      "learning_rate": 6.048094228569445e-07,
      "loss": 0.1254,
      "step": 11835
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4391204341033825,
      "learning_rate": 6.042572698000592e-07,
      "loss": 0.1679,
      "step": 11836
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.404540471495477,
      "learning_rate": 6.037053526891795e-07,
      "loss": 0.1732,
      "step": 11837
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4652753241611205,
      "learning_rate": 6.03153671553931e-07,
      "loss": 0.1885,
      "step": 11838
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2230926520136514,
      "learning_rate": 6.026022264239262e-07,
      "loss": 0.1817,
      "step": 11839
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.009650851859441,
      "learning_rate": 6.020510173287636e-07,
      "loss": 0.5332,
      "step": 11840
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.5444583602084005,
      "learning_rate": 6.015000442980307e-07,
      "loss": 0.6531,
      "step": 11841
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1929639206071763,
      "learning_rate": 6.009493073613004e-07,
      "loss": 0.1538,
      "step": 11842
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3628521254901884,
      "learning_rate": 6.003988065481347e-07,
      "loss": 0.1963,
      "step": 11843
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.361486919745974,
      "learning_rate": 5.998485418880822e-07,
      "loss": 0.1623,
      "step": 11844
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.344447512503299,
      "learning_rate": 5.992985134106805e-07,
      "loss": 0.154,
      "step": 11845
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.337377575003357,
      "learning_rate": 5.987487211454502e-07,
      "loss": 0.1953,
      "step": 11846
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5198571468397921,
      "learning_rate": 5.981991651219043e-07,
      "loss": 0.1819,
      "step": 11847
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2846263944982605,
      "learning_rate": 5.97649845369539e-07,
      "loss": 0.1727,
      "step": 11848
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3924642609089055,
      "learning_rate": 5.971007619178415e-07,
      "loss": 0.2059,
      "step": 11849
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4255315287521797,
      "learning_rate": 5.965519147962823e-07,
      "loss": 0.151,
      "step": 11850
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.488422612454247,
      "learning_rate": 5.960033040343222e-07,
      "loss": 0.1699,
      "step": 11851
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3336444806956729,
      "learning_rate": 5.954549296614082e-07,
      "loss": 0.1725,
      "step": 11852
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0855948361912515,
      "learning_rate": 5.949067917069768e-07,
      "loss": 0.111,
      "step": 11853
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.6050991328470054,
      "learning_rate": 5.943588902004471e-07,
      "loss": 0.6256,
      "step": 11854
    },
    {
      "epoch": 0.85,
      "grad_norm": 10.750108234176988,
      "learning_rate": 5.938112251712308e-07,
      "loss": 0.5607,
      "step": 11855
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.424404622560542,
      "learning_rate": 5.932637966487214e-07,
      "loss": 0.527,
      "step": 11856
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5139715575552102,
      "learning_rate": 5.927166046623062e-07,
      "loss": 0.2156,
      "step": 11857
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4870681731113684,
      "learning_rate": 5.921696492413525e-07,
      "loss": 0.166,
      "step": 11858
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3826070631527563,
      "learning_rate": 5.916229304152215e-07,
      "loss": 0.2079,
      "step": 11859
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2969889297987252,
      "learning_rate": 5.910764482132575e-07,
      "loss": 0.1788,
      "step": 11860
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.519378427113825,
      "learning_rate": 5.905302026647953e-07,
      "loss": 0.1551,
      "step": 11861
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3762181285482564,
      "learning_rate": 5.899841937991529e-07,
      "loss": 0.1719,
      "step": 11862
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2133694256920355,
      "learning_rate": 5.8943842164564e-07,
      "loss": 0.1544,
      "step": 11863
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.155096680459281,
      "learning_rate": 5.888928862335491e-07,
      "loss": 0.5704,
      "step": 11864
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1743679456871643,
      "learning_rate": 5.883475875921646e-07,
      "loss": 0.1289,
      "step": 11865
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.382454117439039,
      "learning_rate": 5.878025257507547e-07,
      "loss": 0.1557,
      "step": 11866
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3101415689207183,
      "learning_rate": 5.87257700738576e-07,
      "loss": 0.1904,
      "step": 11867
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3378263236375791,
      "learning_rate": 5.867131125848729e-07,
      "loss": 0.1817,
      "step": 11868
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3912503329505317,
      "learning_rate": 5.861687613188782e-07,
      "loss": 0.1682,
      "step": 11869
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2552846866997394,
      "learning_rate": 5.856246469698079e-07,
      "loss": 0.1674,
      "step": 11870
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3633125848770604,
      "learning_rate": 5.850807695668692e-07,
      "loss": 0.1914,
      "step": 11871
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.291762433235719,
      "learning_rate": 5.845371291392565e-07,
      "loss": 0.1907,
      "step": 11872
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3103926316467114,
      "learning_rate": 5.839937257161477e-07,
      "loss": 0.142,
      "step": 11873
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.41188789321237,
      "learning_rate": 5.834505593267131e-07,
      "loss": 0.221,
      "step": 11874
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5025406468682707,
      "learning_rate": 5.829076300001052e-07,
      "loss": 0.1781,
      "step": 11875
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4440595148580095,
      "learning_rate": 5.823649377654678e-07,
      "loss": 0.1873,
      "step": 11876
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4167769567823612,
      "learning_rate": 5.818224826519297e-07,
      "loss": 0.1724,
      "step": 11877
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4548155939555418,
      "learning_rate": 5.812802646886095e-07,
      "loss": 0.1742,
      "step": 11878
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3277806643217567,
      "learning_rate": 5.807382839046094e-07,
      "loss": 0.1885,
      "step": 11879
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3327495249796568,
      "learning_rate": 5.801965403290221e-07,
      "loss": 0.1364,
      "step": 11880
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5758238212909672,
      "learning_rate": 5.796550339909245e-07,
      "loss": 0.1929,
      "step": 11881
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3367845920743415,
      "learning_rate": 5.791137649193851e-07,
      "loss": 0.1671,
      "step": 11882
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3365213228439465,
      "learning_rate": 5.785727331434543e-07,
      "loss": 0.1397,
      "step": 11883
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3254073485998006,
      "learning_rate": 5.780319386921745e-07,
      "loss": 0.2032,
      "step": 11884
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3542488455385189,
      "learning_rate": 5.77491381594571e-07,
      "loss": 0.1794,
      "step": 11885
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2519627811573155,
      "learning_rate": 5.76951061879662e-07,
      "loss": 0.1554,
      "step": 11886
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5007446306421866,
      "learning_rate": 5.764109795764478e-07,
      "loss": 0.2027,
      "step": 11887
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5364241678623267,
      "learning_rate": 5.758711347139189e-07,
      "loss": 0.1873,
      "step": 11888
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2885868043790423,
      "learning_rate": 5.753315273210502e-07,
      "loss": 0.1566,
      "step": 11889
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3040430826874798,
      "learning_rate": 5.747921574268078e-07,
      "loss": 0.1877,
      "step": 11890
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3496253365252762,
      "learning_rate": 5.742530250601414e-07,
      "loss": 0.1865,
      "step": 11891
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1181022245888634,
      "learning_rate": 5.73714130249991e-07,
      "loss": 0.1205,
      "step": 11892
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.427943808137582,
      "learning_rate": 5.731754730252797e-07,
      "loss": 0.1966,
      "step": 11893
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.641824724978169,
      "learning_rate": 5.726370534149228e-07,
      "loss": 0.2237,
      "step": 11894
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3459842522768013,
      "learning_rate": 5.720988714478199e-07,
      "loss": 0.1399,
      "step": 11895
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4086366166429791,
      "learning_rate": 5.715609271528594e-07,
      "loss": 0.1783,
      "step": 11896
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3691981343784334,
      "learning_rate": 5.710232205589139e-07,
      "loss": 0.1722,
      "step": 11897
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1988263420978127,
      "learning_rate": 5.704857516948476e-07,
      "loss": 0.1427,
      "step": 11898
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4724141375899162,
      "learning_rate": 5.699485205895078e-07,
      "loss": 0.1796,
      "step": 11899
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.308025678307626,
      "learning_rate": 5.694115272717326e-07,
      "loss": 0.1654,
      "step": 11900
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1472654075812014,
      "learning_rate": 5.688747717703436e-07,
      "loss": 0.1401,
      "step": 11901
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5054923709289376,
      "learning_rate": 5.68338254114153e-07,
      "loss": 0.1842,
      "step": 11902
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3527434717453437,
      "learning_rate": 5.67801974331959e-07,
      "loss": 0.1592,
      "step": 11903
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1855312617240594,
      "learning_rate": 5.67265932452547e-07,
      "loss": 0.152,
      "step": 11904
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3820501160084664,
      "learning_rate": 5.6673012850469e-07,
      "loss": 0.2014,
      "step": 11905
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3302475707209809,
      "learning_rate": 5.661945625171467e-07,
      "loss": 0.1841,
      "step": 11906
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2097195984335025,
      "learning_rate": 5.656592345186657e-07,
      "loss": 0.161,
      "step": 11907
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4124420566582243,
      "learning_rate": 5.65124144537979e-07,
      "loss": 0.1736,
      "step": 11908
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5064478562026378,
      "learning_rate": 5.645892926038105e-07,
      "loss": 0.1689,
      "step": 11909
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4037128965418295,
      "learning_rate": 5.64054678744867e-07,
      "loss": 0.1953,
      "step": 11910
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1588590428448058,
      "learning_rate": 5.635203029898451e-07,
      "loss": 0.1258,
      "step": 11911
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4009223522939578,
      "learning_rate": 5.629861653674279e-07,
      "loss": 0.1888,
      "step": 11912
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.239482489537595,
      "learning_rate": 5.624522659062875e-07,
      "loss": 0.1608,
      "step": 11913
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8389237057585537,
      "learning_rate": 5.619186046350788e-07,
      "loss": 0.1542,
      "step": 11914
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3317044109873473,
      "learning_rate": 5.613851815824489e-07,
      "loss": 0.143,
      "step": 11915
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2578699671643352,
      "learning_rate": 5.608519967770276e-07,
      "loss": 0.1716,
      "step": 11916
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1035110092075342,
      "learning_rate": 5.603190502474365e-07,
      "loss": 0.1357,
      "step": 11917
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3423718411181578,
      "learning_rate": 5.597863420222793e-07,
      "loss": 0.1727,
      "step": 11918
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3441268516481935,
      "learning_rate": 5.592538721301522e-07,
      "loss": 0.1302,
      "step": 11919
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4640847294429042,
      "learning_rate": 5.587216405996343e-07,
      "loss": 0.1807,
      "step": 11920
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5212014573704855,
      "learning_rate": 5.581896474592957e-07,
      "loss": 0.2031,
      "step": 11921
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.215648286912998,
      "learning_rate": 5.576578927376897e-07,
      "loss": 0.1542,
      "step": 11922
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1912643265601766,
      "learning_rate": 5.571263764633606e-07,
      "loss": 0.1372,
      "step": 11923
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4463365351938686,
      "learning_rate": 5.565950986648355e-07,
      "loss": 0.1731,
      "step": 11924
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.008005449135428,
      "learning_rate": 5.56064059370634e-07,
      "loss": 0.3738,
      "step": 11925
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2596222522530265,
      "learning_rate": 5.555332586092583e-07,
      "loss": 0.2166,
      "step": 11926
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2374533258730707,
      "learning_rate": 5.550026964091998e-07,
      "loss": 0.1487,
      "step": 11927
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2301517607904666,
      "learning_rate": 5.544723727989377e-07,
      "loss": 0.18,
      "step": 11928
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2324063556645592,
      "learning_rate": 5.539422878069389e-07,
      "loss": 0.1503,
      "step": 11929
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.970387239803195,
      "learning_rate": 5.534124414616532e-07,
      "loss": 0.6601,
      "step": 11930
    },
    {
      "epoch": 0.85,
      "grad_norm": 8.668238456586904,
      "learning_rate": 5.528828337915237e-07,
      "loss": 0.5707,
      "step": 11931
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.241390059160808,
      "learning_rate": 5.523534648249751e-07,
      "loss": 0.6603,
      "step": 11932
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3193667062433727,
      "learning_rate": 5.518243345904239e-07,
      "loss": 0.1603,
      "step": 11933
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2033244935184737,
      "learning_rate": 5.512954431162698e-07,
      "loss": 0.1925,
      "step": 11934
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.1683665110223105,
      "learning_rate": 5.507667904309027e-07,
      "loss": 0.38,
      "step": 11935
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3531119292411646,
      "learning_rate": 5.502383765626984e-07,
      "loss": 0.1875,
      "step": 11936
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5142867063195167,
      "learning_rate": 5.497102015400197e-07,
      "loss": 0.1644,
      "step": 11937
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2215437916349061,
      "learning_rate": 5.491822653912188e-07,
      "loss": 0.126,
      "step": 11938
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3326128268066109,
      "learning_rate": 5.486545681446303e-07,
      "loss": 0.2022,
      "step": 11939
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.926809551267746,
      "learning_rate": 5.481271098285818e-07,
      "loss": 0.3574,
      "step": 11940
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2702723643377511,
      "learning_rate": 5.475998904713825e-07,
      "loss": 0.149,
      "step": 11941
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.330307227865044,
      "learning_rate": 5.470729101013333e-07,
      "loss": 0.5359,
      "step": 11942
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4493380738552404,
      "learning_rate": 5.465461687467194e-07,
      "loss": 0.1957,
      "step": 11943
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4294740018377265,
      "learning_rate": 5.460196664358141e-07,
      "loss": 0.1504,
      "step": 11944
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3496267056338718,
      "learning_rate": 5.454934031968784e-07,
      "loss": 0.1467,
      "step": 11945
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1514274541346865,
      "learning_rate": 5.449673790581611e-07,
      "loss": 0.1405,
      "step": 11946
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.40909698283343,
      "learning_rate": 5.444415940478948e-07,
      "loss": 0.1658,
      "step": 11947
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3922497019313413,
      "learning_rate": 5.43916048194304e-07,
      "loss": 0.1525,
      "step": 11948
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.2525178338145695,
      "learning_rate": 5.433907415255951e-07,
      "loss": 0.5715,
      "step": 11949
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4388383242469989,
      "learning_rate": 5.428656740699678e-07,
      "loss": 0.1558,
      "step": 11950
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1839678015650463,
      "learning_rate": 5.423408458556023e-07,
      "loss": 0.1806,
      "step": 11951
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3729480158384668,
      "learning_rate": 5.418162569106711e-07,
      "loss": 0.1988,
      "step": 11952
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.322635509044266,
      "learning_rate": 5.412919072633316e-07,
      "loss": 0.4371,
      "step": 11953
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.062667417592494,
      "learning_rate": 5.407677969417302e-07,
      "loss": 0.427,
      "step": 11954
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6011492676449068,
      "learning_rate": 5.402439259739972e-07,
      "loss": 0.1656,
      "step": 11955
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4969033099173448,
      "learning_rate": 5.397202943882529e-07,
      "loss": 0.1851,
      "step": 11956
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.328170343331863,
      "learning_rate": 5.391969022126031e-07,
      "loss": 0.171,
      "step": 11957
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2773728805842328,
      "learning_rate": 5.386737494751421e-07,
      "loss": 0.1395,
      "step": 11958
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.5357603853826465,
      "learning_rate": 5.381508362039495e-07,
      "loss": 0.507,
      "step": 11959
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.329689928153197,
      "learning_rate": 5.376281624270946e-07,
      "loss": 0.1701,
      "step": 11960
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3175595070852766,
      "learning_rate": 5.371057281726311e-07,
      "loss": 0.1808,
      "step": 11961
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4276482219442137,
      "learning_rate": 5.365835334686037e-07,
      "loss": 0.1807,
      "step": 11962
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3182528099412956,
      "learning_rate": 5.360615783430384e-07,
      "loss": 0.1837,
      "step": 11963
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2327930372495695,
      "learning_rate": 5.355398628239544e-07,
      "loss": 0.1477,
      "step": 11964
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.533575832596085,
      "learning_rate": 5.350183869393532e-07,
      "loss": 0.1644,
      "step": 11965
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.377566068246399,
      "learning_rate": 5.34497150717227e-07,
      "loss": 0.1947,
      "step": 11966
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.271684749098478,
      "learning_rate": 5.339761541855537e-07,
      "loss": 0.1498,
      "step": 11967
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2599098109710354,
      "learning_rate": 5.334553973722973e-07,
      "loss": 0.1541,
      "step": 11968
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1781500964320193,
      "learning_rate": 5.329348803054101e-07,
      "loss": 0.1351,
      "step": 11969
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3612329107125847,
      "learning_rate": 5.324146030128325e-07,
      "loss": 0.1537,
      "step": 11970
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3714680404078448,
      "learning_rate": 5.318945655224905e-07,
      "loss": 0.1806,
      "step": 11971
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4767216159160663,
      "learning_rate": 5.313747678622965e-07,
      "loss": 0.1611,
      "step": 11972
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1490626778464939,
      "learning_rate": 5.308552100601533e-07,
      "loss": 0.1363,
      "step": 11973
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2054626630330978,
      "learning_rate": 5.303358921439467e-07,
      "loss": 0.1527,
      "step": 11974
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1832835314945667,
      "learning_rate": 5.298168141415528e-07,
      "loss": 0.1268,
      "step": 11975
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.835606836554857,
      "learning_rate": 5.292979760808326e-07,
      "loss": 0.6086,
      "step": 11976
    },
    {
      "epoch": 0.86,
      "grad_norm": 10.747572637787231,
      "learning_rate": 5.287793779896356e-07,
      "loss": 0.5843,
      "step": 11977
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.675132587305952,
      "learning_rate": 5.282610198957988e-07,
      "loss": 0.6761,
      "step": 11978
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2127876735036949,
      "learning_rate": 5.277429018271457e-07,
      "loss": 0.1555,
      "step": 11979
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2852387539900954,
      "learning_rate": 5.272250238114857e-07,
      "loss": 0.1607,
      "step": 11980
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3277913063459401,
      "learning_rate": 5.267073858766176e-07,
      "loss": 0.1896,
      "step": 11981
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.223891402652373,
      "learning_rate": 5.261899880503246e-07,
      "loss": 0.1525,
      "step": 11982
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2899044232379844,
      "learning_rate": 5.256728303603809e-07,
      "loss": 0.1627,
      "step": 11983
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.286517861900291,
      "learning_rate": 5.251559128345429e-07,
      "loss": 0.5475,
      "step": 11984
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4455629094842346,
      "learning_rate": 5.246392355005591e-07,
      "loss": 0.1665,
      "step": 11985
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3091123115345589,
      "learning_rate": 5.241227983861591e-07,
      "loss": 0.1447,
      "step": 11986
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3418773987432282,
      "learning_rate": 5.236066015190677e-07,
      "loss": 0.1754,
      "step": 11987
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2661347586246718,
      "learning_rate": 5.230906449269895e-07,
      "loss": 0.1606,
      "step": 11988
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3424979374870976,
      "learning_rate": 5.225749286376202e-07,
      "loss": 0.1891,
      "step": 11989
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.213095258539563,
      "learning_rate": 5.220594526786405e-07,
      "loss": 0.5789,
      "step": 11990
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2379775644078874,
      "learning_rate": 5.215442170777196e-07,
      "loss": 0.1811,
      "step": 11991
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.132940374364009,
      "learning_rate": 5.210292218625129e-07,
      "loss": 0.133,
      "step": 11992
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.320505003258881,
      "learning_rate": 5.205144670606643e-07,
      "loss": 0.1855,
      "step": 11993
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3812496057614005,
      "learning_rate": 5.199999526998017e-07,
      "loss": 0.1828,
      "step": 11994
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4366874102412701,
      "learning_rate": 5.194856788075442e-07,
      "loss": 0.1678,
      "step": 11995
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5433334628647148,
      "learning_rate": 5.189716454114946e-07,
      "loss": 0.1797,
      "step": 11996
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.283487865577448,
      "learning_rate": 5.184578525392464e-07,
      "loss": 0.181,
      "step": 11997
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5930113738339486,
      "learning_rate": 5.179443002183754e-07,
      "loss": 0.1579,
      "step": 11998
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.9101051832294145,
      "learning_rate": 5.174309884764478e-07,
      "loss": 0.5179,
      "step": 11999
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3638396987401733,
      "learning_rate": 5.169179173410178e-07,
      "loss": 0.1742,
      "step": 12000
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3041162873666514,
      "learning_rate": 5.164050868396225e-07,
      "loss": 0.1462,
      "step": 12001
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5386301177121489,
      "learning_rate": 5.158924969997908e-07,
      "loss": 0.1603,
      "step": 12002
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3901668082452658,
      "learning_rate": 5.153801478490344e-07,
      "loss": 0.1696,
      "step": 12003
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.109902231292228,
      "learning_rate": 5.148680394148553e-07,
      "loss": 0.1343,
      "step": 12004
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.410708097142772,
      "learning_rate": 5.143561717247414e-07,
      "loss": 0.1729,
      "step": 12005
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1395484611839373,
      "learning_rate": 5.138445448061686e-07,
      "loss": 0.1286,
      "step": 12006
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.323453445405225,
      "learning_rate": 5.13333158686597e-07,
      "loss": 0.1292,
      "step": 12007
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3288784757288983,
      "learning_rate": 5.12822013393478e-07,
      "loss": 0.1591,
      "step": 12008
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.232465708975618,
      "learning_rate": 5.123111089542454e-07,
      "loss": 0.1396,
      "step": 12009
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1555252943829946,
      "learning_rate": 5.118004453963255e-07,
      "loss": 0.1474,
      "step": 12010
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3660256158512312,
      "learning_rate": 5.112900227471257e-07,
      "loss": 0.1821,
      "step": 12011
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.988651188083765,
      "learning_rate": 5.107798410340442e-07,
      "loss": 0.6405,
      "step": 12012
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2369970791967717,
      "learning_rate": 5.102699002844669e-07,
      "loss": 0.1626,
      "step": 12013
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3417701095247825,
      "learning_rate": 5.097602005257651e-07,
      "loss": 0.1947,
      "step": 12014
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.340964331175679,
      "learning_rate": 5.09250741785296e-07,
      "loss": 0.1329,
      "step": 12015
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4358438432050118,
      "learning_rate": 5.087415240904076e-07,
      "loss": 0.1638,
      "step": 12016
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4028479394363496,
      "learning_rate": 5.082325474684303e-07,
      "loss": 0.1605,
      "step": 12017
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5407047053725136,
      "learning_rate": 5.07723811946686e-07,
      "loss": 0.2027,
      "step": 12018
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3479755638075555,
      "learning_rate": 5.072153175524791e-07,
      "loss": 0.1564,
      "step": 12019
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3777309483966722,
      "learning_rate": 5.067070643131056e-07,
      "loss": 0.1539,
      "step": 12020
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1516838456526879,
      "learning_rate": 5.061990522558457e-07,
      "loss": 0.1724,
      "step": 12021
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1735812211725698,
      "learning_rate": 5.05691281407969e-07,
      "loss": 0.1746,
      "step": 12022
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2816709672405695,
      "learning_rate": 5.051837517967284e-07,
      "loss": 0.1886,
      "step": 12023
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3285365148746178,
      "learning_rate": 5.046764634493678e-07,
      "loss": 0.1607,
      "step": 12024
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4605427995706401,
      "learning_rate": 5.041694163931149e-07,
      "loss": 0.1641,
      "step": 12025
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4480998193182022,
      "learning_rate": 5.036626106551879e-07,
      "loss": 0.208,
      "step": 12026
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5182937714709108,
      "learning_rate": 5.031560462627876e-07,
      "loss": 0.2011,
      "step": 12027
    },
    {
      "epoch": 0.86,
      "grad_norm": 10.622260698591672,
      "learning_rate": 5.026497232431055e-07,
      "loss": 0.4959,
      "step": 12028
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3042977109013212,
      "learning_rate": 5.021436416233199e-07,
      "loss": 0.184,
      "step": 12029
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2791144721053482,
      "learning_rate": 5.01637801430594e-07,
      "loss": 0.176,
      "step": 12030
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.3087952987783895,
      "learning_rate": 5.011322026920812e-07,
      "loss": 0.4897,
      "step": 12031
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2479862515079598,
      "learning_rate": 5.006268454349178e-07,
      "loss": 0.1312,
      "step": 12032
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5527835678744886,
      "learning_rate": 5.00121729686231e-07,
      "loss": 0.1611,
      "step": 12033
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.227772118329657,
      "learning_rate": 4.996168554731317e-07,
      "loss": 0.1465,
      "step": 12034
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3805144787449908,
      "learning_rate": 4.991122228227213e-07,
      "loss": 0.1985,
      "step": 12035
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2272353356267884,
      "learning_rate": 4.986078317620852e-07,
      "loss": 0.1445,
      "step": 12036
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.407514163288947,
      "learning_rate": 4.981036823182972e-07,
      "loss": 0.1679,
      "step": 12037
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4485648221990761,
      "learning_rate": 4.975997745184191e-07,
      "loss": 0.2106,
      "step": 12038
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.469848568371563,
      "learning_rate": 4.970961083894982e-07,
      "loss": 0.1835,
      "step": 12039
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3033805816871606,
      "learning_rate": 4.965926839585688e-07,
      "loss": 0.1628,
      "step": 12040
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3073392957575978,
      "learning_rate": 4.960895012526534e-07,
      "loss": 0.1491,
      "step": 12041
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4108953663455581,
      "learning_rate": 4.955865602987598e-07,
      "loss": 0.187,
      "step": 12042
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3368221979909678,
      "learning_rate": 4.950838611238856e-07,
      "loss": 0.1658,
      "step": 12043
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2549017462884158,
      "learning_rate": 4.945814037550117e-07,
      "loss": 0.1591,
      "step": 12044
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5265264799438054,
      "learning_rate": 4.940791882191087e-07,
      "loss": 0.1773,
      "step": 12045
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3065146281181967,
      "learning_rate": 4.935772145431344e-07,
      "loss": 0.1983,
      "step": 12046
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4288383301541023,
      "learning_rate": 4.930754827540324e-07,
      "loss": 0.2015,
      "step": 12047
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3259444683598571,
      "learning_rate": 4.925739928787332e-07,
      "loss": 0.1618,
      "step": 12048
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2555773763850806,
      "learning_rate": 4.920727449441554e-07,
      "loss": 0.176,
      "step": 12049
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2929795110541837,
      "learning_rate": 4.915717389772029e-07,
      "loss": 0.1584,
      "step": 12050
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4207443334974492,
      "learning_rate": 4.910709750047693e-07,
      "loss": 0.1801,
      "step": 12051
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3263765222855792,
      "learning_rate": 4.905704530537314e-07,
      "loss": 0.1303,
      "step": 12052
    },
    {
      "epoch": 0.86,
      "grad_norm": 9.52274707143638,
      "learning_rate": 4.900701731509571e-07,
      "loss": 0.6399,
      "step": 12053
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.279330411279231,
      "learning_rate": 4.895701353232984e-07,
      "loss": 0.127,
      "step": 12054
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2767071359998512,
      "learning_rate": 4.890703395975976e-07,
      "loss": 0.183,
      "step": 12055
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.333287718600134,
      "learning_rate": 4.885707860006783e-07,
      "loss": 0.4976,
      "step": 12056
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2324529264422999,
      "learning_rate": 4.880714745593574e-07,
      "loss": 0.1745,
      "step": 12057
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1560019255779528,
      "learning_rate": 4.875724053004339e-07,
      "loss": 0.1137,
      "step": 12058
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4088661842911823,
      "learning_rate": 4.87073578250698e-07,
      "loss": 0.1639,
      "step": 12059
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3968088815528998,
      "learning_rate": 4.865749934369224e-07,
      "loss": 0.1548,
      "step": 12060
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5418992476934186,
      "learning_rate": 4.860766508858699e-07,
      "loss": 0.1819,
      "step": 12061
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.384114995152617,
      "learning_rate": 4.855785506242899e-07,
      "loss": 0.2188,
      "step": 12062
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4839897990197475,
      "learning_rate": 4.850806926789192e-07,
      "loss": 0.1792,
      "step": 12063
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.363981383431486,
      "learning_rate": 4.845830770764803e-07,
      "loss": 0.16,
      "step": 12064
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.300397110449409,
      "learning_rate": 4.840857038436825e-07,
      "loss": 0.4779,
      "step": 12065
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.272100988282035,
      "learning_rate": 4.835885730072243e-07,
      "loss": 0.1577,
      "step": 12066
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3094773393893258,
      "learning_rate": 4.830916845937872e-07,
      "loss": 0.1715,
      "step": 12067
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3326900119217566,
      "learning_rate": 4.825950386300454e-07,
      "loss": 0.161,
      "step": 12068
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3655936580907897,
      "learning_rate": 4.820986351426537e-07,
      "loss": 0.1886,
      "step": 12069
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2175696161197458,
      "learning_rate": 4.816024741582593e-07,
      "loss": 0.1542,
      "step": 12070
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2580143874975975,
      "learning_rate": 4.811065557034927e-07,
      "loss": 0.1672,
      "step": 12071
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4382745973561353,
      "learning_rate": 4.806108798049752e-07,
      "loss": 0.1982,
      "step": 12072
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1003844880306466,
      "learning_rate": 4.801154464893098e-07,
      "loss": 0.1419,
      "step": 12073
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3682629936777522,
      "learning_rate": 4.796202557830914e-07,
      "loss": 0.1543,
      "step": 12074
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3061279550821403,
      "learning_rate": 4.791253077128988e-07,
      "loss": 0.1545,
      "step": 12075
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.872187046039576,
      "learning_rate": 4.786306023053e-07,
      "loss": 0.5414,
      "step": 12076
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.034567788464287,
      "learning_rate": 4.781361395868467e-07,
      "loss": 0.4311,
      "step": 12077
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2660147875931582,
      "learning_rate": 4.776419195840809e-07,
      "loss": 0.1669,
      "step": 12078
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2303544936107844,
      "learning_rate": 4.771479423235309e-07,
      "loss": 0.1619,
      "step": 12079
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4749718448657627,
      "learning_rate": 4.766542078317121e-07,
      "loss": 0.2042,
      "step": 12080
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3076443156159239,
      "learning_rate": 4.7616071613512394e-07,
      "loss": 0.1586,
      "step": 12081
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.388036453829973,
      "learning_rate": 4.756674672602574e-07,
      "loss": 0.1478,
      "step": 12082
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.251372578781535,
      "learning_rate": 4.7517446123358566e-07,
      "loss": 0.1719,
      "step": 12083
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1686357435779169,
      "learning_rate": 4.746816980815738e-07,
      "loss": 0.1228,
      "step": 12084
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2481857885447187,
      "learning_rate": 4.7418917783066956e-07,
      "loss": 0.1815,
      "step": 12085
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4546438282829435,
      "learning_rate": 4.736969005073105e-07,
      "loss": 0.1735,
      "step": 12086
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2393235951392096,
      "learning_rate": 4.7320486613791904e-07,
      "loss": 0.1515,
      "step": 12087
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4670446776653145,
      "learning_rate": 4.727130747489067e-07,
      "loss": 0.1651,
      "step": 12088
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.346087553083383,
      "learning_rate": 4.722215263666702e-07,
      "loss": 0.1577,
      "step": 12089
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4215527901831406,
      "learning_rate": 4.717302210175956e-07,
      "loss": 0.1732,
      "step": 12090
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2358803005810628,
      "learning_rate": 4.712391587280518e-07,
      "loss": 0.1483,
      "step": 12091
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.212986776261093,
      "learning_rate": 4.707483395243989e-07,
      "loss": 0.1601,
      "step": 12092
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.188765331400923,
      "learning_rate": 4.7025776343298023e-07,
      "loss": 0.1384,
      "step": 12093
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4146078484558355,
      "learning_rate": 4.697674304801297e-07,
      "loss": 0.2004,
      "step": 12094
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2555503325213395,
      "learning_rate": 4.6927734069216633e-07,
      "loss": 0.1341,
      "step": 12095
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3997454774395999,
      "learning_rate": 4.6878749409539515e-07,
      "loss": 0.2178,
      "step": 12096
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3440107209857164,
      "learning_rate": 4.6829789071610955e-07,
      "loss": 0.1355,
      "step": 12097
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3818965012018785,
      "learning_rate": 4.678085305805896e-07,
      "loss": 0.1654,
      "step": 12098
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2401974217328104,
      "learning_rate": 4.6731941371510324e-07,
      "loss": 0.2002,
      "step": 12099
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3624157940766028,
      "learning_rate": 4.668305401459022e-07,
      "loss": 0.1627,
      "step": 12100
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.141679817638267,
      "learning_rate": 4.6634190989922947e-07,
      "loss": 0.537,
      "step": 12101
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5026080738426155,
      "learning_rate": 4.6585352300131113e-07,
      "loss": 0.1778,
      "step": 12102
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.193631947611065,
      "learning_rate": 4.6536537947836356e-07,
      "loss": 0.1548,
      "step": 12103
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4887071873815263,
      "learning_rate": 4.6487747935658576e-07,
      "loss": 0.1846,
      "step": 12104
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5976807457233757,
      "learning_rate": 4.6438982266216793e-07,
      "loss": 0.1856,
      "step": 12105
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4101931936809016,
      "learning_rate": 4.6390240942128573e-07,
      "loss": 0.2156,
      "step": 12106
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3742170074788973,
      "learning_rate": 4.634152396601016e-07,
      "loss": 0.1803,
      "step": 12107
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.352882633235185,
      "learning_rate": 4.6292831340476406e-07,
      "loss": 0.1483,
      "step": 12108
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4546025549351913,
      "learning_rate": 4.6244163068141057e-07,
      "loss": 0.1453,
      "step": 12109
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.33421094366108,
      "learning_rate": 4.61955191516163e-07,
      "loss": 0.1662,
      "step": 12110
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4363937926170578,
      "learning_rate": 4.614689959351326e-07,
      "loss": 0.1947,
      "step": 12111
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4674549401563293,
      "learning_rate": 4.6098304396441474e-07,
      "loss": 0.2094,
      "step": 12112
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5929176435326713,
      "learning_rate": 4.6049733563009515e-07,
      "loss": 0.1854,
      "step": 12113
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3610287350878238,
      "learning_rate": 4.600118709582435e-07,
      "loss": 0.1604,
      "step": 12114
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3249234947239992,
      "learning_rate": 4.595266499749196e-07,
      "loss": 0.1686,
      "step": 12115
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.822404599675312,
      "learning_rate": 4.5904167270616527e-07,
      "loss": 0.4698,
      "step": 12116
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2964428425537264,
      "learning_rate": 4.5855693917801537e-07,
      "loss": 0.1649,
      "step": 12117
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2477238355064513,
      "learning_rate": 4.580724494164851e-07,
      "loss": 0.1611,
      "step": 12118
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3805966659786733,
      "learning_rate": 4.5758820344758317e-07,
      "loss": 0.1944,
      "step": 12119
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4177783647063877,
      "learning_rate": 4.571042012972993e-07,
      "loss": 0.1699,
      "step": 12120
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3519721452161142,
      "learning_rate": 4.566204429916138e-07,
      "loss": 0.1789,
      "step": 12121
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2065207392224284,
      "learning_rate": 4.5613692855649316e-07,
      "loss": 0.138,
      "step": 12122
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5021114733730614,
      "learning_rate": 4.556536580178911e-07,
      "loss": 0.1909,
      "step": 12123
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2686956146474246,
      "learning_rate": 4.551706314017462e-07,
      "loss": 0.1824,
      "step": 12124
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3342065932239426,
      "learning_rate": 4.5468784873398733e-07,
      "loss": 0.1406,
      "step": 12125
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1378828874411688,
      "learning_rate": 4.5420531004052584e-07,
      "loss": 0.156,
      "step": 12126
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4866697701096736,
      "learning_rate": 4.537230153472638e-07,
      "loss": 0.1755,
      "step": 12127
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2237092356301258,
      "learning_rate": 4.5324096468009006e-07,
      "loss": 0.1278,
      "step": 12128
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2340765133303346,
      "learning_rate": 4.527591580648771e-07,
      "loss": 0.1543,
      "step": 12129
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.175850339966986,
      "learning_rate": 4.5227759552748716e-07,
      "loss": 0.1296,
      "step": 12130
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2948577160880215,
      "learning_rate": 4.5179627709376893e-07,
      "loss": 0.1857,
      "step": 12131
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.157772732853982,
      "learning_rate": 4.5131520278955785e-07,
      "loss": 0.156,
      "step": 12132
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1985947240732575,
      "learning_rate": 4.5083437264067555e-07,
      "loss": 0.1598,
      "step": 12133
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3342352506877397,
      "learning_rate": 4.503537866729318e-07,
      "loss": 0.1823,
      "step": 12134
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1872918917230555,
      "learning_rate": 4.4987344491212117e-07,
      "loss": 0.0958,
      "step": 12135
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.798849286541649,
      "learning_rate": 4.493933473840284e-07,
      "loss": 0.4926,
      "step": 12136
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2694586966163712,
      "learning_rate": 4.4891349411442074e-07,
      "loss": 0.1614,
      "step": 12137
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2484985522759429,
      "learning_rate": 4.484338851290565e-07,
      "loss": 0.1433,
      "step": 12138
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4386659174564058,
      "learning_rate": 4.479545204536789e-07,
      "loss": 0.1483,
      "step": 12139
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4166666368515797,
      "learning_rate": 4.4747540011401913e-07,
      "loss": 0.1854,
      "step": 12140
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.198211637051815,
      "learning_rate": 4.469965241357932e-07,
      "loss": 0.1526,
      "step": 12141
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3860905054221737,
      "learning_rate": 4.465178925447061e-07,
      "loss": 0.1931,
      "step": 12142
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.246253008483393,
      "learning_rate": 4.460395053664479e-07,
      "loss": 0.469,
      "step": 12143
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4453350166592382,
      "learning_rate": 4.45561362626698e-07,
      "loss": 0.1603,
      "step": 12144
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.169594590585564,
      "learning_rate": 4.450834643511193e-07,
      "loss": 0.7001,
      "step": 12145
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.666100207694078,
      "learning_rate": 4.446058105653644e-07,
      "loss": 0.7823,
      "step": 12146
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6227531398697075,
      "learning_rate": 4.4412840129507194e-07,
      "loss": 0.167,
      "step": 12147
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2583613161113556,
      "learning_rate": 4.4365123656586853e-07,
      "loss": 0.1375,
      "step": 12148
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2330254891650425,
      "learning_rate": 4.431743164033647e-07,
      "loss": 0.179,
      "step": 12149
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3854823515773331,
      "learning_rate": 4.426976408331607e-07,
      "loss": 0.1765,
      "step": 12150
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2045150808063168,
      "learning_rate": 4.422212098808415e-07,
      "loss": 0.1384,
      "step": 12151
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2721965538461268,
      "learning_rate": 4.417450235719811e-07,
      "loss": 0.1426,
      "step": 12152
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2991215730071053,
      "learning_rate": 4.4126908193213856e-07,
      "loss": 0.1698,
      "step": 12153
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.323325836148321,
      "learning_rate": 4.4079338498686063e-07,
      "loss": 0.1383,
      "step": 12154
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4289252337746643,
      "learning_rate": 4.403179327616808e-07,
      "loss": 0.1432,
      "step": 12155
    },
    {
      "epoch": 0.87,
      "grad_norm": 13.01925522703635,
      "learning_rate": 4.3984272528212077e-07,
      "loss": 0.6366,
      "step": 12156
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2177147900210736,
      "learning_rate": 4.3936776257368586e-07,
      "loss": 0.1409,
      "step": 12157
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5411701783366438,
      "learning_rate": 4.3889304466187055e-07,
      "loss": 0.1927,
      "step": 12158
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2327111905230808,
      "learning_rate": 4.384185715721578e-07,
      "loss": 0.1541,
      "step": 12159
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.306575622650174,
      "learning_rate": 4.379443433300129e-07,
      "loss": 0.1449,
      "step": 12160
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.03960602568097,
      "learning_rate": 4.3747035996089206e-07,
      "loss": 0.4088,
      "step": 12161
    },
    {
      "epoch": 0.87,
      "grad_norm": 14.100764281235646,
      "learning_rate": 4.36996621490236e-07,
      "loss": 0.6425,
      "step": 12162
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.550904632861906,
      "learning_rate": 4.3652312794347284e-07,
      "loss": 0.1977,
      "step": 12163
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2751742127626762,
      "learning_rate": 4.360498793460188e-07,
      "loss": 0.1552,
      "step": 12164
    },
    {
      "epoch": 0.87,
      "grad_norm": 13.055919294893094,
      "learning_rate": 4.3557687572327644e-07,
      "loss": 0.5961,
      "step": 12165
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4806812405444445,
      "learning_rate": 4.3510411710063316e-07,
      "loss": 0.1692,
      "step": 12166
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.998049834843293,
      "learning_rate": 4.3463160350346644e-07,
      "loss": 0.4372,
      "step": 12167
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3366537001986256,
      "learning_rate": 4.341593349571366e-07,
      "loss": 0.1555,
      "step": 12168
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4205419887384076,
      "learning_rate": 4.3368731148699604e-07,
      "loss": 0.1699,
      "step": 12169
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4155081990278664,
      "learning_rate": 4.3321553311837847e-07,
      "loss": 0.1567,
      "step": 12170
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3087785368213412,
      "learning_rate": 4.327439998766081e-07,
      "loss": 0.1866,
      "step": 12171
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4719709901256772,
      "learning_rate": 4.322727117869951e-07,
      "loss": 0.1528,
      "step": 12172
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.329574533977938,
      "learning_rate": 4.3180166887483723e-07,
      "loss": 0.1544,
      "step": 12173
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.351566962012528,
      "learning_rate": 4.313308711654163e-07,
      "loss": 0.1771,
      "step": 12174
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.748832165787509,
      "learning_rate": 4.308603186840049e-07,
      "loss": 0.5493,
      "step": 12175
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.328431481581902,
      "learning_rate": 4.3039001145585843e-07,
      "loss": 0.1639,
      "step": 12176
    },
    {
      "epoch": 0.87,
      "grad_norm": 16.261083758165544,
      "learning_rate": 4.299199495062234e-07,
      "loss": 0.366,
      "step": 12177
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2546935342771806,
      "learning_rate": 4.294501328603279e-07,
      "loss": 0.1681,
      "step": 12178
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.24761627420577,
      "learning_rate": 4.2898056154339174e-07,
      "loss": 0.1658,
      "step": 12179
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4273639524174486,
      "learning_rate": 4.2851123558061927e-07,
      "loss": 0.2006,
      "step": 12180
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2367723589706219,
      "learning_rate": 4.2804215499720304e-07,
      "loss": 0.1719,
      "step": 12181
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.776615709933504,
      "learning_rate": 4.275733198183196e-07,
      "loss": 0.4312,
      "step": 12182
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1599368773771237,
      "learning_rate": 4.271047300691361e-07,
      "loss": 0.1534,
      "step": 12183
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2594040577272472,
      "learning_rate": 4.266363857748024e-07,
      "loss": 0.151,
      "step": 12184
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4163832549097195,
      "learning_rate": 4.2616828696045943e-07,
      "loss": 0.1687,
      "step": 12185
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2516124972585618,
      "learning_rate": 4.2570043365123105e-07,
      "loss": 0.1704,
      "step": 12186
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2640239092462386,
      "learning_rate": 4.2523282587223216e-07,
      "loss": 0.1232,
      "step": 12187
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.200582172097428,
      "learning_rate": 4.2476546364855875e-07,
      "loss": 0.1745,
      "step": 12188
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3052542189004455,
      "learning_rate": 4.2429834700529914e-07,
      "loss": 0.1988,
      "step": 12189
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3091665092389297,
      "learning_rate": 4.238314759675266e-07,
      "loss": 0.1619,
      "step": 12190
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3700322187550933,
      "learning_rate": 4.233648505602994e-07,
      "loss": 0.1994,
      "step": 12191
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7742948656314352,
      "learning_rate": 4.228984708086664e-07,
      "loss": 0.1447,
      "step": 12192
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2091943714488227,
      "learning_rate": 4.224323367376587e-07,
      "loss": 0.1689,
      "step": 12193
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.5959588072957525,
      "learning_rate": 4.219664483722979e-07,
      "loss": 0.738,
      "step": 12194
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1875479298262428,
      "learning_rate": 4.215008057375902e-07,
      "loss": 0.1281,
      "step": 12195
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2140445212908904,
      "learning_rate": 4.2103540885853055e-07,
      "loss": 0.1431,
      "step": 12196
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4146790789458403,
      "learning_rate": 4.2057025776009787e-07,
      "loss": 0.1467,
      "step": 12197
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3520571174870055,
      "learning_rate": 4.20105352467261e-07,
      "loss": 0.1601,
      "step": 12198
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2280093935850074,
      "learning_rate": 4.1964069300497345e-07,
      "loss": 0.1563,
      "step": 12199
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4581271619862008,
      "learning_rate": 4.1917627939817793e-07,
      "loss": 0.1659,
      "step": 12200
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.557918726414298,
      "learning_rate": 4.1871211167180015e-07,
      "loss": 0.1643,
      "step": 12201
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.251951033704364,
      "learning_rate": 4.182481898507562e-07,
      "loss": 0.158,
      "step": 12202
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3338568506650184,
      "learning_rate": 4.1778451395994677e-07,
      "loss": 0.1909,
      "step": 12203
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.175865882056589,
      "learning_rate": 4.1732108402426087e-07,
      "loss": 0.1349,
      "step": 12204
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3227403096633223,
      "learning_rate": 4.168579000685724e-07,
      "loss": 0.1537,
      "step": 12205
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4735030953649493,
      "learning_rate": 4.1639496211774434e-07,
      "loss": 0.1797,
      "step": 12206
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2156417629751475,
      "learning_rate": 4.1593227019662453e-07,
      "loss": 0.1553,
      "step": 12207
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.373289203466091,
      "learning_rate": 4.1546982433005036e-07,
      "loss": 0.1936,
      "step": 12208
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3709082679424485,
      "learning_rate": 4.1500762454284136e-07,
      "loss": 0.1421,
      "step": 12209
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.416322985715025,
      "learning_rate": 4.145456708598089e-07,
      "loss": 0.1563,
      "step": 12210
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3632375154018952,
      "learning_rate": 4.1408396330574643e-07,
      "loss": 0.1587,
      "step": 12211
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3615236367905432,
      "learning_rate": 4.136225019054391e-07,
      "loss": 0.1705,
      "step": 12212
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.323329634548248,
      "learning_rate": 4.1316128668365483e-07,
      "loss": 0.1623,
      "step": 12213
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.285195743266254,
      "learning_rate": 4.1270031766514953e-07,
      "loss": 0.4025,
      "step": 12214
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.244070687686029,
      "learning_rate": 4.1223959487466713e-07,
      "loss": 0.1362,
      "step": 12215
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.200025885286412,
      "learning_rate": 4.117791183369374e-07,
      "loss": 0.1452,
      "step": 12216
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2431502237774237,
      "learning_rate": 4.1131888807667607e-07,
      "loss": 0.1436,
      "step": 12217
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2384751176492876,
      "learning_rate": 4.1085890411858785e-07,
      "loss": 0.1558,
      "step": 12218
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3940139174592552,
      "learning_rate": 4.1039916648736133e-07,
      "loss": 0.1792,
      "step": 12219
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3904645131808844,
      "learning_rate": 4.0993967520767455e-07,
      "loss": 0.1792,
      "step": 12220
    },
    {
      "epoch": 0.87,
      "grad_norm": 12.457294363268492,
      "learning_rate": 4.094804303041899e-07,
      "loss": 0.7482,
      "step": 12221
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5941108175601084,
      "learning_rate": 4.090214318015584e-07,
      "loss": 0.204,
      "step": 12222
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3024176362045392,
      "learning_rate": 4.0856267972441797e-07,
      "loss": 0.1675,
      "step": 12223
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4440827087074304,
      "learning_rate": 4.081041740973918e-07,
      "loss": 0.1441,
      "step": 12224
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2829166128395488,
      "learning_rate": 4.0764591494509176e-07,
      "loss": 0.1669,
      "step": 12225
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2300453122373085,
      "learning_rate": 4.071879022921138e-07,
      "loss": 0.1433,
      "step": 12226
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4664215629290163,
      "learning_rate": 4.0673013616304326e-07,
      "loss": 0.1513,
      "step": 12227
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.416449360875045,
      "learning_rate": 4.062726165824504e-07,
      "loss": 0.19,
      "step": 12228
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3411980862560648,
      "learning_rate": 4.058153435748946e-07,
      "loss": 0.1809,
      "step": 12229
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.892496377199309,
      "learning_rate": 4.053583171649189e-07,
      "loss": 0.6597,
      "step": 12230
    },
    {
      "epoch": 0.87,
      "grad_norm": 11.590427617530558,
      "learning_rate": 4.0490153737705486e-07,
      "loss": 0.5039,
      "step": 12231
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.205068764583334,
      "learning_rate": 4.0444500423582057e-07,
      "loss": 0.5764,
      "step": 12232
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.331187991214813,
      "learning_rate": 4.0398871776572256e-07,
      "loss": 0.175,
      "step": 12233
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5120773480108451,
      "learning_rate": 4.035326779912502e-07,
      "loss": 0.1504,
      "step": 12234
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3599119941773794,
      "learning_rate": 4.0307688493688377e-07,
      "loss": 0.1944,
      "step": 12235
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3876240506926127,
      "learning_rate": 4.026213386270866e-07,
      "loss": 0.1893,
      "step": 12236
    },
    {
      "epoch": 0.88,
      "grad_norm": 8.298959646081002,
      "learning_rate": 4.021660390863125e-07,
      "loss": 0.5796,
      "step": 12237
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.231696624177417,
      "learning_rate": 4.017109863389984e-07,
      "loss": 0.1365,
      "step": 12238
    },
    {
      "epoch": 0.88,
      "grad_norm": 10.476024619507395,
      "learning_rate": 4.012561804095705e-07,
      "loss": 0.6054,
      "step": 12239
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3826891841009685,
      "learning_rate": 4.008016213224408e-07,
      "loss": 0.1911,
      "step": 12240
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3798521154519907,
      "learning_rate": 4.003473091020094e-07,
      "loss": 0.172,
      "step": 12241
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1508323425007327,
      "learning_rate": 3.9989324377266046e-07,
      "loss": 0.1457,
      "step": 12242
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2192145747985534,
      "learning_rate": 3.9943942535876744e-07,
      "loss": 0.157,
      "step": 12243
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1033435724988767,
      "learning_rate": 3.98985853884688e-07,
      "loss": 0.1199,
      "step": 12244
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.393878738043813,
      "learning_rate": 3.985325293747705e-07,
      "loss": 0.1764,
      "step": 12245
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4144657662132667,
      "learning_rate": 3.980794518533448e-07,
      "loss": 0.208,
      "step": 12246
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.392002203773193,
      "learning_rate": 3.9762662134473163e-07,
      "loss": 0.1839,
      "step": 12247
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2444680437920506,
      "learning_rate": 3.971740378732375e-07,
      "loss": 0.1409,
      "step": 12248
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4910329498955077,
      "learning_rate": 3.967217014631558e-07,
      "loss": 0.1797,
      "step": 12249
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6977258899730152,
      "learning_rate": 3.9626961213876433e-07,
      "loss": 0.1954,
      "step": 12250
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.676296993636367,
      "learning_rate": 3.9581776992433096e-07,
      "loss": 0.6332,
      "step": 12251
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3785641666925728,
      "learning_rate": 3.953661748441079e-07,
      "loss": 0.1758,
      "step": 12252
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2439100582036597,
      "learning_rate": 3.949148269223363e-07,
      "loss": 0.1449,
      "step": 12253
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2186158784700016,
      "learning_rate": 3.9446372618324014e-07,
      "loss": 0.13,
      "step": 12254
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.493630704725571,
      "learning_rate": 3.9401287265103504e-07,
      "loss": 0.6286,
      "step": 12255
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.647808266787209,
      "learning_rate": 3.9356226634992e-07,
      "loss": 0.5098,
      "step": 12256
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1761827044722613,
      "learning_rate": 3.9311190730408233e-07,
      "loss": 0.1508,
      "step": 12257
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4434775606417454,
      "learning_rate": 3.92661795537696e-07,
      "loss": 0.1656,
      "step": 12258
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3342629015040315,
      "learning_rate": 3.922119310749195e-07,
      "loss": 0.1635,
      "step": 12259
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3305860297153327,
      "learning_rate": 3.9176231393990183e-07,
      "loss": 0.1972,
      "step": 12260
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3772213508150388,
      "learning_rate": 3.913129441567748e-07,
      "loss": 0.1617,
      "step": 12261
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.948450420518056,
      "learning_rate": 3.908638217496602e-07,
      "loss": 0.3538,
      "step": 12262
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.262322033590518,
      "learning_rate": 3.904149467426638e-07,
      "loss": 0.1616,
      "step": 12263
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.632738746918453,
      "learning_rate": 3.8996631915988015e-07,
      "loss": 0.647,
      "step": 12264
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3390360530271688,
      "learning_rate": 3.895179390253895e-07,
      "loss": 0.1816,
      "step": 12265
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.433933385951657,
      "learning_rate": 3.8906980636326084e-07,
      "loss": 0.1772,
      "step": 12266
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4605461032137121,
      "learning_rate": 3.886219211975456e-07,
      "loss": 0.178,
      "step": 12267
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3999869830628127,
      "learning_rate": 3.8817428355228617e-07,
      "loss": 0.1605,
      "step": 12268
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3918101029623047,
      "learning_rate": 3.8772689345150885e-07,
      "loss": 0.1365,
      "step": 12269
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2761329244596857,
      "learning_rate": 3.8727975091922885e-07,
      "loss": 0.1666,
      "step": 12270
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2871996959620147,
      "learning_rate": 3.8683285597944597e-07,
      "loss": 0.1684,
      "step": 12271
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.131599000441692,
      "learning_rate": 3.8638620865614816e-07,
      "loss": 0.1772,
      "step": 12272
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2263705961535334,
      "learning_rate": 3.8593980897330964e-07,
      "loss": 0.1502,
      "step": 12273
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.5382787792814065,
      "learning_rate": 3.854936569548923e-07,
      "loss": 0.5408,
      "step": 12274
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2366509309891942,
      "learning_rate": 3.850477526248425e-07,
      "loss": 0.1883,
      "step": 12275
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4034854012144553,
      "learning_rate": 3.846020960070956e-07,
      "loss": 0.1532,
      "step": 12276
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4540425842688196,
      "learning_rate": 3.8415668712557144e-07,
      "loss": 0.2122,
      "step": 12277
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4070341664050143,
      "learning_rate": 3.837115260041796e-07,
      "loss": 0.1591,
      "step": 12278
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2806264333576889,
      "learning_rate": 3.832666126668122e-07,
      "loss": 0.1637,
      "step": 12279
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4815025974692486,
      "learning_rate": 3.8282194713735286e-07,
      "loss": 0.1688,
      "step": 12280
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5266397018111026,
      "learning_rate": 3.823775294396664e-07,
      "loss": 0.1791,
      "step": 12281
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5880528232541373,
      "learning_rate": 3.819333595976116e-07,
      "loss": 0.2031,
      "step": 12282
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.228769227001511,
      "learning_rate": 3.814894376350259e-07,
      "loss": 0.143,
      "step": 12283
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5483599195029787,
      "learning_rate": 3.8104576357573975e-07,
      "loss": 0.1913,
      "step": 12284
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5119780247374408,
      "learning_rate": 3.8060233744356634e-07,
      "loss": 0.181,
      "step": 12285
    },
    {
      "epoch": 0.88,
      "grad_norm": 11.331125935441687,
      "learning_rate": 3.8015915926230774e-07,
      "loss": 0.4936,
      "step": 12286
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3576696371095958,
      "learning_rate": 3.797162290557521e-07,
      "loss": 0.1656,
      "step": 12287
    },
    {
      "epoch": 0.88,
      "grad_norm": 9.668643520280467,
      "learning_rate": 3.7927354684767323e-07,
      "loss": 0.4131,
      "step": 12288
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4170356259058676,
      "learning_rate": 3.7883111266183435e-07,
      "loss": 0.1774,
      "step": 12289
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.103271131856995,
      "learning_rate": 3.7838892652198036e-07,
      "loss": 0.568,
      "step": 12290
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2053350626947632,
      "learning_rate": 3.7794698845184994e-07,
      "loss": 0.1617,
      "step": 12291
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3965087454268106,
      "learning_rate": 3.77505298475162e-07,
      "loss": 0.1764,
      "step": 12292
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2970032203006339,
      "learning_rate": 3.7706385661562596e-07,
      "loss": 0.1693,
      "step": 12293
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4355161302530306,
      "learning_rate": 3.766226628969355e-07,
      "loss": 0.1895,
      "step": 12294
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.442672401575189,
      "learning_rate": 3.7618171734277396e-07,
      "loss": 0.1612,
      "step": 12295
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5211361710003917,
      "learning_rate": 3.757410199768069e-07,
      "loss": 0.2211,
      "step": 12296
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1370303397238015,
      "learning_rate": 3.75300570822692e-07,
      "loss": 0.137,
      "step": 12297
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3717828731318171,
      "learning_rate": 3.7486036990406873e-07,
      "loss": 0.152,
      "step": 12298
    },
    {
      "epoch": 0.88,
      "grad_norm": 9.13957036649532,
      "learning_rate": 3.744204172445653e-07,
      "loss": 0.5733,
      "step": 12299
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3861206028774824,
      "learning_rate": 3.739807128677986e-07,
      "loss": 0.1615,
      "step": 12300
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3022870386754126,
      "learning_rate": 3.7354125679736897e-07,
      "loss": 0.157,
      "step": 12301
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1157176515128262,
      "learning_rate": 3.7310204905686433e-07,
      "loss": 0.1351,
      "step": 12302
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3918472910864714,
      "learning_rate": 3.7266308966986074e-07,
      "loss": 0.1835,
      "step": 12303
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4909868167616538,
      "learning_rate": 3.722243786599178e-07,
      "loss": 0.1998,
      "step": 12304
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.280873808077874,
      "learning_rate": 3.717859160505866e-07,
      "loss": 0.1512,
      "step": 12305
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.418460877360948,
      "learning_rate": 3.7134770186539935e-07,
      "loss": 0.1757,
      "step": 12306
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.9956190497297035,
      "learning_rate": 3.709097361278785e-07,
      "loss": 0.5067,
      "step": 12307
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.276270539250999,
      "learning_rate": 3.7047201886153296e-07,
      "loss": 0.1484,
      "step": 12308
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.73469025182058,
      "learning_rate": 3.700345500898583e-07,
      "loss": 0.4954,
      "step": 12309
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2986115732855688,
      "learning_rate": 3.695973298363337e-07,
      "loss": 0.1285,
      "step": 12310
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.263406964778152,
      "learning_rate": 3.691603581244302e-07,
      "loss": 0.5711,
      "step": 12311
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9668219167265268,
      "learning_rate": 3.687236349776002e-07,
      "loss": 0.1906,
      "step": 12312
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3439456633990847,
      "learning_rate": 3.6828716041928723e-07,
      "loss": 0.1873,
      "step": 12313
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2352730566493975,
      "learning_rate": 3.678509344729181e-07,
      "loss": 0.1593,
      "step": 12314
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2483539040552933,
      "learning_rate": 3.674149571619079e-07,
      "loss": 0.1687,
      "step": 12315
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.635784358384336,
      "learning_rate": 3.6697922850965906e-07,
      "loss": 0.5316,
      "step": 12316
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3312229282458148,
      "learning_rate": 3.6654374853955844e-07,
      "loss": 0.1733,
      "step": 12317
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4175391552464351,
      "learning_rate": 3.6610851727498343e-07,
      "loss": 0.1654,
      "step": 12318
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3066140238651576,
      "learning_rate": 3.65673534739292e-07,
      "loss": 0.141,
      "step": 12319
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0928493779892046,
      "learning_rate": 3.6523880095583554e-07,
      "loss": 0.1452,
      "step": 12320
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.358086193285075,
      "learning_rate": 3.648043159479459e-07,
      "loss": 0.1878,
      "step": 12321
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.373036638653411,
      "learning_rate": 3.643700797389471e-07,
      "loss": 0.1705,
      "step": 12322
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.294043813270681,
      "learning_rate": 3.6393609235214513e-07,
      "loss": 0.2,
      "step": 12323
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3459247005783943,
      "learning_rate": 3.6350235381083563e-07,
      "loss": 0.1836,
      "step": 12324
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4733115237766263,
      "learning_rate": 3.630688641383001e-07,
      "loss": 0.1675,
      "step": 12325
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2128790661257387,
      "learning_rate": 3.626356233578071e-07,
      "loss": 0.1361,
      "step": 12326
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.49515009735866,
      "learning_rate": 3.622026314926097e-07,
      "loss": 0.1649,
      "step": 12327
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3636222685705321,
      "learning_rate": 3.6176988856595095e-07,
      "loss": 0.1702,
      "step": 12328
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.348972954270638,
      "learning_rate": 3.6133739460105675e-07,
      "loss": 0.1772,
      "step": 12329
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2885129841539755,
      "learning_rate": 3.609051496211441e-07,
      "loss": 0.1929,
      "step": 12330
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4773227816967902,
      "learning_rate": 3.604731536494116e-07,
      "loss": 0.2028,
      "step": 12331
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4681410774497143,
      "learning_rate": 3.6004140670904853e-07,
      "loss": 0.1661,
      "step": 12332
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2093552741892046,
      "learning_rate": 3.596099088232291e-07,
      "loss": 0.1414,
      "step": 12333
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.459888258865435,
      "learning_rate": 3.591786600151154e-07,
      "loss": 0.1731,
      "step": 12334
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.47142068635466,
      "learning_rate": 3.5874766030785327e-07,
      "loss": 0.1858,
      "step": 12335
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.239522511716702,
      "learning_rate": 3.583169097245792e-07,
      "loss": 0.7484,
      "step": 12336
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.539485995845976,
      "learning_rate": 3.5788640828841205e-07,
      "loss": 0.5582,
      "step": 12337
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.295000261721078,
      "learning_rate": 3.57456156022461e-07,
      "loss": 0.1604,
      "step": 12338
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2865617902982753,
      "learning_rate": 3.570261529498187e-07,
      "loss": 0.1805,
      "step": 12339
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2564835669419616,
      "learning_rate": 3.5659639909356725e-07,
      "loss": 0.1433,
      "step": 12340
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.388334398910148,
      "learning_rate": 3.561668944767738e-07,
      "loss": 0.1787,
      "step": 12341
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.196559703772662,
      "learning_rate": 3.5573763912249313e-07,
      "loss": 0.1585,
      "step": 12342
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.342659361808913,
      "learning_rate": 3.553086330537647e-07,
      "loss": 0.5447,
      "step": 12343
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3256640899410945,
      "learning_rate": 3.5487987629361676e-07,
      "loss": 0.1379,
      "step": 12344
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3964545508681,
      "learning_rate": 3.544513688650625e-07,
      "loss": 0.1668,
      "step": 12345
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1967479338941163,
      "learning_rate": 3.540231107911041e-07,
      "loss": 0.1419,
      "step": 12346
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2371701858481159,
      "learning_rate": 3.5359510209472646e-07,
      "loss": 0.1421,
      "step": 12347
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3382407833683225,
      "learning_rate": 3.531673427989046e-07,
      "loss": 0.1747,
      "step": 12348
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5690017815328545,
      "learning_rate": 3.527398329265991e-07,
      "loss": 0.24,
      "step": 12349
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2353217629407596,
      "learning_rate": 3.523125725007565e-07,
      "loss": 0.1932,
      "step": 12350
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5311822924626806,
      "learning_rate": 3.5188556154431174e-07,
      "loss": 0.1802,
      "step": 12351
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2659777934358274,
      "learning_rate": 3.514588000801833e-07,
      "loss": 0.1538,
      "step": 12352
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2078433176260877,
      "learning_rate": 3.5103228813127944e-07,
      "loss": 0.1584,
      "step": 12353
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2115604846805677,
      "learning_rate": 3.506060257204924e-07,
      "loss": 0.1332,
      "step": 12354
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3624529911697794,
      "learning_rate": 3.5018001287070336e-07,
      "loss": 0.1769,
      "step": 12355
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.273772827219816,
      "learning_rate": 3.497542496047779e-07,
      "loss": 0.1443,
      "step": 12356
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4465989168929785,
      "learning_rate": 3.493287359455694e-07,
      "loss": 0.233,
      "step": 12357
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.912982676040247,
      "learning_rate": 3.4890347191591843e-07,
      "loss": 0.5574,
      "step": 12358
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4832753212448433,
      "learning_rate": 3.484784575386524e-07,
      "loss": 0.1867,
      "step": 12359
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4330048010668202,
      "learning_rate": 3.480536928365824e-07,
      "loss": 0.1871,
      "step": 12360
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3703808225425294,
      "learning_rate": 3.476291778325097e-07,
      "loss": 0.1639,
      "step": 12361
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2772496372971203,
      "learning_rate": 3.472049125492188e-07,
      "loss": 0.1357,
      "step": 12362
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.149637745702199,
      "learning_rate": 3.4678089700948434e-07,
      "loss": 0.1473,
      "step": 12363
    },
    {
      "epoch": 0.88,
      "grad_norm": 13.172816609200712,
      "learning_rate": 3.4635713123606475e-07,
      "loss": 0.5569,
      "step": 12364
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.9600074701916546,
      "learning_rate": 3.4593361525170623e-07,
      "loss": 0.5062,
      "step": 12365
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1533178814690603,
      "learning_rate": 3.455103490791417e-07,
      "loss": 0.1259,
      "step": 12366
    },
    {
      "epoch": 0.88,
      "grad_norm": 9.597160896139815,
      "learning_rate": 3.4508733274109087e-07,
      "loss": 0.6245,
      "step": 12367
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2706524349221564,
      "learning_rate": 3.4466456626025824e-07,
      "loss": 0.162,
      "step": 12368
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4204386749026654,
      "learning_rate": 3.4424204965933795e-07,
      "loss": 0.1549,
      "step": 12369
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4020773809171883,
      "learning_rate": 3.438197829610074e-07,
      "loss": 0.1515,
      "step": 12370
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3773032257163889,
      "learning_rate": 3.4339776618793344e-07,
      "loss": 0.1732,
      "step": 12371
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.323601648147998,
      "learning_rate": 3.429759993627674e-07,
      "loss": 0.1584,
      "step": 12372
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3434001656029488,
      "learning_rate": 3.4255448250814784e-07,
      "loss": 0.1811,
      "step": 12373
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.051427180575806,
      "learning_rate": 3.421332156467005e-07,
      "loss": 0.6618,
      "step": 12374
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2132310916298552,
      "learning_rate": 3.4171219880103857e-07,
      "loss": 0.1462,
      "step": 12375
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3358766688809016,
      "learning_rate": 3.4129143199375823e-07,
      "loss": 0.1476,
      "step": 12376
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4334694027367691,
      "learning_rate": 3.4087091524744706e-07,
      "loss": 0.1986,
      "step": 12377
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.477152860097439,
      "learning_rate": 3.4045064858467426e-07,
      "loss": 0.2167,
      "step": 12378
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1574427847598912,
      "learning_rate": 3.40030632028e-07,
      "loss": 0.1471,
      "step": 12379
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1979415157825948,
      "learning_rate": 3.39610865599968e-07,
      "loss": 0.1523,
      "step": 12380
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3489180679672341,
      "learning_rate": 3.391913493231097e-07,
      "loss": 0.1605,
      "step": 12381
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.203006222351786,
      "learning_rate": 3.3877208321994316e-07,
      "loss": 0.4289,
      "step": 12382
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.348096782614331,
      "learning_rate": 3.3835306731297313e-07,
      "loss": 0.1652,
      "step": 12383
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.765061639221064,
      "learning_rate": 3.3793430162469166e-07,
      "loss": 0.5733,
      "step": 12384
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2827393879554507,
      "learning_rate": 3.375157861775746e-07,
      "loss": 0.1691,
      "step": 12385
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5540421573430727,
      "learning_rate": 3.37097520994088e-07,
      "loss": 0.166,
      "step": 12386
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4090949388514569,
      "learning_rate": 3.36679506096681e-07,
      "loss": 0.1397,
      "step": 12387
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2611459387350195,
      "learning_rate": 3.362617415077923e-07,
      "loss": 0.1515,
      "step": 12388
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3037005367631884,
      "learning_rate": 3.3584422724984465e-07,
      "loss": 0.167,
      "step": 12389
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3567670838458108,
      "learning_rate": 3.3542696334525e-07,
      "loss": 0.1662,
      "step": 12390
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.439067013285684,
      "learning_rate": 3.3500994981640387e-07,
      "loss": 0.1825,
      "step": 12391
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4258390338582885,
      "learning_rate": 3.3459318668569053e-07,
      "loss": 0.214,
      "step": 12392
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3264378498860108,
      "learning_rate": 3.341766739754798e-07,
      "loss": 0.1536,
      "step": 12393
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2356530717824874,
      "learning_rate": 3.3376041170813003e-07,
      "loss": 0.1426,
      "step": 12394
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.479249458709933,
      "learning_rate": 3.3334439990598264e-07,
      "loss": 0.2046,
      "step": 12395
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3094962782956234,
      "learning_rate": 3.329286385913688e-07,
      "loss": 0.1829,
      "step": 12396
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.067125408659858,
      "learning_rate": 3.3251312778660394e-07,
      "loss": 0.1222,
      "step": 12397
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2422038297816622,
      "learning_rate": 3.320978675139919e-07,
      "loss": 0.1768,
      "step": 12398
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3766425132407423,
      "learning_rate": 3.3168285779582144e-07,
      "loss": 0.2059,
      "step": 12399
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.75046398306239,
      "learning_rate": 3.3126809865436817e-07,
      "loss": 0.5133,
      "step": 12400
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.367680696745612,
      "learning_rate": 3.308535901118959e-07,
      "loss": 0.2001,
      "step": 12401
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3004700846865003,
      "learning_rate": 3.3043933219065404e-07,
      "loss": 0.1758,
      "step": 12402
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.54481663369894,
      "learning_rate": 3.300253249128771e-07,
      "loss": 0.1982,
      "step": 12403
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3224219478602848,
      "learning_rate": 3.2961156830078836e-07,
      "loss": 0.1884,
      "step": 12404
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2267261418082813,
      "learning_rate": 3.291980623765956e-07,
      "loss": 0.1285,
      "step": 12405
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3056063682518488,
      "learning_rate": 3.287848071624955e-07,
      "loss": 0.1837,
      "step": 12406
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3657374754509184,
      "learning_rate": 3.283718026806687e-07,
      "loss": 0.2021,
      "step": 12407
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3441918433307374,
      "learning_rate": 3.2795904895328356e-07,
      "loss": 0.1691,
      "step": 12408
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.59482928195363,
      "learning_rate": 3.275465460024957e-07,
      "loss": 0.5248,
      "step": 12409
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.335134705645474,
      "learning_rate": 3.271342938504479e-07,
      "loss": 0.1436,
      "step": 12410
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3406152387376735,
      "learning_rate": 3.267222925192659e-07,
      "loss": 0.189,
      "step": 12411
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.413920952316013,
      "learning_rate": 3.263105420310658e-07,
      "loss": 0.1599,
      "step": 12412
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3959980563531655,
      "learning_rate": 3.258990424079472e-07,
      "loss": 0.1451,
      "step": 12413
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2718813359317076,
      "learning_rate": 3.2548779367199965e-07,
      "loss": 0.1419,
      "step": 12414
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3769974828851124,
      "learning_rate": 3.250767958452966e-07,
      "loss": 0.1918,
      "step": 12415
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1817027545956562,
      "learning_rate": 3.246660489498982e-07,
      "loss": 0.1452,
      "step": 12416
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3086829248549918,
      "learning_rate": 3.242555530078517e-07,
      "loss": 0.1847,
      "step": 12417
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2626044260275835,
      "learning_rate": 3.2384530804119185e-07,
      "loss": 0.1683,
      "step": 12418
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5208098424671461,
      "learning_rate": 3.234353140719393e-07,
      "loss": 0.2158,
      "step": 12419
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.6812279838177036,
      "learning_rate": 3.230255711220992e-07,
      "loss": 0.6642,
      "step": 12420
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.848773243185552,
      "learning_rate": 3.226160792136662e-07,
      "loss": 0.624,
      "step": 12421
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4532321794848184,
      "learning_rate": 3.222068383686194e-07,
      "loss": 0.2123,
      "step": 12422
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3661002088818466,
      "learning_rate": 3.217978486089263e-07,
      "loss": 0.1875,
      "step": 12423
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4092594637819513,
      "learning_rate": 3.2138910995653863e-07,
      "loss": 0.2075,
      "step": 12424
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2910739822320736,
      "learning_rate": 3.209806224333961e-07,
      "loss": 0.1241,
      "step": 12425
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3783685604183196,
      "learning_rate": 3.205723860614252e-07,
      "loss": 0.1776,
      "step": 12426
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0438665324735945,
      "learning_rate": 3.2016440086253873e-07,
      "loss": 0.117,
      "step": 12427
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.195459556146762,
      "learning_rate": 3.1975666685863494e-07,
      "loss": 0.1312,
      "step": 12428
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3882636245512612,
      "learning_rate": 3.193491840716001e-07,
      "loss": 0.1765,
      "step": 12429
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.364730778036446,
      "learning_rate": 3.1894195252330506e-07,
      "loss": 0.1315,
      "step": 12430
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4064565023172253,
      "learning_rate": 3.185349722356101e-07,
      "loss": 0.1445,
      "step": 12431
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.340913204651421,
      "learning_rate": 3.1812824323035843e-07,
      "loss": 0.1634,
      "step": 12432
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.3619176645894555,
      "learning_rate": 3.1772176552938237e-07,
      "loss": 0.5294,
      "step": 12433
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3168796741115623,
      "learning_rate": 3.1731553915450077e-07,
      "loss": 0.1465,
      "step": 12434
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3897909799715649,
      "learning_rate": 3.1690956412751774e-07,
      "loss": 0.1595,
      "step": 12435
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3792811248485628,
      "learning_rate": 3.165038404702242e-07,
      "loss": 0.151,
      "step": 12436
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3494970740900445,
      "learning_rate": 3.160983682043983e-07,
      "loss": 0.1913,
      "step": 12437
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3835658274110567,
      "learning_rate": 3.156931473518038e-07,
      "loss": 0.1576,
      "step": 12438
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.6748384678749755,
      "learning_rate": 3.1528817793419154e-07,
      "loss": 0.4537,
      "step": 12439
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3404998653831366,
      "learning_rate": 3.1488345997329806e-07,
      "loss": 0.1405,
      "step": 12440
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.401175191074733,
      "learning_rate": 3.144789934908471e-07,
      "loss": 0.5149,
      "step": 12441
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3316726052927388,
      "learning_rate": 3.140747785085496e-07,
      "loss": 0.1837,
      "step": 12442
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4888731829927793,
      "learning_rate": 3.136708150481022e-07,
      "loss": 0.2071,
      "step": 12443
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4279632262611879,
      "learning_rate": 3.132671031311868e-07,
      "loss": 0.1426,
      "step": 12444
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4498420793390037,
      "learning_rate": 3.12863642779474e-07,
      "loss": 0.1647,
      "step": 12445
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3946268108917923,
      "learning_rate": 3.124604340146209e-07,
      "loss": 0.1826,
      "step": 12446
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1673593961733186,
      "learning_rate": 3.120574768582685e-07,
      "loss": 0.1214,
      "step": 12447
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3578403325048316,
      "learning_rate": 3.1165477133204667e-07,
      "loss": 0.1721,
      "step": 12448
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.563163070820605,
      "learning_rate": 3.1125231745757045e-07,
      "loss": 0.1877,
      "step": 12449
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.377470444069363,
      "learning_rate": 3.108501152564425e-07,
      "loss": 0.1571,
      "step": 12450
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.279067843436056,
      "learning_rate": 3.1044816475025106e-07,
      "loss": 0.194,
      "step": 12451
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6144859106476261,
      "learning_rate": 3.100464659605729e-07,
      "loss": 0.2168,
      "step": 12452
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.253188148166966,
      "learning_rate": 3.096450189089667e-07,
      "loss": 0.1624,
      "step": 12453
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.488688662069539,
      "learning_rate": 3.092438236169837e-07,
      "loss": 0.1603,
      "step": 12454
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3393158505834106,
      "learning_rate": 3.088428801061555e-07,
      "loss": 0.1395,
      "step": 12455
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2418327503431368,
      "learning_rate": 3.0844218839800545e-07,
      "loss": 0.1383,
      "step": 12456
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.194457668720862,
      "learning_rate": 3.0804174851403914e-07,
      "loss": 0.1461,
      "step": 12457
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.920598916824305,
      "learning_rate": 3.076415604757521e-07,
      "loss": 0.6875,
      "step": 12458
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4053592451727894,
      "learning_rate": 3.072416243046239e-07,
      "loss": 0.1586,
      "step": 12459
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2897200257329038,
      "learning_rate": 3.0684194002212287e-07,
      "loss": 0.1695,
      "step": 12460
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.7529251989028,
      "learning_rate": 3.064425076497007e-07,
      "loss": 0.5261,
      "step": 12461
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.433009955482248,
      "learning_rate": 3.0604332720879924e-07,
      "loss": 0.1553,
      "step": 12462
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4001146457754277,
      "learning_rate": 3.0564439872084293e-07,
      "loss": 0.1791,
      "step": 12463
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.142190149933078,
      "learning_rate": 3.052457222072469e-07,
      "loss": 0.4043,
      "step": 12464
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3443850312846948,
      "learning_rate": 3.048472976894079e-07,
      "loss": 0.1638,
      "step": 12465
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2442280654811086,
      "learning_rate": 3.044491251887133e-07,
      "loss": 0.1642,
      "step": 12466
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3138916736364092,
      "learning_rate": 3.0405120472653483e-07,
      "loss": 0.1702,
      "step": 12467
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2591893810302524,
      "learning_rate": 3.036535363242327e-07,
      "loss": 0.1768,
      "step": 12468
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2499275416547715,
      "learning_rate": 3.0325612000315084e-07,
      "loss": 0.159,
      "step": 12469
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3831302487444126,
      "learning_rate": 3.028589557846218e-07,
      "loss": 0.1692,
      "step": 12470
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4390109897537258,
      "learning_rate": 3.0246204368996225e-07,
      "loss": 0.1711,
      "step": 12471
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.304005878619856,
      "learning_rate": 3.0206538374047857e-07,
      "loss": 0.1645,
      "step": 12472
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2519095276043892,
      "learning_rate": 3.016689759574604e-07,
      "loss": 0.1325,
      "step": 12473
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3715711453842334,
      "learning_rate": 3.012728203621862e-07,
      "loss": 0.1935,
      "step": 12474
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.263307314409951,
      "learning_rate": 3.0087691697592014e-07,
      "loss": 0.1607,
      "step": 12475
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3312634813017985,
      "learning_rate": 3.0048126581991353e-07,
      "loss": 0.1322,
      "step": 12476
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.261362369410002,
      "learning_rate": 3.000858669154011e-07,
      "loss": 0.172,
      "step": 12477
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3611724398878218,
      "learning_rate": 2.9969072028360804e-07,
      "loss": 0.1351,
      "step": 12478
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5007552151191472,
      "learning_rate": 2.9929582594574403e-07,
      "loss": 0.1987,
      "step": 12479
    },
    {
      "epoch": 0.89,
      "grad_norm": 9.559493919121897,
      "learning_rate": 2.9890118392300493e-07,
      "loss": 0.4451,
      "step": 12480
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3368832252066805,
      "learning_rate": 2.985067942365738e-07,
      "loss": 0.1782,
      "step": 12481
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.32638849344922,
      "learning_rate": 2.981126569076198e-07,
      "loss": 0.1814,
      "step": 12482
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.837124975944917,
      "learning_rate": 2.9771877195729883e-07,
      "loss": 0.5101,
      "step": 12483
    },
    {
      "epoch": 0.89,
      "grad_norm": 13.22147042289944,
      "learning_rate": 2.9732513940675276e-07,
      "loss": 0.5297,
      "step": 12484
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1666480087203432,
      "learning_rate": 2.9693175927711093e-07,
      "loss": 0.1313,
      "step": 12485
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.684537497491347,
      "learning_rate": 2.96538631589488e-07,
      "loss": 0.6459,
      "step": 12486
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4085592456680074,
      "learning_rate": 2.9614575636498557e-07,
      "loss": 0.1806,
      "step": 12487
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3629126695917761,
      "learning_rate": 2.9575313362469105e-07,
      "loss": 0.1523,
      "step": 12488
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5276165162006734,
      "learning_rate": 2.9536076338967987e-07,
      "loss": 0.2187,
      "step": 12489
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2869717020421376,
      "learning_rate": 2.949686456810119e-07,
      "loss": 0.1621,
      "step": 12490
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.209368106952642,
      "learning_rate": 2.945767805197353e-07,
      "loss": 0.5035,
      "step": 12491
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3687655137467514,
      "learning_rate": 2.941851679268831e-07,
      "loss": 0.1702,
      "step": 12492
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2776005894333442,
      "learning_rate": 2.937938079234753e-07,
      "loss": 0.1747,
      "step": 12493
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3555054933350938,
      "learning_rate": 2.934027005305201e-07,
      "loss": 0.1554,
      "step": 12494
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5697083387951232,
      "learning_rate": 2.9301184576900954e-07,
      "loss": 0.2058,
      "step": 12495
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3318681327793407,
      "learning_rate": 2.9262124365992294e-07,
      "loss": 0.1759,
      "step": 12496
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5419473085227686,
      "learning_rate": 2.922308942242269e-07,
      "loss": 0.1948,
      "step": 12497
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5323509408420828,
      "learning_rate": 2.9184079748287297e-07,
      "loss": 0.1516,
      "step": 12498
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.393141092397405,
      "learning_rate": 2.9145095345680105e-07,
      "loss": 0.1776,
      "step": 12499
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.193024294597759,
      "learning_rate": 2.910613621669356e-07,
      "loss": 0.5073,
      "step": 12500
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3108665252595617,
      "learning_rate": 2.906720236341881e-07,
      "loss": 0.1403,
      "step": 12501
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1726288023083062,
      "learning_rate": 2.902829378794575e-07,
      "loss": 0.1391,
      "step": 12502
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3738962089123448,
      "learning_rate": 2.898941049236287e-07,
      "loss": 0.1967,
      "step": 12503
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.5844703246899,
      "learning_rate": 2.895055247875717e-07,
      "loss": 0.6976,
      "step": 12504
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2556502697049867,
      "learning_rate": 2.891171974921447e-07,
      "loss": 0.1452,
      "step": 12505
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.301032256629275,
      "learning_rate": 2.887291230581901e-07,
      "loss": 0.2031,
      "step": 12506
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4341399008956541,
      "learning_rate": 2.883413015065406e-07,
      "loss": 0.1787,
      "step": 12507
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3590387892800633,
      "learning_rate": 2.8795373285801067e-07,
      "loss": 0.1708,
      "step": 12508
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3139818584917724,
      "learning_rate": 2.875664171334047e-07,
      "loss": 0.1835,
      "step": 12509
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3777335355500027,
      "learning_rate": 2.8717935435351174e-07,
      "loss": 0.1949,
      "step": 12510
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3364779053597309,
      "learning_rate": 2.867925445391079e-07,
      "loss": 0.1696,
      "step": 12511
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6997505026451962,
      "learning_rate": 2.864059877109565e-07,
      "loss": 0.1937,
      "step": 12512
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4251723104736207,
      "learning_rate": 2.860196838898044e-07,
      "loss": 0.1785,
      "step": 12513
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.336513246438921,
      "learning_rate": 2.856336330963894e-07,
      "loss": 0.1632,
      "step": 12514
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1555101125920735,
      "learning_rate": 2.85247835351431e-07,
      "loss": 0.1618,
      "step": 12515
    },
    {
      "epoch": 0.9,
      "grad_norm": 10.964823186769275,
      "learning_rate": 2.848622906756382e-07,
      "loss": 0.509,
      "step": 12516
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3465780416192414,
      "learning_rate": 2.8447699908970506e-07,
      "loss": 0.203,
      "step": 12517
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.252086372121904,
      "learning_rate": 2.840919606143133e-07,
      "loss": 0.1646,
      "step": 12518
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.307387248344108,
      "learning_rate": 2.8370717527012925e-07,
      "loss": 0.1521,
      "step": 12519
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4588793145150496,
      "learning_rate": 2.83322643077808e-07,
      "loss": 0.1685,
      "step": 12520
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4584636313246884,
      "learning_rate": 2.8293836405798805e-07,
      "loss": 0.2085,
      "step": 12521
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2486457371904838,
      "learning_rate": 2.8255433823129787e-07,
      "loss": 0.1453,
      "step": 12522
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.042719641988343,
      "learning_rate": 2.821705656183482e-07,
      "loss": 0.4748,
      "step": 12523
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.338185186893009,
      "learning_rate": 2.8178704623974094e-07,
      "loss": 0.1717,
      "step": 12524
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4081695583690395,
      "learning_rate": 2.8140378011605904e-07,
      "loss": 0.1796,
      "step": 12525
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.227685634090469,
      "learning_rate": 2.810207672678772e-07,
      "loss": 0.1512,
      "step": 12526
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4557141248542993,
      "learning_rate": 2.8063800771575224e-07,
      "loss": 0.1511,
      "step": 12527
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.321647147921995,
      "learning_rate": 2.802555014802311e-07,
      "loss": 0.1591,
      "step": 12528
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3627614152594545,
      "learning_rate": 2.798732485818434e-07,
      "loss": 0.1591,
      "step": 12529
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3983385210594073,
      "learning_rate": 2.7949124904110836e-07,
      "loss": 0.1863,
      "step": 12530
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5056646731198842,
      "learning_rate": 2.7910950287852903e-07,
      "loss": 0.164,
      "step": 12531
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3238623817356592,
      "learning_rate": 2.787280101145967e-07,
      "loss": 0.2073,
      "step": 12532
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2687150559654465,
      "learning_rate": 2.7834677076978787e-07,
      "loss": 0.1814,
      "step": 12533
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.272581854399197,
      "learning_rate": 2.779657848645656e-07,
      "loss": 0.1471,
      "step": 12534
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.171771105732968,
      "learning_rate": 2.775850524193807e-07,
      "loss": 0.1523,
      "step": 12535
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4164942157037956,
      "learning_rate": 2.772045734546702e-07,
      "loss": 0.1686,
      "step": 12536
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.339114043030958,
      "learning_rate": 2.7682434799085444e-07,
      "loss": 0.1356,
      "step": 12537
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4290427416069456,
      "learning_rate": 2.764443760483443e-07,
      "loss": 0.1466,
      "step": 12538
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2918517412285264,
      "learning_rate": 2.7606465764753345e-07,
      "loss": 0.1673,
      "step": 12539
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2736457837521273,
      "learning_rate": 2.756851928088056e-07,
      "loss": 0.1137,
      "step": 12540
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3602898770286478,
      "learning_rate": 2.7530598155252666e-07,
      "loss": 0.187,
      "step": 12541
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4249664016452768,
      "learning_rate": 2.7492702389905256e-07,
      "loss": 0.1952,
      "step": 12542
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3048012602183272,
      "learning_rate": 2.745483198687243e-07,
      "loss": 0.173,
      "step": 12543
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3374903248710193,
      "learning_rate": 2.7416986948186897e-07,
      "loss": 0.1781,
      "step": 12544
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.341796236212721,
      "learning_rate": 2.737916727588014e-07,
      "loss": 0.4947,
      "step": 12545
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4704387419518583,
      "learning_rate": 2.7341372971981917e-07,
      "loss": 0.2267,
      "step": 12546
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.364986000832111,
      "learning_rate": 2.730360403852117e-07,
      "loss": 0.5715,
      "step": 12547
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2590148768463814,
      "learning_rate": 2.7265860477524884e-07,
      "loss": 0.1869,
      "step": 12548
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.184235382412954,
      "learning_rate": 2.722814229101928e-07,
      "loss": 0.1426,
      "step": 12549
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4046107910298908,
      "learning_rate": 2.719044948102867e-07,
      "loss": 0.16,
      "step": 12550
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3447899315524046,
      "learning_rate": 2.7152782049576333e-07,
      "loss": 0.1569,
      "step": 12551
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4537091126228423,
      "learning_rate": 2.711513999868415e-07,
      "loss": 0.1751,
      "step": 12552
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.289879925660578,
      "learning_rate": 2.707752333037267e-07,
      "loss": 0.1703,
      "step": 12553
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3761626848003403,
      "learning_rate": 2.7039932046660833e-07,
      "loss": 0.1559,
      "step": 12554
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.229135617513562,
      "learning_rate": 2.700236614956653e-07,
      "loss": 0.1694,
      "step": 12555
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.355308492082857,
      "learning_rate": 2.696482564110603e-07,
      "loss": 0.1715,
      "step": 12556
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4285438957038499,
      "learning_rate": 2.6927310523294503e-07,
      "loss": 0.1789,
      "step": 12557
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.021920864952697,
      "learning_rate": 2.688982079814545e-07,
      "loss": 0.4165,
      "step": 12558
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.246655775470798,
      "learning_rate": 2.685235646767126e-07,
      "loss": 0.1437,
      "step": 12559
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8937377906460031,
      "learning_rate": 2.681491753388282e-07,
      "loss": 0.2287,
      "step": 12560
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.363767789495913,
      "learning_rate": 2.677750399878987e-07,
      "loss": 0.1924,
      "step": 12561
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3829496459611403,
      "learning_rate": 2.6740115864400396e-07,
      "loss": 0.1996,
      "step": 12562
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3203493974584029,
      "learning_rate": 2.6702753132721414e-07,
      "loss": 0.1776,
      "step": 12563
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1857892031064974,
      "learning_rate": 2.6665415805758264e-07,
      "loss": 0.1285,
      "step": 12564
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3542116209574886,
      "learning_rate": 2.662810388551518e-07,
      "loss": 0.2093,
      "step": 12565
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.17659787964073,
      "learning_rate": 2.6590817373994826e-07,
      "loss": 0.1233,
      "step": 12566
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.438280177912832,
      "learning_rate": 2.655355627319861e-07,
      "loss": 0.1243,
      "step": 12567
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3023538663901006,
      "learning_rate": 2.6516320585126653e-07,
      "loss": 0.188,
      "step": 12568
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.238514189539441,
      "learning_rate": 2.6479110311777576e-07,
      "loss": 0.1545,
      "step": 12569
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.411360869702086,
      "learning_rate": 2.644192545514862e-07,
      "loss": 0.1769,
      "step": 12570
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5939163650426575,
      "learning_rate": 2.6404766017235794e-07,
      "loss": 0.1798,
      "step": 12571
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2626736258315665,
      "learning_rate": 2.636763200003356e-07,
      "loss": 0.1879,
      "step": 12572
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2500473212776095,
      "learning_rate": 2.6330523405535267e-07,
      "loss": 0.1413,
      "step": 12573
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2682458474736469,
      "learning_rate": 2.6293440235732717e-07,
      "loss": 0.1551,
      "step": 12574
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2972426190502278,
      "learning_rate": 2.625638249261625e-07,
      "loss": 0.1674,
      "step": 12575
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3334537112900413,
      "learning_rate": 2.621935017817512e-07,
      "loss": 0.1661,
      "step": 12576
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3876207690264928,
      "learning_rate": 2.6182343294397063e-07,
      "loss": 0.2015,
      "step": 12577
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3558097484135483,
      "learning_rate": 2.614536184326849e-07,
      "loss": 0.1615,
      "step": 12578
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3438174049366505,
      "learning_rate": 2.610840582677426e-07,
      "loss": 0.1822,
      "step": 12579
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3193474495388444,
      "learning_rate": 2.607147524689829e-07,
      "loss": 0.1766,
      "step": 12580
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2161152651702318,
      "learning_rate": 2.6034570105622604e-07,
      "loss": 0.1247,
      "step": 12581
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2819576709581733,
      "learning_rate": 2.599769040492828e-07,
      "loss": 0.1763,
      "step": 12582
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.361645412362775,
      "learning_rate": 2.5960836146794734e-07,
      "loss": 0.1604,
      "step": 12583
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.376547176144151,
      "learning_rate": 2.5924007333200387e-07,
      "loss": 0.1495,
      "step": 12584
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2778420966634354,
      "learning_rate": 2.588720396612171e-07,
      "loss": 0.1741,
      "step": 12585
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4570081720208612,
      "learning_rate": 2.585042604753457e-07,
      "loss": 0.1673,
      "step": 12586
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3521435576405392,
      "learning_rate": 2.5813673579412724e-07,
      "loss": 0.1996,
      "step": 12587
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.416755350975266,
      "learning_rate": 2.57769465637292e-07,
      "loss": 0.1778,
      "step": 12588
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2967252316492108,
      "learning_rate": 2.574024500245503e-07,
      "loss": 0.1435,
      "step": 12589
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.152672295231197,
      "learning_rate": 2.5703568897560535e-07,
      "loss": 0.4505,
      "step": 12590
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4130034582149782,
      "learning_rate": 2.566691825101403e-07,
      "loss": 0.196,
      "step": 12591
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.316328585841141,
      "learning_rate": 2.563029306478304e-07,
      "loss": 0.1072,
      "step": 12592
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.381329074036769,
      "learning_rate": 2.559369334083328e-07,
      "loss": 0.1501,
      "step": 12593
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.13616648166765,
      "learning_rate": 2.5557119081129343e-07,
      "loss": 0.6089,
      "step": 12594
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2667537866810106,
      "learning_rate": 2.5520570287634383e-07,
      "loss": 0.1715,
      "step": 12595
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3751099081153582,
      "learning_rate": 2.548404696231027e-07,
      "loss": 0.1911,
      "step": 12596
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.214955755228897,
      "learning_rate": 2.544754910711733e-07,
      "loss": 0.1575,
      "step": 12597
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2761027544301025,
      "learning_rate": 2.5411076724014715e-07,
      "loss": 0.1935,
      "step": 12598
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4992574399977376,
      "learning_rate": 2.537462981495997e-07,
      "loss": 0.1735,
      "step": 12599
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2901441287426927,
      "learning_rate": 2.533820838190959e-07,
      "loss": 0.1676,
      "step": 12600
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3562405522916454,
      "learning_rate": 2.530181242681834e-07,
      "loss": 0.1528,
      "step": 12601
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1891191079285433,
      "learning_rate": 2.526544195163999e-07,
      "loss": 0.1317,
      "step": 12602
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1358061463714986,
      "learning_rate": 2.522909695832665e-07,
      "loss": 0.1646,
      "step": 12603
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4563545500093997,
      "learning_rate": 2.519277744882931e-07,
      "loss": 0.1795,
      "step": 12604
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3757213132838086,
      "learning_rate": 2.51564834250973e-07,
      "loss": 0.1664,
      "step": 12605
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3615375691049343,
      "learning_rate": 2.5120214889078785e-07,
      "loss": 0.1404,
      "step": 12606
    },
    {
      "epoch": 0.9,
      "grad_norm": 8.115128303924921,
      "learning_rate": 2.5083971842720654e-07,
      "loss": 0.6011,
      "step": 12607
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3777547141643542,
      "learning_rate": 2.5047754287968075e-07,
      "loss": 0.1835,
      "step": 12608
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4550370281847165,
      "learning_rate": 2.5011562226765205e-07,
      "loss": 0.1729,
      "step": 12609
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.772965394498877,
      "learning_rate": 2.4975395661054614e-07,
      "loss": 0.5983,
      "step": 12610
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.452928597215935,
      "learning_rate": 2.4939254592777573e-07,
      "loss": 0.5357,
      "step": 12611
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2980012496121758,
      "learning_rate": 2.490313902387409e-07,
      "loss": 0.1391,
      "step": 12612
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3526996219450143,
      "learning_rate": 2.486704895628267e-07,
      "loss": 0.1504,
      "step": 12613
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3600802230817879,
      "learning_rate": 2.483098439194037e-07,
      "loss": 0.1849,
      "step": 12614
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.323109597117248,
      "learning_rate": 2.4794945332783205e-07,
      "loss": 0.4746,
      "step": 12615
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2898502608589355,
      "learning_rate": 2.47589317807454e-07,
      "loss": 0.1806,
      "step": 12616
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4890459380942849,
      "learning_rate": 2.472294373776013e-07,
      "loss": 0.1567,
      "step": 12617
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2869522820229289,
      "learning_rate": 2.468698120575902e-07,
      "loss": 0.1382,
      "step": 12618
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.374671484906084,
      "learning_rate": 2.465104418667247e-07,
      "loss": 0.1589,
      "step": 12619
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.465379589483349,
      "learning_rate": 2.461513268242938e-07,
      "loss": 0.1438,
      "step": 12620
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.268249103254443,
      "learning_rate": 2.4579246694957426e-07,
      "loss": 0.1795,
      "step": 12621
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2648286021344564,
      "learning_rate": 2.4543386226182686e-07,
      "loss": 0.1856,
      "step": 12622
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1864698921227268,
      "learning_rate": 2.4507551278030173e-07,
      "loss": 0.1345,
      "step": 12623
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4440622907300216,
      "learning_rate": 2.447174185242324e-07,
      "loss": 0.1621,
      "step": 12624
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.798272506632053,
      "learning_rate": 2.443595795128406e-07,
      "loss": 0.4578,
      "step": 12625
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3258373802044958,
      "learning_rate": 2.440019957653328e-07,
      "loss": 0.1712,
      "step": 12626
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3808373471207442,
      "learning_rate": 2.4364466730090355e-07,
      "loss": 0.219,
      "step": 12627
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3594122353703182,
      "learning_rate": 2.4328759413873194e-07,
      "loss": 0.1769,
      "step": 12628
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2489932798227668,
      "learning_rate": 2.42930776297986e-07,
      "loss": 0.1419,
      "step": 12629
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3075258853304108,
      "learning_rate": 2.4257421379781654e-07,
      "loss": 0.1391,
      "step": 12630
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5263125558689432,
      "learning_rate": 2.4221790665736325e-07,
      "loss": 0.2145,
      "step": 12631
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.362030507587289,
      "learning_rate": 2.4186185489575076e-07,
      "loss": 0.1567,
      "step": 12632
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2511353758378456,
      "learning_rate": 2.4150605853209164e-07,
      "loss": 0.1594,
      "step": 12633
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.287665169396671,
      "learning_rate": 2.411505175854817e-07,
      "loss": 0.1951,
      "step": 12634
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2157894373998381,
      "learning_rate": 2.407952320750062e-07,
      "loss": 0.1753,
      "step": 12635
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4326108665458335,
      "learning_rate": 2.404402020197355e-07,
      "loss": 0.1756,
      "step": 12636
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4382546042767963,
      "learning_rate": 2.4008542743872597e-07,
      "loss": 0.1896,
      "step": 12637
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.180132108354165,
      "learning_rate": 2.397309083510213e-07,
      "loss": 0.1233,
      "step": 12638
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2945141482590476,
      "learning_rate": 2.3937664477564905e-07,
      "loss": 0.1647,
      "step": 12639
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1998086146363998,
      "learning_rate": 2.390226367316262e-07,
      "loss": 0.1615,
      "step": 12640
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.367934762975672,
      "learning_rate": 2.386688842379531e-07,
      "loss": 0.1894,
      "step": 12641
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.361737265873703,
      "learning_rate": 2.383153873136196e-07,
      "loss": 0.1412,
      "step": 12642
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4918394390969996,
      "learning_rate": 2.3796214597759825e-07,
      "loss": 0.1641,
      "step": 12643
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4099074226139585,
      "learning_rate": 2.3760916024885007e-07,
      "loss": 0.1919,
      "step": 12644
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6334725342104919,
      "learning_rate": 2.3725643014632205e-07,
      "loss": 0.2573,
      "step": 12645
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4623518045673327,
      "learning_rate": 2.3690395568894853e-07,
      "loss": 0.1873,
      "step": 12646
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2134875684352853,
      "learning_rate": 2.3655173689564714e-07,
      "loss": 0.1618,
      "step": 12647
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2261628665144881,
      "learning_rate": 2.36199773785325e-07,
      "loss": 0.1381,
      "step": 12648
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3269781824631393,
      "learning_rate": 2.3584806637687252e-07,
      "loss": 0.1367,
      "step": 12649
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.385156350246033,
      "learning_rate": 2.3549661468916963e-07,
      "loss": 0.1514,
      "step": 12650
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2972375857117198,
      "learning_rate": 2.3514541874107954e-07,
      "loss": 0.1677,
      "step": 12651
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.088758649983839,
      "learning_rate": 2.3479447855145388e-07,
      "loss": 0.611,
      "step": 12652
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2714985403771593,
      "learning_rate": 2.3444379413912865e-07,
      "loss": 0.1595,
      "step": 12653
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3952634400063104,
      "learning_rate": 2.3409336552292937e-07,
      "loss": 0.2054,
      "step": 12654
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2878854804319144,
      "learning_rate": 2.3374319272166324e-07,
      "loss": 0.1565,
      "step": 12655
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.390845028815124,
      "learning_rate": 2.3339327575412852e-07,
      "loss": 0.1471,
      "step": 12656
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3455684222514772,
      "learning_rate": 2.3304361463910463e-07,
      "loss": 0.1444,
      "step": 12657
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7130207936078783,
      "learning_rate": 2.3269420939536213e-07,
      "loss": 0.1845,
      "step": 12658
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2157470666127261,
      "learning_rate": 2.3234506004165492e-07,
      "loss": 0.149,
      "step": 12659
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.210971755825889,
      "learning_rate": 2.3199616659672352e-07,
      "loss": 0.5211,
      "step": 12660
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.275369519662811,
      "learning_rate": 2.3164752907929522e-07,
      "loss": 0.1495,
      "step": 12661
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3570798167669922,
      "learning_rate": 2.3129914750808503e-07,
      "loss": 0.1345,
      "step": 12662
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.355340402076586,
      "learning_rate": 2.309510219017913e-07,
      "loss": 0.1689,
      "step": 12663
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4251630076528905,
      "learning_rate": 2.3060315227910023e-07,
      "loss": 0.1748,
      "step": 12664
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2901020450337168,
      "learning_rate": 2.3025553865868412e-07,
      "loss": 0.1731,
      "step": 12665
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3455511759679402,
      "learning_rate": 2.2990818105920132e-07,
      "loss": 0.1597,
      "step": 12666
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5308538427693263,
      "learning_rate": 2.2956107949929696e-07,
      "loss": 0.2,
      "step": 12667
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3524891383994382,
      "learning_rate": 2.2921423399760168e-07,
      "loss": 0.1624,
      "step": 12668
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1494135250024697,
      "learning_rate": 2.288676445727328e-07,
      "loss": 0.1094,
      "step": 12669
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3464442488383974,
      "learning_rate": 2.285213112432938e-07,
      "loss": 0.1502,
      "step": 12670
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.074477184718176,
      "learning_rate": 2.281752340278759e-07,
      "loss": 0.4917,
      "step": 12671
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7279358846116963,
      "learning_rate": 2.2782941294505312e-07,
      "loss": 0.2067,
      "step": 12672
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1919272580318094,
      "learning_rate": 2.2748384801338953e-07,
      "loss": 0.1452,
      "step": 12673
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.72844635430164,
      "learning_rate": 2.271385392514319e-07,
      "loss": 0.6493,
      "step": 12674
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5019772434597523,
      "learning_rate": 2.2679348667771662e-07,
      "loss": 0.1417,
      "step": 12675
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.391474200074532,
      "learning_rate": 2.264486903107632e-07,
      "loss": 0.2061,
      "step": 12676
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.1715158813366795,
      "learning_rate": 2.2610415016907973e-07,
      "loss": 0.6574,
      "step": 12677
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2851411197759597,
      "learning_rate": 2.257598662711602e-07,
      "loss": 0.1671,
      "step": 12678
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.489362709641089,
      "learning_rate": 2.2541583863548433e-07,
      "loss": 0.1735,
      "step": 12679
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4497275435218506,
      "learning_rate": 2.2507206728051732e-07,
      "loss": 0.1777,
      "step": 12680
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2438421670019086,
      "learning_rate": 2.2472855222471225e-07,
      "loss": 0.1335,
      "step": 12681
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3333968076745413,
      "learning_rate": 2.2438529348650707e-07,
      "loss": 0.1608,
      "step": 12682
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3080865554542618,
      "learning_rate": 2.2404229108432762e-07,
      "loss": 0.1746,
      "step": 12683
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3560911245380336,
      "learning_rate": 2.2369954503658308e-07,
      "loss": 0.1493,
      "step": 12684
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.34216046791445,
      "learning_rate": 2.2335705536167263e-07,
      "loss": 0.1697,
      "step": 12685
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2421169726672066,
      "learning_rate": 2.2301482207797765e-07,
      "loss": 0.1602,
      "step": 12686
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4440396982952837,
      "learning_rate": 2.226728452038701e-07,
      "loss": 0.1917,
      "step": 12687
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4842486618204693,
      "learning_rate": 2.2233112475770425e-07,
      "loss": 0.1508,
      "step": 12688
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.31412708120763,
      "learning_rate": 2.2198966075782368e-07,
      "loss": 0.4765,
      "step": 12689
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.401507444196246,
      "learning_rate": 2.2164845322255545e-07,
      "loss": 0.1554,
      "step": 12690
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.593578750664929,
      "learning_rate": 2.2130750217021546e-07,
      "loss": 0.1843,
      "step": 12691
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.387124870363444,
      "learning_rate": 2.2096680761910349e-07,
      "loss": 0.5165,
      "step": 12692
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1720870993437176,
      "learning_rate": 2.206263695875077e-07,
      "loss": 0.1271,
      "step": 12693
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.29065390664548,
      "learning_rate": 2.2028618809370073e-07,
      "loss": 0.5026,
      "step": 12694
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3341181283120755,
      "learning_rate": 2.1994626315594181e-07,
      "loss": 0.1362,
      "step": 12695
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.166459844987831,
      "learning_rate": 2.1960659479247749e-07,
      "loss": 0.1416,
      "step": 12696
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3180970280588613,
      "learning_rate": 2.1926718302154037e-07,
      "loss": 0.1438,
      "step": 12697
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.205945768344295,
      "learning_rate": 2.189280278613476e-07,
      "loss": 0.1575,
      "step": 12698
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5407504146541176,
      "learning_rate": 2.1858912933010456e-07,
      "loss": 0.1715,
      "step": 12699
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.240102079430423,
      "learning_rate": 2.1825048744600062e-07,
      "loss": 0.1398,
      "step": 12700
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3946866416980106,
      "learning_rate": 2.1791210222721405e-07,
      "loss": 0.1655,
      "step": 12701
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3432019843021943,
      "learning_rate": 2.1757397369190802e-07,
      "loss": 0.1773,
      "step": 12702
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.848624915876966,
      "learning_rate": 2.1723610185823085e-07,
      "loss": 0.4256,
      "step": 12703
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3644903532423924,
      "learning_rate": 2.1689848674431912e-07,
      "loss": 0.2049,
      "step": 12704
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2659687446524321,
      "learning_rate": 2.165611283682939e-07,
      "loss": 0.1793,
      "step": 12705
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2604631340377983,
      "learning_rate": 2.162240267482646e-07,
      "loss": 0.158,
      "step": 12706
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2062277517590558,
      "learning_rate": 2.1588718190232395e-07,
      "loss": 0.1449,
      "step": 12707
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.0812775361243,
      "learning_rate": 2.1555059384855358e-07,
      "loss": 0.5837,
      "step": 12708
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2652800820562675,
      "learning_rate": 2.1521426260501965e-07,
      "loss": 0.1547,
      "step": 12709
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.292272420480728,
      "learning_rate": 2.1487818818977545e-07,
      "loss": 0.1642,
      "step": 12710
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4300264731553758,
      "learning_rate": 2.145423706208588e-07,
      "loss": 0.1786,
      "step": 12711
    },
    {
      "epoch": 0.91,
      "grad_norm": 12.608328991227296,
      "learning_rate": 2.142068099162964e-07,
      "loss": 0.6462,
      "step": 12712
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2223192099129898,
      "learning_rate": 2.1387150609409991e-07,
      "loss": 0.1537,
      "step": 12713
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2811856241231114,
      "learning_rate": 2.135364591722666e-07,
      "loss": 0.1405,
      "step": 12714
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2919615344656545,
      "learning_rate": 2.132016691687805e-07,
      "loss": 0.1696,
      "step": 12715
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4143280558702906,
      "learning_rate": 2.1286713610161215e-07,
      "loss": 0.1627,
      "step": 12716
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3171704607557675,
      "learning_rate": 2.1253285998871719e-07,
      "loss": 0.1483,
      "step": 12717
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3888484234119711,
      "learning_rate": 2.121988408480391e-07,
      "loss": 0.1935,
      "step": 12718
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.389201521309338,
      "learning_rate": 2.1186507869750573e-07,
      "loss": 0.1567,
      "step": 12719
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.939276780240387,
      "learning_rate": 2.1153157355503274e-07,
      "loss": 0.4929,
      "step": 12720
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2849870808185841,
      "learning_rate": 2.1119832543852137e-07,
      "loss": 0.1423,
      "step": 12721
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4064810631375324,
      "learning_rate": 2.108653343658601e-07,
      "loss": 0.1909,
      "step": 12722
    },
    {
      "epoch": 0.91,
      "grad_norm": 11.933941021940372,
      "learning_rate": 2.1053260035492017e-07,
      "loss": 0.5475,
      "step": 12723
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.400823236097534,
      "learning_rate": 2.1020012342356344e-07,
      "loss": 0.1914,
      "step": 12724
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.194602002841027,
      "learning_rate": 2.0986790358963449e-07,
      "loss": 0.1618,
      "step": 12725
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4286086054600553,
      "learning_rate": 2.095359408709674e-07,
      "loss": 0.1613,
      "step": 12726
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2315076608409024,
      "learning_rate": 2.0920423528537848e-07,
      "loss": 0.1501,
      "step": 12727
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3306083422762518,
      "learning_rate": 2.0887278685067348e-07,
      "loss": 0.1787,
      "step": 12728
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0917576309404418,
      "learning_rate": 2.0854159558464316e-07,
      "loss": 0.1468,
      "step": 12729
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4355306159506678,
      "learning_rate": 2.0821066150506497e-07,
      "loss": 0.1825,
      "step": 12730
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4181918270894824,
      "learning_rate": 2.078799846297014e-07,
      "loss": 0.1582,
      "step": 12731
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3544895023806705,
      "learning_rate": 2.0754956497630262e-07,
      "loss": 0.1685,
      "step": 12732
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3791696139548792,
      "learning_rate": 2.072194025626034e-07,
      "loss": 0.191,
      "step": 12733
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1845903456790687,
      "learning_rate": 2.068894974063257e-07,
      "loss": 0.1266,
      "step": 12734
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4915368865117113,
      "learning_rate": 2.0655984952517806e-07,
      "loss": 0.2032,
      "step": 12735
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.244248709550514,
      "learning_rate": 2.0623045893685413e-07,
      "loss": 0.1474,
      "step": 12736
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3279174530285651,
      "learning_rate": 2.0590132565903475e-07,
      "loss": 0.1817,
      "step": 12737
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5801327434120815,
      "learning_rate": 2.0557244970938583e-07,
      "loss": 0.1975,
      "step": 12738
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4573533212946175,
      "learning_rate": 2.0524383110556156e-07,
      "loss": 0.1518,
      "step": 12739
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.302924610356093,
      "learning_rate": 2.0491546986519896e-07,
      "loss": 0.1998,
      "step": 12740
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2155938743662618,
      "learning_rate": 2.04587366005925e-07,
      "loss": 0.1548,
      "step": 12741
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4973755541503786,
      "learning_rate": 2.0425951954534896e-07,
      "loss": 0.1731,
      "step": 12742
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2550542012626813,
      "learning_rate": 2.0393193050107064e-07,
      "loss": 0.1497,
      "step": 12743
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.9313190551921675,
      "learning_rate": 2.0360459889067153e-07,
      "loss": 0.4868,
      "step": 12744
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3243342929166981,
      "learning_rate": 2.0327752473172256e-07,
      "loss": 0.1713,
      "step": 12745
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.176956660120435,
      "learning_rate": 2.0295070804177974e-07,
      "loss": 0.1319,
      "step": 12746
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3982550048429367,
      "learning_rate": 2.0262414883838565e-07,
      "loss": 0.158,
      "step": 12747
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4570055716443895,
      "learning_rate": 2.0229784713906796e-07,
      "loss": 0.1825,
      "step": 12748
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1312398118018703,
      "learning_rate": 2.019718029613421e-07,
      "loss": 0.1762,
      "step": 12749
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4268647707584796,
      "learning_rate": 2.0164601632270797e-07,
      "loss": 0.195,
      "step": 12750
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3915743114432548,
      "learning_rate": 2.0132048724065323e-07,
      "loss": 0.1792,
      "step": 12751
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.222932772242943,
      "learning_rate": 2.0099521573265e-07,
      "loss": 0.1559,
      "step": 12752
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5233677766233629,
      "learning_rate": 2.006702018161577e-07,
      "loss": 0.1866,
      "step": 12753
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3666646012117158,
      "learning_rate": 2.003454455086229e-07,
      "loss": 0.1668,
      "step": 12754
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3783722747849652,
      "learning_rate": 2.0002094682747663e-07,
      "loss": 0.1447,
      "step": 12755
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.261074088016899,
      "learning_rate": 1.996967057901361e-07,
      "loss": 0.1507,
      "step": 12756
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2987466133276455,
      "learning_rate": 1.9937272241400685e-07,
      "loss": 0.2055,
      "step": 12757
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2160074800438412,
      "learning_rate": 1.9904899671647715e-07,
      "loss": 0.1551,
      "step": 12758
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2916984396955824,
      "learning_rate": 1.9872552871492478e-07,
      "loss": 0.158,
      "step": 12759
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4241466276475483,
      "learning_rate": 1.9840231842671087e-07,
      "loss": 0.194,
      "step": 12760
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3674066382133252,
      "learning_rate": 1.9807936586918485e-07,
      "loss": 0.167,
      "step": 12761
    },
    {
      "epoch": 0.91,
      "grad_norm": 14.339278699785922,
      "learning_rate": 1.977566710596812e-07,
      "loss": 0.478,
      "step": 12762
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3730802100404944,
      "learning_rate": 1.974342340155222e-07,
      "loss": 0.1869,
      "step": 12763
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.393961963912052,
      "learning_rate": 1.9711205475401286e-07,
      "loss": 0.2109,
      "step": 12764
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3612032996038137,
      "learning_rate": 1.9679013329244766e-07,
      "loss": 0.1632,
      "step": 12765
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.37050486797022,
      "learning_rate": 1.9646846964810672e-07,
      "loss": 0.1549,
      "step": 12766
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2643573707724556,
      "learning_rate": 1.9614706383825454e-07,
      "loss": 0.1857,
      "step": 12767
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5286755492706403,
      "learning_rate": 1.9582591588014344e-07,
      "loss": 0.1764,
      "step": 12768
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3630176807988492,
      "learning_rate": 1.955050257910107e-07,
      "loss": 0.1762,
      "step": 12769
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4368104550676726,
      "learning_rate": 1.951843935880815e-07,
      "loss": 0.1868,
      "step": 12770
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4312590684300504,
      "learning_rate": 1.948640192885648e-07,
      "loss": 0.1897,
      "step": 12771
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3363325117674558,
      "learning_rate": 1.9454390290965853e-07,
      "loss": 0.1665,
      "step": 12772
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1983137451812842,
      "learning_rate": 1.9422404446854393e-07,
      "loss": 0.1426,
      "step": 12773
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.364344406207192,
      "learning_rate": 1.9390444398239117e-07,
      "loss": 0.1422,
      "step": 12774
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2735006406516576,
      "learning_rate": 1.9358510146835318e-07,
      "loss": 0.1645,
      "step": 12775
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.42555570748825,
      "learning_rate": 1.9326601694357296e-07,
      "loss": 0.4525,
      "step": 12776
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3865371850103634,
      "learning_rate": 1.9294719042517617e-07,
      "loss": 0.1404,
      "step": 12777
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2844491232657287,
      "learning_rate": 1.92628621930277e-07,
      "loss": 0.1591,
      "step": 12778
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5943022315536373,
      "learning_rate": 1.9231031147597446e-07,
      "loss": 0.1891,
      "step": 12779
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1856540736482608,
      "learning_rate": 1.9199225907935492e-07,
      "loss": 0.1619,
      "step": 12780
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.384493799844268,
      "learning_rate": 1.916744647574892e-07,
      "loss": 0.124,
      "step": 12781
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5264940808139795,
      "learning_rate": 1.9135692852743637e-07,
      "loss": 0.1865,
      "step": 12782
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3477894799329293,
      "learning_rate": 1.910396504062395e-07,
      "loss": 0.1455,
      "step": 12783
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4738063550447547,
      "learning_rate": 1.9072263041092998e-07,
      "loss": 0.1654,
      "step": 12784
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.273004954345331,
      "learning_rate": 1.904058685585225e-07,
      "loss": 0.1499,
      "step": 12785
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3647244268406307,
      "learning_rate": 1.900893648660207e-07,
      "loss": 0.1918,
      "step": 12786
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3147723721912674,
      "learning_rate": 1.897731193504132e-07,
      "loss": 0.1733,
      "step": 12787
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5137023341174125,
      "learning_rate": 1.894571320286742e-07,
      "loss": 0.1452,
      "step": 12788
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2739711704495216,
      "learning_rate": 1.8914140291776516e-07,
      "loss": 0.184,
      "step": 12789
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.317344106338329,
      "learning_rate": 1.888259320346336e-07,
      "loss": 0.189,
      "step": 12790
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3675545493382846,
      "learning_rate": 1.8851071939621157e-07,
      "loss": 0.1554,
      "step": 12791
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3532705112489385,
      "learning_rate": 1.8819576501941995e-07,
      "loss": 0.1442,
      "step": 12792
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.320639052149005,
      "learning_rate": 1.8788106892116243e-07,
      "loss": 0.1368,
      "step": 12793
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4273980400305475,
      "learning_rate": 1.8756663111833274e-07,
      "loss": 0.1602,
      "step": 12794
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2804584621244166,
      "learning_rate": 1.8725245162780625e-07,
      "loss": 0.15,
      "step": 12795
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5179165168103461,
      "learning_rate": 1.8693853046644838e-07,
      "loss": 0.1871,
      "step": 12796
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5290061771363928,
      "learning_rate": 1.86624867651109e-07,
      "loss": 0.1785,
      "step": 12797
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3524513085804102,
      "learning_rate": 1.8631146319862459e-07,
      "loss": 0.1447,
      "step": 12798
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.195468698768072,
      "learning_rate": 1.8599831712581728e-07,
      "loss": 0.1433,
      "step": 12799
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1825816397123174,
      "learning_rate": 1.8568542944949474e-07,
      "loss": 0.1728,
      "step": 12800
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.882215261329323,
      "learning_rate": 1.8537280018645243e-07,
      "loss": 0.559,
      "step": 12801
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3086463135748683,
      "learning_rate": 1.8506042935347025e-07,
      "loss": 0.1493,
      "step": 12802
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.433916015712827,
      "learning_rate": 1.8474831696731588e-07,
      "loss": 0.4228,
      "step": 12803
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4149473796727927,
      "learning_rate": 1.8443646304474206e-07,
      "loss": 0.2147,
      "step": 12804
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.378455823490417,
      "learning_rate": 1.8412486760248705e-07,
      "loss": 0.5911,
      "step": 12805
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3333284540094448,
      "learning_rate": 1.8381353065727636e-07,
      "loss": 0.1965,
      "step": 12806
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3424792342914524,
      "learning_rate": 1.8350245222582274e-07,
      "loss": 0.1687,
      "step": 12807
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3298522610183259,
      "learning_rate": 1.831916323248223e-07,
      "loss": 0.1667,
      "step": 12808
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.354789719890319,
      "learning_rate": 1.8288107097095887e-07,
      "loss": 0.1641,
      "step": 12809
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3871104225019073,
      "learning_rate": 1.8257076818090192e-07,
      "loss": 0.1926,
      "step": 12810
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2672795185829975,
      "learning_rate": 1.8226072397130755e-07,
      "loss": 0.1344,
      "step": 12811
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4348601102312597,
      "learning_rate": 1.819509383588175e-07,
      "loss": 0.1654,
      "step": 12812
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5196719280027633,
      "learning_rate": 1.816414113600601e-07,
      "loss": 0.1881,
      "step": 12813
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3362287003305124,
      "learning_rate": 1.8133214299164925e-07,
      "loss": 0.1777,
      "step": 12814
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2797063776665907,
      "learning_rate": 1.8102313327018617e-07,
      "loss": 0.2087,
      "step": 12815
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4723686324634557,
      "learning_rate": 1.807143822122559e-07,
      "loss": 0.1877,
      "step": 12816
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4246960496045753,
      "learning_rate": 1.8040588983443185e-07,
      "loss": 0.1662,
      "step": 12817
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.568065016145491,
      "learning_rate": 1.800976561532719e-07,
      "loss": 0.1817,
      "step": 12818
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1955192728492101,
      "learning_rate": 1.7978968118532226e-07,
      "loss": 0.1448,
      "step": 12819
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4478351914573695,
      "learning_rate": 1.794819649471119e-07,
      "loss": 0.2073,
      "step": 12820
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4682813705173308,
      "learning_rate": 1.7917450745515874e-07,
      "loss": 0.2157,
      "step": 12821
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.378405298776126,
      "learning_rate": 1.7886730872596625e-07,
      "loss": 0.1707,
      "step": 12822
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4207753909467031,
      "learning_rate": 1.7856036877602344e-07,
      "loss": 0.1865,
      "step": 12823
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5527661084262234,
      "learning_rate": 1.7825368762180496e-07,
      "loss": 0.1582,
      "step": 12824
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3120611712944024,
      "learning_rate": 1.779472652797737e-07,
      "loss": 0.1528,
      "step": 12825
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3689170371802326,
      "learning_rate": 1.776411017663754e-07,
      "loss": 0.1731,
      "step": 12826
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2977832575145447,
      "learning_rate": 1.7733519709804525e-07,
      "loss": 0.1533,
      "step": 12827
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5177522975788869,
      "learning_rate": 1.7702955129120125e-07,
      "loss": 0.1631,
      "step": 12828
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5444551598888603,
      "learning_rate": 1.7672416436225082e-07,
      "loss": 0.2274,
      "step": 12829
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3082135917193725,
      "learning_rate": 1.764190363275853e-07,
      "loss": 0.1905,
      "step": 12830
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2353417737363328,
      "learning_rate": 1.7611416720358266e-07,
      "loss": 0.1112,
      "step": 12831
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.376544398794052,
      "learning_rate": 1.7580955700660762e-07,
      "loss": 0.1965,
      "step": 12832
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2546797642941734,
      "learning_rate": 1.755052057530099e-07,
      "loss": 0.1611,
      "step": 12833
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3318230988802007,
      "learning_rate": 1.7520111345912637e-07,
      "loss": 0.1901,
      "step": 12834
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1471092268359164,
      "learning_rate": 1.7489728014127905e-07,
      "loss": 0.1308,
      "step": 12835
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2609314058257748,
      "learning_rate": 1.745937058157765e-07,
      "loss": 0.1285,
      "step": 12836
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3912497761656284,
      "learning_rate": 1.742903904989135e-07,
      "loss": 0.1809,
      "step": 12837
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.34030083386807,
      "learning_rate": 1.7398733420697033e-07,
      "loss": 0.175,
      "step": 12838
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3641488291375587,
      "learning_rate": 1.7368453695621456e-07,
      "loss": 0.1559,
      "step": 12839
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.52920484075741,
      "learning_rate": 1.7338199876289984e-07,
      "loss": 0.2003,
      "step": 12840
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.691867749837039,
      "learning_rate": 1.730797196432632e-07,
      "loss": 0.5169,
      "step": 12841
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5431078769645992,
      "learning_rate": 1.7277769961353165e-07,
      "loss": 0.1891,
      "step": 12842
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.538845839325577,
      "learning_rate": 1.7247593868991496e-07,
      "loss": 0.4731,
      "step": 12843
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2867288000206403,
      "learning_rate": 1.721744368886119e-07,
      "loss": 0.1478,
      "step": 12844
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3257499301183564,
      "learning_rate": 1.7187319422580451e-07,
      "loss": 0.1589,
      "step": 12845
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.905120048463217,
      "learning_rate": 1.7157221071766316e-07,
      "loss": 0.4166,
      "step": 12846
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2254771028913514,
      "learning_rate": 1.7127148638034274e-07,
      "loss": 0.1323,
      "step": 12847
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4773835885105777,
      "learning_rate": 1.7097102122998588e-07,
      "loss": 0.2281,
      "step": 12848
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2417969338478139,
      "learning_rate": 1.7067081528271966e-07,
      "loss": 0.1393,
      "step": 12849
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1834486968188433,
      "learning_rate": 1.7037086855465902e-07,
      "loss": 0.1433,
      "step": 12850
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3465693673928576,
      "learning_rate": 1.7007118106190213e-07,
      "loss": 0.1823,
      "step": 12851
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.626746650203718,
      "learning_rate": 1.6977175282053672e-07,
      "loss": 0.5873,
      "step": 12852
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.437169898472335,
      "learning_rate": 1.6947258384663322e-07,
      "loss": 0.1736,
      "step": 12853
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.38632346129207,
      "learning_rate": 1.6917367415625107e-07,
      "loss": 0.1793,
      "step": 12854
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.246604364869626,
      "learning_rate": 1.6887502376543407e-07,
      "loss": 0.1614,
      "step": 12855
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4662093112879977,
      "learning_rate": 1.685766326902133e-07,
      "loss": 0.1781,
      "step": 12856
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5890817347206132,
      "learning_rate": 1.6827850094660424e-07,
      "loss": 0.2021,
      "step": 12857
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3837649229609714,
      "learning_rate": 1.6798062855061025e-07,
      "loss": 0.1832,
      "step": 12858
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.367038361183578,
      "learning_rate": 1.6768301551821908e-07,
      "loss": 0.1491,
      "step": 12859
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3541996635276008,
      "learning_rate": 1.6738566186540628e-07,
      "loss": 0.1538,
      "step": 12860
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2440045619088251,
      "learning_rate": 1.6708856760813187e-07,
      "loss": 0.1324,
      "step": 12861
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5797297998894513,
      "learning_rate": 1.667917327623425e-07,
      "loss": 0.1566,
      "step": 12862
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2573092740785436,
      "learning_rate": 1.6649515734397214e-07,
      "loss": 0.1422,
      "step": 12863
    },
    {
      "epoch": 0.92,
      "grad_norm": 11.432367036544106,
      "learning_rate": 1.6619884136893915e-07,
      "loss": 0.5574,
      "step": 12864
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.163731921719996,
      "learning_rate": 1.6590278485314914e-07,
      "loss": 0.161,
      "step": 12865
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3408682819827338,
      "learning_rate": 1.6560698781249218e-07,
      "loss": 0.1515,
      "step": 12866
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.245681832770281,
      "learning_rate": 1.6531145026284667e-07,
      "loss": 0.1698,
      "step": 12867
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4037636132878548,
      "learning_rate": 1.6501617222007492e-07,
      "loss": 0.1935,
      "step": 12868
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5000233103807037,
      "learning_rate": 1.6472115370002705e-07,
      "loss": 0.1645,
      "step": 12869
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.916082454841228,
      "learning_rate": 1.6442639471853762e-07,
      "loss": 0.4796,
      "step": 12870
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.611564504019174,
      "learning_rate": 1.6413189529142838e-07,
      "loss": 0.6148,
      "step": 12871
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5505699569923206,
      "learning_rate": 1.6383765543450724e-07,
      "loss": 0.1863,
      "step": 12872
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4012195447185236,
      "learning_rate": 1.635436751635683e-07,
      "loss": 0.2008,
      "step": 12873
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5143417399237455,
      "learning_rate": 1.6324995449439053e-07,
      "loss": 0.1381,
      "step": 12874
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.732242935702377,
      "learning_rate": 1.6295649344273968e-07,
      "loss": 0.4083,
      "step": 12875
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4580546232263518,
      "learning_rate": 1.6266329202436758e-07,
      "loss": 0.2346,
      "step": 12876
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1821854518223598,
      "learning_rate": 1.623703502550128e-07,
      "loss": 0.1464,
      "step": 12877
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6275301531509294,
      "learning_rate": 1.620776681503977e-07,
      "loss": 0.2043,
      "step": 12878
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4274674232524396,
      "learning_rate": 1.6178524572623367e-07,
      "loss": 0.137,
      "step": 12879
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2871975339199262,
      "learning_rate": 1.6149308299821643e-07,
      "loss": 0.1903,
      "step": 12880
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4624109661668023,
      "learning_rate": 1.61201179982029e-07,
      "loss": 0.1624,
      "step": 12881
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4062327411793167,
      "learning_rate": 1.6090953669333775e-07,
      "loss": 0.138,
      "step": 12882
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2791369612992154,
      "learning_rate": 1.606181531477985e-07,
      "loss": 0.1568,
      "step": 12883
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3992412508669223,
      "learning_rate": 1.603270293610504e-07,
      "loss": 0.1435,
      "step": 12884
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3364573951294005,
      "learning_rate": 1.600361653487209e-07,
      "loss": 0.1704,
      "step": 12885
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4880208469135259,
      "learning_rate": 1.5974556112642147e-07,
      "loss": 0.1942,
      "step": 12886
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3665210421700824,
      "learning_rate": 1.5945521670975128e-07,
      "loss": 0.1728,
      "step": 12887
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3651330828997572,
      "learning_rate": 1.591651321142945e-07,
      "loss": 0.192,
      "step": 12888
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.3734949761380655,
      "learning_rate": 1.5887530735562152e-07,
      "loss": 0.5842,
      "step": 12889
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.39587384754957,
      "learning_rate": 1.5858574244928926e-07,
      "loss": 0.19,
      "step": 12890
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.083456310151006,
      "learning_rate": 1.582964374108409e-07,
      "loss": 0.1289,
      "step": 12891
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2269041243617134,
      "learning_rate": 1.5800739225580452e-07,
      "loss": 0.1427,
      "step": 12892
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3823273105647054,
      "learning_rate": 1.5771860699969445e-07,
      "loss": 0.1472,
      "step": 12893
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6096440171397919,
      "learning_rate": 1.574300816580132e-07,
      "loss": 0.1607,
      "step": 12894
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2169577171172035,
      "learning_rate": 1.5714181624624625e-07,
      "loss": 0.1522,
      "step": 12895
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2899281018294328,
      "learning_rate": 1.5685381077986727e-07,
      "loss": 0.1369,
      "step": 12896
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3504956597421125,
      "learning_rate": 1.565660652743345e-07,
      "loss": 0.1625,
      "step": 12897
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4600053487346356,
      "learning_rate": 1.562785797450933e-07,
      "loss": 0.1768,
      "step": 12898
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.350876291909039,
      "learning_rate": 1.559913542075747e-07,
      "loss": 0.1684,
      "step": 12899
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.492894935337053,
      "learning_rate": 1.5570438867719695e-07,
      "loss": 0.1697,
      "step": 12900
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4524982054585263,
      "learning_rate": 1.5541768316936157e-07,
      "loss": 0.1975,
      "step": 12901
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.148422425918754,
      "learning_rate": 1.5513123769945847e-07,
      "loss": 0.5124,
      "step": 12902
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3452216940275108,
      "learning_rate": 1.5484505228286317e-07,
      "loss": 0.1688,
      "step": 12903
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3194292911203649,
      "learning_rate": 1.5455912693493668e-07,
      "loss": 0.1784,
      "step": 12904
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3770426527454505,
      "learning_rate": 1.5427346167102564e-07,
      "loss": 0.1713,
      "step": 12905
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3357997850634453,
      "learning_rate": 1.5398805650646443e-07,
      "loss": 0.1716,
      "step": 12906
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3538916176121085,
      "learning_rate": 1.5370291145657247e-07,
      "loss": 0.1634,
      "step": 12907
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.367050146082185,
      "learning_rate": 1.5341802653665527e-07,
      "loss": 0.1646,
      "step": 12908
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3863426338217126,
      "learning_rate": 1.5313340176200285e-07,
      "loss": 0.1965,
      "step": 12909
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3606178826830897,
      "learning_rate": 1.5284903714789523e-07,
      "loss": 0.1785,
      "step": 12910
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4613775082693043,
      "learning_rate": 1.525649327095935e-07,
      "loss": 0.1677,
      "step": 12911
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5540744289976354,
      "learning_rate": 1.522810884623488e-07,
      "loss": 0.1972,
      "step": 12912
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2192449938417365,
      "learning_rate": 1.5199750442139615e-07,
      "loss": 0.1488,
      "step": 12913
    },
    {
      "epoch": 0.92,
      "grad_norm": 12.703233971197784,
      "learning_rate": 1.517141806019573e-07,
      "loss": 0.5076,
      "step": 12914
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.330231492019149,
      "learning_rate": 1.5143111701924006e-07,
      "loss": 0.1505,
      "step": 12915
    },
    {
      "epoch": 0.92,
      "grad_norm": 11.610203280691792,
      "learning_rate": 1.511483136884384e-07,
      "loss": 0.7352,
      "step": 12916
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2303204031852368,
      "learning_rate": 1.5086577062473128e-07,
      "loss": 0.1521,
      "step": 12917
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.311486357413655,
      "learning_rate": 1.5058348784328547e-07,
      "loss": 0.146,
      "step": 12918
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3313614159749305,
      "learning_rate": 1.5030146535925217e-07,
      "loss": 0.1771,
      "step": 12919
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7519745892173373,
      "learning_rate": 1.500197031877698e-07,
      "loss": 0.1779,
      "step": 12920
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3438452715000122,
      "learning_rate": 1.4973820134396079e-07,
      "loss": 0.1651,
      "step": 12921
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5677381084250293,
      "learning_rate": 1.4945695984293685e-07,
      "loss": 0.1817,
      "step": 12922
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.8272386030734693,
      "learning_rate": 1.491759786997926e-07,
      "loss": 0.4609,
      "step": 12923
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4328138791713092,
      "learning_rate": 1.4889525792961103e-07,
      "loss": 0.1798,
      "step": 12924
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.339541059012164,
      "learning_rate": 1.4861479754745943e-07,
      "loss": 0.4788,
      "step": 12925
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.350559882043178,
      "learning_rate": 1.483345975683914e-07,
      "loss": 0.1603,
      "step": 12926
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.379839200113693,
      "learning_rate": 1.4805465800744877e-07,
      "loss": 0.4229,
      "step": 12927
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3792472209690887,
      "learning_rate": 1.4777497887965508e-07,
      "loss": 0.1896,
      "step": 12928
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.36359365528784,
      "learning_rate": 1.4749556020002443e-07,
      "loss": 0.1766,
      "step": 12929
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.326540428353443,
      "learning_rate": 1.4721640198355424e-07,
      "loss": 0.2028,
      "step": 12930
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4913169154942265,
      "learning_rate": 1.4693750424522757e-07,
      "loss": 0.1711,
      "step": 12931
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.39884134525124,
      "learning_rate": 1.4665886700001632e-07,
      "loss": 0.1852,
      "step": 12932
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.323525708225339,
      "learning_rate": 1.4638049026287572e-07,
      "loss": 0.1414,
      "step": 12933
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0932652740903774,
      "learning_rate": 1.4610237404874773e-07,
      "loss": 0.1265,
      "step": 12934
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4207941555917307,
      "learning_rate": 1.4582451837256095e-07,
      "loss": 0.1636,
      "step": 12935
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.40865501645935,
      "learning_rate": 1.4554692324922903e-07,
      "loss": 0.193,
      "step": 12936
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5067522653309107,
      "learning_rate": 1.4526958869365338e-07,
      "loss": 0.1843,
      "step": 12937
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3997914897268995,
      "learning_rate": 1.4499251472071873e-07,
      "loss": 0.1569,
      "step": 12938
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5249571025971718,
      "learning_rate": 1.4471570134529768e-07,
      "loss": 0.1998,
      "step": 12939
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4433698743624355,
      "learning_rate": 1.4443914858224938e-07,
      "loss": 0.1479,
      "step": 12940
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5523992832047322,
      "learning_rate": 1.441628564464176e-07,
      "loss": 0.1948,
      "step": 12941
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6378516115813424,
      "learning_rate": 1.438868249526315e-07,
      "loss": 0.2051,
      "step": 12942
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2128528613426908,
      "learning_rate": 1.436110541157093e-07,
      "loss": 0.1391,
      "step": 12943
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9841802211260668,
      "learning_rate": 1.433355439504519e-07,
      "loss": 0.1794,
      "step": 12944
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3325814272476917,
      "learning_rate": 1.4306029447164858e-07,
      "loss": 0.1966,
      "step": 12945
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.347697119996661,
      "learning_rate": 1.4278530569407257e-07,
      "loss": 0.1622,
      "step": 12946
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2773133500884537,
      "learning_rate": 1.425105776324842e-07,
      "loss": 0.1969,
      "step": 12947
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.356618109075472,
      "learning_rate": 1.4223611030163064e-07,
      "loss": 0.1892,
      "step": 12948
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4095478445827514,
      "learning_rate": 1.4196190371624397e-07,
      "loss": 0.1879,
      "step": 12949
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.097802976395776,
      "learning_rate": 1.4168795789104184e-07,
      "loss": 0.6606,
      "step": 12950
    },
    {
      "epoch": 0.93,
      "grad_norm": 12.31442018266902,
      "learning_rate": 1.4141427284072973e-07,
      "loss": 0.7216,
      "step": 12951
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2610130126756305,
      "learning_rate": 1.4114084857999698e-07,
      "loss": 0.1399,
      "step": 12952
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4560254381956967,
      "learning_rate": 1.4086768512352079e-07,
      "loss": 0.1583,
      "step": 12953
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3040342171574046,
      "learning_rate": 1.4059478248596215e-07,
      "loss": 0.1184,
      "step": 12954
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3316654826481873,
      "learning_rate": 1.4032214068197048e-07,
      "loss": 0.19,
      "step": 12955
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3402568257667655,
      "learning_rate": 1.4004975972617962e-07,
      "loss": 0.142,
      "step": 12956
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.160935222611165,
      "learning_rate": 1.397776396332101e-07,
      "loss": 0.1529,
      "step": 12957
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5175251881684066,
      "learning_rate": 1.3950578041766917e-07,
      "loss": 0.1673,
      "step": 12958
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.786461694698364,
      "learning_rate": 1.392341820941473e-07,
      "loss": 0.6165,
      "step": 12959
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3569909910398614,
      "learning_rate": 1.3896284467722398e-07,
      "loss": 0.1821,
      "step": 12960
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.975597444480245,
      "learning_rate": 1.3869176818146314e-07,
      "loss": 0.5669,
      "step": 12961
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3533362659141104,
      "learning_rate": 1.3842095262141585e-07,
      "loss": 0.15,
      "step": 12962
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.960541401546087,
      "learning_rate": 1.3815039801161723e-07,
      "loss": 0.3474,
      "step": 12963
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2680083681014855,
      "learning_rate": 1.3788010436659006e-07,
      "loss": 0.1714,
      "step": 12964
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.8324390497248215,
      "learning_rate": 1.3761007170084273e-07,
      "loss": 0.5217,
      "step": 12965
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3337206639634662,
      "learning_rate": 1.3734030002886978e-07,
      "loss": 0.1924,
      "step": 12966
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3467838888854387,
      "learning_rate": 1.3707078936515071e-07,
      "loss": 0.1906,
      "step": 12967
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.44934871803956,
      "learning_rate": 1.3680153972415343e-07,
      "loss": 0.5684,
      "step": 12968
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4529880658353362,
      "learning_rate": 1.3653255112032803e-07,
      "loss": 0.1908,
      "step": 12969
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3412948556016269,
      "learning_rate": 1.3626382356811408e-07,
      "loss": 0.1447,
      "step": 12970
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2658208042288468,
      "learning_rate": 1.35995357081935e-07,
      "loss": 0.1814,
      "step": 12971
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4584534496212038,
      "learning_rate": 1.357271516762021e-07,
      "loss": 0.1586,
      "step": 12972
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4108451912653879,
      "learning_rate": 1.354592073653105e-07,
      "loss": 0.2214,
      "step": 12973
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.63946851920727,
      "learning_rate": 1.3519152416364312e-07,
      "loss": 0.2029,
      "step": 12974
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.393962865895861,
      "learning_rate": 1.3492410208556795e-07,
      "loss": 0.6108,
      "step": 12975
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.536757571289754,
      "learning_rate": 1.3465694114543903e-07,
      "loss": 0.1624,
      "step": 12976
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2069847101656013,
      "learning_rate": 1.343900413575966e-07,
      "loss": 0.1486,
      "step": 12977
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.304099853911999,
      "learning_rate": 1.3412340273636636e-07,
      "loss": 0.1323,
      "step": 12978
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2626633941477252,
      "learning_rate": 1.3385702529606082e-07,
      "loss": 0.1864,
      "step": 12979
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3512974962584914,
      "learning_rate": 1.335909090509785e-07,
      "loss": 0.1931,
      "step": 12980
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3962725520776282,
      "learning_rate": 1.3332505401540186e-07,
      "loss": 0.1736,
      "step": 12981
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3317482919784012,
      "learning_rate": 1.3305946020360338e-07,
      "loss": 0.1753,
      "step": 12982
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3582307673986647,
      "learning_rate": 1.3279412762983668e-07,
      "loss": 0.1698,
      "step": 12983
    },
    {
      "epoch": 0.93,
      "grad_norm": 14.790905680492724,
      "learning_rate": 1.325290563083459e-07,
      "loss": 0.4909,
      "step": 12984
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5316658739383067,
      "learning_rate": 1.3226424625335686e-07,
      "loss": 0.1684,
      "step": 12985
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3640900131410525,
      "learning_rate": 1.3199969747908546e-07,
      "loss": 0.1767,
      "step": 12986
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0082411426812892,
      "learning_rate": 1.3173540999973034e-07,
      "loss": 0.0983,
      "step": 12987
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6212404149523039,
      "learning_rate": 1.3147138382947788e-07,
      "loss": 0.1743,
      "step": 12988
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3052263993813262,
      "learning_rate": 1.3120761898250013e-07,
      "loss": 0.153,
      "step": 12989
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.388540708638378,
      "learning_rate": 1.3094411547295406e-07,
      "loss": 0.1475,
      "step": 12990
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3532659090054722,
      "learning_rate": 1.3068087331498447e-07,
      "loss": 0.1556,
      "step": 12991
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.30881558198817,
      "learning_rate": 1.3041789252272063e-07,
      "loss": 0.1464,
      "step": 12992
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.392115111204779,
      "learning_rate": 1.3015517311027904e-07,
      "loss": 0.1915,
      "step": 12993
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2764244182103412,
      "learning_rate": 1.2989271509176004e-07,
      "loss": 0.157,
      "step": 12994
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.271604139215707,
      "learning_rate": 1.2963051848125352e-07,
      "loss": 0.1547,
      "step": 12995
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.703866424221616,
      "learning_rate": 1.2936858329283043e-07,
      "loss": 0.6039,
      "step": 12996
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.329588261496409,
      "learning_rate": 1.2910690954055282e-07,
      "loss": 0.1415,
      "step": 12997
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.415173397206805,
      "learning_rate": 1.2884549723846451e-07,
      "loss": 0.6107,
      "step": 12998
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.342267609363397,
      "learning_rate": 1.2858434640059814e-07,
      "loss": 0.1666,
      "step": 12999
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.304897421337096,
      "learning_rate": 1.2832345704097082e-07,
      "loss": 0.1644,
      "step": 13000
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4212017982709633,
      "learning_rate": 1.2806282917358692e-07,
      "loss": 0.1751,
      "step": 13001
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3177373540991115,
      "learning_rate": 1.2780246281243414e-07,
      "loss": 0.1407,
      "step": 13002
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4478646716571963,
      "learning_rate": 1.275423579714896e-07,
      "loss": 0.1891,
      "step": 13003
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1882960944908658,
      "learning_rate": 1.2728251466471387e-07,
      "loss": 0.1588,
      "step": 13004
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.399788493914658,
      "learning_rate": 1.270229329060546e-07,
      "loss": 0.1692,
      "step": 13005
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.269340039774937,
      "learning_rate": 1.267636127094446e-07,
      "loss": 0.1525,
      "step": 13006
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2735409203175627,
      "learning_rate": 1.265045540888038e-07,
      "loss": 0.142,
      "step": 13007
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4037757526766568,
      "learning_rate": 1.2624575705803666e-07,
      "loss": 0.1597,
      "step": 13008
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.883200284065154,
      "learning_rate": 1.2598722163103593e-07,
      "loss": 0.4594,
      "step": 13009
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.400190402513513,
      "learning_rate": 1.2572894782167667e-07,
      "loss": 0.1355,
      "step": 13010
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4207291454800643,
      "learning_rate": 1.2547093564382384e-07,
      "loss": 0.1754,
      "step": 13011
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4315867529724424,
      "learning_rate": 1.252131851113253e-07,
      "loss": 0.1528,
      "step": 13012
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.458978036337528,
      "learning_rate": 1.249556962380166e-07,
      "loss": 0.1596,
      "step": 13013
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3523558623896803,
      "learning_rate": 1.2469846903771843e-07,
      "loss": 0.1627,
      "step": 13014
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5271121669354406,
      "learning_rate": 1.24441503524238e-07,
      "loss": 0.1558,
      "step": 13015
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3453344696713097,
      "learning_rate": 1.2418479971136822e-07,
      "loss": 0.1551,
      "step": 13016
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.332534543665476,
      "learning_rate": 1.2392835761288802e-07,
      "loss": 0.1338,
      "step": 13017
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2941825192215042,
      "learning_rate": 1.2367217724256141e-07,
      "loss": 0.1598,
      "step": 13018
    },
    {
      "epoch": 0.93,
      "grad_norm": 11.537320213274038,
      "learning_rate": 1.234162586141402e-07,
      "loss": 0.6438,
      "step": 13019
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4544338958215004,
      "learning_rate": 1.2316060174136e-07,
      "loss": 0.1787,
      "step": 13020
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.747512231125022,
      "learning_rate": 1.2290520663794435e-07,
      "loss": 0.4764,
      "step": 13021
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.420694078192601,
      "learning_rate": 1.226500733176017e-07,
      "loss": 0.5657,
      "step": 13022
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2658350758569819,
      "learning_rate": 1.223952017940261e-07,
      "loss": 0.1225,
      "step": 13023
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2537207318497374,
      "learning_rate": 1.2214059208089825e-07,
      "loss": 0.1565,
      "step": 13024
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3872971259085132,
      "learning_rate": 1.2188624419188499e-07,
      "loss": 0.1808,
      "step": 13025
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2658627968217848,
      "learning_rate": 1.216321581406388e-07,
      "loss": 0.1321,
      "step": 13026
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.624023964992862,
      "learning_rate": 1.2137833394079701e-07,
      "loss": 0.583,
      "step": 13027
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3453485750899448,
      "learning_rate": 1.2112477160598546e-07,
      "loss": 0.162,
      "step": 13028
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4765884147791575,
      "learning_rate": 1.2087147114981214e-07,
      "loss": 0.1821,
      "step": 13029
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3711788968569627,
      "learning_rate": 1.2061843258587557e-07,
      "loss": 0.166,
      "step": 13030
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.894817986312573,
      "learning_rate": 1.2036565592775606e-07,
      "loss": 0.5937,
      "step": 13031
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3342730369248716,
      "learning_rate": 1.2011314118902273e-07,
      "loss": 0.1907,
      "step": 13032
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3948150810909734,
      "learning_rate": 1.1986088838322919e-07,
      "loss": 0.1707,
      "step": 13033
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2974649541602692,
      "learning_rate": 1.1960889752391512e-07,
      "loss": 0.1683,
      "step": 13034
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2191271656826363,
      "learning_rate": 1.19357168624607e-07,
      "loss": 0.1342,
      "step": 13035
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.50397157272611,
      "learning_rate": 1.1910570169881675e-07,
      "loss": 0.1661,
      "step": 13036
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2310629315843262,
      "learning_rate": 1.188544967600408e-07,
      "loss": 0.1559,
      "step": 13037
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2869831494558506,
      "learning_rate": 1.1860355382176447e-07,
      "loss": 0.1435,
      "step": 13038
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.508035417979797,
      "learning_rate": 1.1835287289745589e-07,
      "loss": 0.2021,
      "step": 13039
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4614234319272879,
      "learning_rate": 1.1810245400057152e-07,
      "loss": 0.1846,
      "step": 13040
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.24864616097747,
      "learning_rate": 1.1785229714455227e-07,
      "loss": 0.675,
      "step": 13041
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3413861124674022,
      "learning_rate": 1.1760240234282683e-07,
      "loss": 0.1278,
      "step": 13042
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3674041012899993,
      "learning_rate": 1.1735276960880726e-07,
      "loss": 0.1646,
      "step": 13043
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3581006225710643,
      "learning_rate": 1.171033989558934e-07,
      "loss": 0.1754,
      "step": 13044
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3013411966835415,
      "learning_rate": 1.1685429039747065e-07,
      "loss": 0.1612,
      "step": 13045
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.478292729082162,
      "learning_rate": 1.1660544394690943e-07,
      "loss": 0.2076,
      "step": 13046
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2814085519438763,
      "learning_rate": 1.1635685961756737e-07,
      "loss": 0.1709,
      "step": 13047
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.445485746596271,
      "learning_rate": 1.1610853742278716e-07,
      "loss": 0.1568,
      "step": 13048
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2202765683122585,
      "learning_rate": 1.1586047737589811e-07,
      "loss": 0.1518,
      "step": 13049
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.314784589660231,
      "learning_rate": 1.156126794902157e-07,
      "loss": 0.1313,
      "step": 13050
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3029084909703432,
      "learning_rate": 1.1536514377903929e-07,
      "loss": 0.1637,
      "step": 13051
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4339914262765812,
      "learning_rate": 1.1511787025565657e-07,
      "loss": 0.1434,
      "step": 13052
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2844191853008895,
      "learning_rate": 1.1487085893333971e-07,
      "loss": 0.1312,
      "step": 13053
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.029903553624552,
      "learning_rate": 1.1462410982534755e-07,
      "loss": 0.6679,
      "step": 13054
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2381020716243722,
      "learning_rate": 1.1437762294492505e-07,
      "loss": 0.1691,
      "step": 13055
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.35257495098502,
      "learning_rate": 1.1413139830530218e-07,
      "loss": 0.1633,
      "step": 13056
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4669209733120736,
      "learning_rate": 1.1388543591969447e-07,
      "loss": 0.1659,
      "step": 13057
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.461701080708909,
      "learning_rate": 1.1363973580130583e-07,
      "loss": 0.172,
      "step": 13058
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5460519686090854,
      "learning_rate": 1.1339429796332346e-07,
      "loss": 0.166,
      "step": 13059
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.531885829681735,
      "learning_rate": 1.1314912241892184e-07,
      "loss": 0.1714,
      "step": 13060
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.808847205768016,
      "learning_rate": 1.12904209181261e-07,
      "loss": 0.5302,
      "step": 13061
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3035162116687118,
      "learning_rate": 1.1265955826348651e-07,
      "loss": 0.1274,
      "step": 13062
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.373753561739822,
      "learning_rate": 1.1241516967873123e-07,
      "loss": 0.1557,
      "step": 13063
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3210867360016143,
      "learning_rate": 1.1217104344011188e-07,
      "loss": 0.1594,
      "step": 13064
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2385621757311034,
      "learning_rate": 1.1192717956073185e-07,
      "loss": 0.1477,
      "step": 13065
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.167584155431501,
      "learning_rate": 1.1168357805368235e-07,
      "loss": 0.1785,
      "step": 13066
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3436188375363232,
      "learning_rate": 1.1144023893203793e-07,
      "loss": 0.1818,
      "step": 13067
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3352287358052948,
      "learning_rate": 1.1119716220886034e-07,
      "loss": 0.1747,
      "step": 13068
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4915024396218557,
      "learning_rate": 1.1095434789719694e-07,
      "loss": 0.1707,
      "step": 13069
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3193940317117399,
      "learning_rate": 1.107117960100812e-07,
      "loss": 0.1335,
      "step": 13070
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2944119861840084,
      "learning_rate": 1.1046950656053213e-07,
      "loss": 0.1559,
      "step": 13071
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3431899582036462,
      "learning_rate": 1.1022747956155432e-07,
      "loss": 0.152,
      "step": 13072
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3819211347768412,
      "learning_rate": 1.0998571502613964e-07,
      "loss": 0.1694,
      "step": 13073
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2411462118299283,
      "learning_rate": 1.0974421296726435e-07,
      "loss": 0.1435,
      "step": 13074
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.5324546334017,
      "learning_rate": 1.0950297339789252e-07,
      "loss": 0.5558,
      "step": 13075
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3834918229447388,
      "learning_rate": 1.0926199633097156e-07,
      "loss": 0.2221,
      "step": 13076
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.5197715092846,
      "learning_rate": 1.0902128177943783e-07,
      "loss": 0.4249,
      "step": 13077
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4913011682842394,
      "learning_rate": 1.0878082975620985e-07,
      "loss": 0.1902,
      "step": 13078
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3248632542578571,
      "learning_rate": 1.0854064027419564e-07,
      "loss": 0.1698,
      "step": 13079
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.030048043291224,
      "learning_rate": 1.0830071334628655e-07,
      "loss": 0.4913,
      "step": 13080
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5813138430132825,
      "learning_rate": 1.0806104898536229e-07,
      "loss": 0.1792,
      "step": 13081
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.378566173273803,
      "learning_rate": 1.0782164720428479e-07,
      "loss": 0.1338,
      "step": 13082
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.504559037991383,
      "learning_rate": 1.0758250801590709e-07,
      "loss": 0.2034,
      "step": 13083
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3848468434242853,
      "learning_rate": 1.0734363143306337e-07,
      "loss": 0.187,
      "step": 13084
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3329826609720354,
      "learning_rate": 1.071050174685756e-07,
      "loss": 0.1686,
      "step": 13085
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4426649824191622,
      "learning_rate": 1.0686666613525298e-07,
      "loss": 0.1842,
      "step": 13086
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.395489446400578,
      "learning_rate": 1.0662857744588751e-07,
      "loss": 0.1446,
      "step": 13087
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2792686249295209,
      "learning_rate": 1.0639075141326061e-07,
      "loss": 0.1193,
      "step": 13088
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3984795156823275,
      "learning_rate": 1.0615318805013596e-07,
      "loss": 0.1508,
      "step": 13089
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4382458851959075,
      "learning_rate": 1.0591588736926672e-07,
      "loss": 0.1755,
      "step": 13090
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3533279641194829,
      "learning_rate": 1.0567884938338879e-07,
      "loss": 0.1698,
      "step": 13091
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2702161509420813,
      "learning_rate": 1.0544207410522644e-07,
      "loss": 0.1773,
      "step": 13092
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5583326868147065,
      "learning_rate": 1.0520556154748785e-07,
      "loss": 0.1861,
      "step": 13093
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5483130044613096,
      "learning_rate": 1.0496931172287006e-07,
      "loss": 0.184,
      "step": 13094
    },
    {
      "epoch": 0.94,
      "grad_norm": 9.708388351396467,
      "learning_rate": 1.0473332464405184e-07,
      "loss": 0.6168,
      "step": 13095
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2889044943779253,
      "learning_rate": 1.0449760032370082e-07,
      "loss": 0.1496,
      "step": 13096
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4123455661993207,
      "learning_rate": 1.0426213877447022e-07,
      "loss": 0.2114,
      "step": 13097
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7067079570069033,
      "learning_rate": 1.0402694000899827e-07,
      "loss": 0.2053,
      "step": 13098
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3205238824992984,
      "learning_rate": 1.037920040399093e-07,
      "loss": 0.1385,
      "step": 13099
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6385177648235962,
      "learning_rate": 1.035573308798138e-07,
      "loss": 0.2219,
      "step": 13100
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3263431913695964,
      "learning_rate": 1.0332292054130777e-07,
      "loss": 0.1548,
      "step": 13101
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.159165278064142,
      "learning_rate": 1.0308877303697507e-07,
      "loss": 0.1278,
      "step": 13102
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.595434251741846,
      "learning_rate": 1.0285488837938174e-07,
      "loss": 0.5559,
      "step": 13103
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3752630909287211,
      "learning_rate": 1.0262126658108329e-07,
      "loss": 0.1631,
      "step": 13104
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2840145265315719,
      "learning_rate": 1.0238790765461858e-07,
      "loss": 0.1285,
      "step": 13105
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.107591290111134,
      "learning_rate": 1.0215481161251428e-07,
      "loss": 0.6798,
      "step": 13106
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3693171368534323,
      "learning_rate": 1.0192197846728147e-07,
      "loss": 0.2093,
      "step": 13107
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.277469947011179,
      "learning_rate": 1.0168940823141737e-07,
      "loss": 0.6557,
      "step": 13108
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3852171817117205,
      "learning_rate": 1.014571009174059e-07,
      "loss": 0.1811,
      "step": 13109
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1329181902241747,
      "learning_rate": 1.0122505653771708e-07,
      "loss": 0.1242,
      "step": 13110
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.353506512698889,
      "learning_rate": 1.0099327510480483e-07,
      "loss": 0.1876,
      "step": 13111
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4337532917646107,
      "learning_rate": 1.0076175663111087e-07,
      "loss": 0.1348,
      "step": 13112
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2477611070074937,
      "learning_rate": 1.0053050112906248e-07,
      "loss": 0.1321,
      "step": 13113
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3912910857740561,
      "learning_rate": 1.0029950861107196e-07,
      "loss": 0.1741,
      "step": 13114
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3512404767831534,
      "learning_rate": 1.0006877908953826e-07,
      "loss": 0.1386,
      "step": 13115
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.110986870385391,
      "learning_rate": 9.983831257684651e-08,
      "loss": 0.5485,
      "step": 13116
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5483293940956269,
      "learning_rate": 9.960810908536622e-08,
      "loss": 0.1896,
      "step": 13117
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.229605540318505,
      "learning_rate": 9.937816862745475e-08,
      "loss": 0.1407,
      "step": 13118
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3297474618040226,
      "learning_rate": 9.914849121545388e-08,
      "loss": 0.1843,
      "step": 13119
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5567899971390182,
      "learning_rate": 9.891907686169211e-08,
      "loss": 0.189,
      "step": 13120
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4312646734644974,
      "learning_rate": 9.868992557848401e-08,
      "loss": 0.1659,
      "step": 13121
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.341316879436319,
      "learning_rate": 9.846103737812807e-08,
      "loss": 0.1651,
      "step": 13122
    },
    {
      "epoch": 0.94,
      "grad_norm": 14.065110660810747,
      "learning_rate": 9.823241227291113e-08,
      "loss": 0.5843,
      "step": 13123
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.382805628650766,
      "learning_rate": 9.800405027510396e-08,
      "loss": 0.1682,
      "step": 13124
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.585446288251537,
      "learning_rate": 9.777595139696506e-08,
      "loss": 0.1742,
      "step": 13125
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2576152579774451,
      "learning_rate": 9.754811565073796e-08,
      "loss": 0.1685,
      "step": 13126
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4499602127577298,
      "learning_rate": 9.73205430486518e-08,
      "loss": 0.1851,
      "step": 13127
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4824911039300384,
      "learning_rate": 9.709323360292067e-08,
      "loss": 0.1733,
      "step": 13128
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3227094630150207,
      "learning_rate": 9.68661873257476e-08,
      "loss": 0.1767,
      "step": 13129
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2209886216560952,
      "learning_rate": 9.663940422931783e-08,
      "loss": 0.159,
      "step": 13130
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.337536080474091,
      "learning_rate": 9.641288432580498e-08,
      "loss": 0.1852,
      "step": 13131
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5227611830354462,
      "learning_rate": 9.618662762736764e-08,
      "loss": 0.1981,
      "step": 13132
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2972669414577704,
      "learning_rate": 9.596063414614998e-08,
      "loss": 0.1865,
      "step": 13133
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3156695845908088,
      "learning_rate": 9.573490389428286e-08,
      "loss": 0.1651,
      "step": 13134
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.357873390491458,
      "learning_rate": 9.550943688388325e-08,
      "loss": 0.2129,
      "step": 13135
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2288554871393065,
      "learning_rate": 9.528423312705148e-08,
      "loss": 0.1716,
      "step": 13136
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2303274837130422,
      "learning_rate": 9.505929263587733e-08,
      "loss": 0.1337,
      "step": 13137
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2889038268444117,
      "learning_rate": 9.483461542243388e-08,
      "loss": 0.1568,
      "step": 13138
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.521683357271531,
      "learning_rate": 9.461020149878153e-08,
      "loss": 0.1709,
      "step": 13139
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3005790355300526,
      "learning_rate": 9.43860508769645e-08,
      "loss": 0.189,
      "step": 13140
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2474637291205568,
      "learning_rate": 9.416216356901597e-08,
      "loss": 0.1988,
      "step": 13141
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4133639339308066,
      "learning_rate": 9.393853958695242e-08,
      "loss": 0.162,
      "step": 13142
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3785053020309725,
      "learning_rate": 9.37151789427776e-08,
      "loss": 0.1567,
      "step": 13143
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1666104222537332,
      "learning_rate": 9.349208164847967e-08,
      "loss": 0.1444,
      "step": 13144
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3307109058678352,
      "learning_rate": 9.326924771603518e-08,
      "loss": 0.1706,
      "step": 13145
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.6088935611880295,
      "learning_rate": 9.304667715740345e-08,
      "loss": 0.6629,
      "step": 13146
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3319720250609979,
      "learning_rate": 9.282436998453215e-08,
      "loss": 0.1507,
      "step": 13147
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4114046449541982,
      "learning_rate": 9.26023262093534e-08,
      "loss": 0.2,
      "step": 13148
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2603244406999339,
      "learning_rate": 9.238054584378541e-08,
      "loss": 0.1785,
      "step": 13149
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4976525073715505,
      "learning_rate": 9.21590288997326e-08,
      "loss": 0.1987,
      "step": 13150
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2864213327436262,
      "learning_rate": 9.193777538908599e-08,
      "loss": 0.1713,
      "step": 13151
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2462573364831886,
      "learning_rate": 9.171678532372052e-08,
      "loss": 0.1589,
      "step": 13152
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.26175273957759,
      "learning_rate": 9.149605871549838e-08,
      "loss": 0.1415,
      "step": 13153
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1182522289505583,
      "learning_rate": 9.127559557626786e-08,
      "loss": 0.1251,
      "step": 13154
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.904515872118644,
      "learning_rate": 9.105539591786172e-08,
      "loss": 0.5161,
      "step": 13155
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.357110053473666,
      "learning_rate": 9.083545975209995e-08,
      "loss": 0.1946,
      "step": 13156
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.772578343180989,
      "learning_rate": 9.0615787090787e-08,
      "loss": 0.402,
      "step": 13157
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1188050197733188,
      "learning_rate": 9.03963779457151e-08,
      "loss": 0.1253,
      "step": 13158
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4707379574131976,
      "learning_rate": 9.017723232866039e-08,
      "loss": 0.1639,
      "step": 13159
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2273041047292015,
      "learning_rate": 8.995835025138677e-08,
      "loss": 0.1577,
      "step": 13160
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0926319135109355,
      "learning_rate": 8.97397317256421e-08,
      "loss": 0.1207,
      "step": 13161
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3867029982926145,
      "learning_rate": 8.952137676316141e-08,
      "loss": 0.1692,
      "step": 13162
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4240445258121894,
      "learning_rate": 8.93032853756648e-08,
      "loss": 0.1505,
      "step": 13163
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4157623775107033,
      "learning_rate": 8.908545757485843e-08,
      "loss": 0.1938,
      "step": 13164
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2080908690891115,
      "learning_rate": 8.886789337243462e-08,
      "loss": 0.1409,
      "step": 13165
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2990010372879015,
      "learning_rate": 8.865059278007182e-08,
      "loss": 0.1402,
      "step": 13166
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2229264905357322,
      "learning_rate": 8.843355580943291e-08,
      "loss": 0.1613,
      "step": 13167
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2919422351630574,
      "learning_rate": 8.821678247216858e-08,
      "loss": 0.1653,
      "step": 13168
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.720174039662592,
      "learning_rate": 8.800027277991341e-08,
      "loss": 0.7503,
      "step": 13169
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2591293754937085,
      "learning_rate": 8.778402674429031e-08,
      "loss": 0.1202,
      "step": 13170
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6820175215716104,
      "learning_rate": 8.756804437690447e-08,
      "loss": 0.1728,
      "step": 13171
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2476043944204145,
      "learning_rate": 8.735232568935049e-08,
      "loss": 0.1819,
      "step": 13172
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.28675531022988,
      "learning_rate": 8.713687069320631e-08,
      "loss": 0.1468,
      "step": 13173
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3475176998565799,
      "learning_rate": 8.692167940003715e-08,
      "loss": 0.1872,
      "step": 13174
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1563297716167171,
      "learning_rate": 8.670675182139321e-08,
      "loss": 0.1285,
      "step": 13175
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2527017363892579,
      "learning_rate": 8.649208796881248e-08,
      "loss": 0.1331,
      "step": 13176
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.464795728271567,
      "learning_rate": 8.627768785381518e-08,
      "loss": 0.1875,
      "step": 13177
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3890714342467665,
      "learning_rate": 8.606355148791101e-08,
      "loss": 0.1699,
      "step": 13178
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3710217749381677,
      "learning_rate": 8.584967888259299e-08,
      "loss": 0.2052,
      "step": 13179
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.424173515550315,
      "learning_rate": 8.563607004934193e-08,
      "loss": 0.1825,
      "step": 13180
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3525908163031133,
      "learning_rate": 8.5422724999622e-08,
      "loss": 0.1731,
      "step": 13181
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3889769580591278,
      "learning_rate": 8.520964374488627e-08,
      "loss": 0.1657,
      "step": 13182
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3917039236269346,
      "learning_rate": 8.499682629657114e-08,
      "loss": 0.1768,
      "step": 13183
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2745733567079207,
      "learning_rate": 8.478427266609968e-08,
      "loss": 0.1807,
      "step": 13184
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2221171778586912,
      "learning_rate": 8.457198286488222e-08,
      "loss": 0.1639,
      "step": 13185
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2338872376690642,
      "learning_rate": 8.435995690431242e-08,
      "loss": 0.1546,
      "step": 13186
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3188212572258886,
      "learning_rate": 8.414819479577118e-08,
      "loss": 0.1622,
      "step": 13187
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.379511274167154,
      "learning_rate": 8.393669655062553e-08,
      "loss": 0.124,
      "step": 13188
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5580262105535592,
      "learning_rate": 8.372546218022747e-08,
      "loss": 0.1702,
      "step": 13189
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.570275028084891,
      "learning_rate": 8.351449169591463e-08,
      "loss": 0.616,
      "step": 13190
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3419742066889624,
      "learning_rate": 8.330378510901238e-08,
      "loss": 0.1589,
      "step": 13191
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4778512504421364,
      "learning_rate": 8.309334243083e-08,
      "loss": 0.1426,
      "step": 13192
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.268436198876092,
      "learning_rate": 8.288316367266291e-08,
      "loss": 0.1395,
      "step": 13193
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3528786332494176,
      "learning_rate": 8.267324884579264e-08,
      "loss": 0.1865,
      "step": 13194
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2932904597220947,
      "learning_rate": 8.24635979614874e-08,
      "loss": 0.137,
      "step": 13195
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.411943736392523,
      "learning_rate": 8.225421103099929e-08,
      "loss": 0.5426,
      "step": 13196
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.414172526109825,
      "learning_rate": 8.204508806556877e-08,
      "loss": 0.5679,
      "step": 13197
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3660109446424185,
      "learning_rate": 8.183622907641963e-08,
      "loss": 0.1765,
      "step": 13198
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3240350017822546,
      "learning_rate": 8.162763407476293e-08,
      "loss": 0.1685,
      "step": 13199
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2602024284896012,
      "learning_rate": 8.141930307179468e-08,
      "loss": 0.1734,
      "step": 13200
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4054850099068794,
      "learning_rate": 8.121123607869818e-08,
      "loss": 0.1495,
      "step": 13201
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3272310905770894,
      "learning_rate": 8.100343310664061e-08,
      "loss": 0.1457,
      "step": 13202
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4742246384552529,
      "learning_rate": 8.079589416677747e-08,
      "loss": 0.1761,
      "step": 13203
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3407467158577455,
      "learning_rate": 8.058861927024708e-08,
      "loss": 0.1699,
      "step": 13204
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2833551866324244,
      "learning_rate": 8.038160842817666e-08,
      "loss": 0.1673,
      "step": 13205
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4261462553203497,
      "learning_rate": 8.017486165167621e-08,
      "loss": 0.1407,
      "step": 13206
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1562713087317606,
      "learning_rate": 7.996837895184406e-08,
      "loss": 0.1265,
      "step": 13207
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4636628870217436,
      "learning_rate": 7.976216033976303e-08,
      "loss": 0.1919,
      "step": 13208
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3401967494704117,
      "learning_rate": 7.955620582650204e-08,
      "loss": 0.1413,
      "step": 13209
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2935427297686013,
      "learning_rate": 7.935051542311556e-08,
      "loss": 0.1695,
      "step": 13210
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5016092546494715,
      "learning_rate": 7.91450891406459e-08,
      "loss": 0.1721,
      "step": 13211
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4751814581786167,
      "learning_rate": 7.893992699011754e-08,
      "loss": 0.176,
      "step": 13212
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.339140025214084,
      "learning_rate": 7.873502898254337e-08,
      "loss": 0.1601,
      "step": 13213
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2450203871782686,
      "learning_rate": 7.853039512892235e-08,
      "loss": 0.1422,
      "step": 13214
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.862964689344508,
      "learning_rate": 7.832602544023737e-08,
      "loss": 0.6403,
      "step": 13215
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4564716105323856,
      "learning_rate": 7.812191992745854e-08,
      "loss": 0.1768,
      "step": 13216
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.284168528774165,
      "learning_rate": 7.791807860154155e-08,
      "loss": 0.1395,
      "step": 13217
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0470929117950873,
      "learning_rate": 7.771450147342763e-08,
      "loss": 0.1026,
      "step": 13218
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4016408515146388,
      "learning_rate": 7.751118855404416e-08,
      "loss": 0.206,
      "step": 13219
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3856990364992687,
      "learning_rate": 7.730813985430407e-08,
      "loss": 0.1543,
      "step": 13220
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4213896760697655,
      "learning_rate": 7.710535538510589e-08,
      "loss": 0.2111,
      "step": 13221
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.012461332937578,
      "learning_rate": 7.690283515733532e-08,
      "loss": 0.5598,
      "step": 13222
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.560209140015982,
      "learning_rate": 7.670057918186147e-08,
      "loss": 0.6105,
      "step": 13223
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3324723368649825,
      "learning_rate": 7.649858746954176e-08,
      "loss": 0.1433,
      "step": 13224
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4768086776168086,
      "learning_rate": 7.62968600312175e-08,
      "loss": 0.1561,
      "step": 13225
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2798724942726052,
      "learning_rate": 7.609539687771727e-08,
      "loss": 0.1304,
      "step": 13226
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7855314924497487,
      "learning_rate": 7.589419801985409e-08,
      "loss": 0.1865,
      "step": 13227
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.296955319019752,
      "learning_rate": 7.569326346842876e-08,
      "loss": 0.1718,
      "step": 13228
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4851335929835605,
      "learning_rate": 7.549259323422486e-08,
      "loss": 0.1884,
      "step": 13229
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3677643098864833,
      "learning_rate": 7.529218732801491e-08,
      "loss": 0.1738,
      "step": 13230
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2484973443639664,
      "learning_rate": 7.509204576055529e-08,
      "loss": 0.1725,
      "step": 13231
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2846549468903667,
      "learning_rate": 7.489216854258963e-08,
      "loss": 0.1445,
      "step": 13232
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2658728955844833,
      "learning_rate": 7.469255568484545e-08,
      "loss": 0.1443,
      "step": 13233
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2605448486544473,
      "learning_rate": 7.449320719803755e-08,
      "loss": 0.2019,
      "step": 13234
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.429345167294272,
      "learning_rate": 7.429412309286676e-08,
      "loss": 0.1707,
      "step": 13235
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.49159151752236,
      "learning_rate": 7.409530338001846e-08,
      "loss": 0.1815,
      "step": 13236
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2420970692119704,
      "learning_rate": 7.389674807016466e-08,
      "loss": 0.1207,
      "step": 13237
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3931910733036952,
      "learning_rate": 7.369845717396296e-08,
      "loss": 0.1632,
      "step": 13238
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3788759808527267,
      "learning_rate": 7.350043070205704e-08,
      "loss": 0.2028,
      "step": 13239
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3691547178452361,
      "learning_rate": 7.330266866507618e-08,
      "loss": 0.1773,
      "step": 13240
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1552786896711713,
      "learning_rate": 7.310517107363524e-08,
      "loss": 0.14,
      "step": 13241
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1923613077957342,
      "learning_rate": 7.290793793833517e-08,
      "loss": 0.1523,
      "step": 13242
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.6902423487841105,
      "learning_rate": 7.271096926976251e-08,
      "loss": 0.5808,
      "step": 13243
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3509350546262486,
      "learning_rate": 7.25142650784899e-08,
      "loss": 0.1672,
      "step": 13244
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4719310867624726,
      "learning_rate": 7.231782537507614e-08,
      "loss": 0.1538,
      "step": 13245
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4686892159235314,
      "learning_rate": 7.212165017006445e-08,
      "loss": 0.1893,
      "step": 13246
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4935687993719293,
      "learning_rate": 7.192573947398529e-08,
      "loss": 0.203,
      "step": 13247
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3848343846267883,
      "learning_rate": 7.173009329735415e-08,
      "loss": 0.1637,
      "step": 13248
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.460053041703305,
      "learning_rate": 7.153471165067316e-08,
      "loss": 0.1717,
      "step": 13249
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4102758891736003,
      "learning_rate": 7.13395945444284e-08,
      "loss": 0.183,
      "step": 13250
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3119573478711892,
      "learning_rate": 7.114474198909372e-08,
      "loss": 0.151,
      "step": 13251
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.2454437782653445,
      "learning_rate": 7.095015399512795e-08,
      "loss": 0.6151,
      "step": 13252
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4925098994958246,
      "learning_rate": 7.075583057297664e-08,
      "loss": 0.1974,
      "step": 13253
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2899647110576729,
      "learning_rate": 7.056177173306866e-08,
      "loss": 0.1221,
      "step": 13254
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6524785038332643,
      "learning_rate": 7.03679774858218e-08,
      "loss": 0.206,
      "step": 13255
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4669521814104454,
      "learning_rate": 7.017444784163663e-08,
      "loss": 0.1706,
      "step": 13256
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2481972039563185,
      "learning_rate": 6.998118281090316e-08,
      "loss": 0.1214,
      "step": 13257
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.371040592234891,
      "learning_rate": 6.978818240399309e-08,
      "loss": 0.1599,
      "step": 13258
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3964108047019743,
      "learning_rate": 6.95954466312665e-08,
      "loss": 0.1815,
      "step": 13259
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4784656137991061,
      "learning_rate": 6.940297550306895e-08,
      "loss": 0.1566,
      "step": 13260
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.18117714834306,
      "learning_rate": 6.921076902973167e-08,
      "loss": 0.1418,
      "step": 13261
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.238292358755321,
      "learning_rate": 6.901882722157138e-08,
      "loss": 0.1704,
      "step": 13262
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.43456800477493,
      "learning_rate": 6.882715008889096e-08,
      "loss": 0.1551,
      "step": 13263
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4521725723769963,
      "learning_rate": 6.863573764197828e-08,
      "loss": 0.1766,
      "step": 13264
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2621969287598642,
      "learning_rate": 6.844458989110792e-08,
      "loss": 0.1443,
      "step": 13265
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.328167882481714,
      "learning_rate": 6.825370684653998e-08,
      "loss": 0.1896,
      "step": 13266
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3201820530043873,
      "learning_rate": 6.806308851852017e-08,
      "loss": 0.1502,
      "step": 13267
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2808754072918478,
      "learning_rate": 6.78727349172803e-08,
      "loss": 0.184,
      "step": 13268
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3209254841369964,
      "learning_rate": 6.768264605303776e-08,
      "loss": 0.1727,
      "step": 13269
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.572085168199533,
      "learning_rate": 6.749282193599549e-08,
      "loss": 0.1663,
      "step": 13270
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.459129359779204,
      "learning_rate": 6.730326257634312e-08,
      "loss": 0.182,
      "step": 13271
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2546965359437803,
      "learning_rate": 6.711396798425473e-08,
      "loss": 0.1728,
      "step": 13272
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6376378247208394,
      "learning_rate": 6.692493816989165e-08,
      "loss": 0.2236,
      "step": 13273
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.4782200603387725,
      "learning_rate": 6.673617314339964e-08,
      "loss": 0.3883,
      "step": 13274
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4372954261811717,
      "learning_rate": 6.65476729149106e-08,
      "loss": 0.1601,
      "step": 13275
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3228751621797696,
      "learning_rate": 6.63594374945431e-08,
      "loss": 0.1667,
      "step": 13276
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.377892444882691,
      "learning_rate": 6.617146689240072e-08,
      "loss": 0.1839,
      "step": 13277
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.976187994249105,
      "learning_rate": 6.598376111857319e-08,
      "loss": 0.694,
      "step": 13278
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.115276875479222,
      "learning_rate": 6.579632018313519e-08,
      "loss": 0.6909,
      "step": 13279
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3358222860011357,
      "learning_rate": 6.560914409614872e-08,
      "loss": 0.1642,
      "step": 13280
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4880771477728123,
      "learning_rate": 6.542223286765959e-08,
      "loss": 0.1902,
      "step": 13281
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.321907913587343,
      "learning_rate": 6.523558650770146e-08,
      "loss": 0.1681,
      "step": 13282
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3677737122692795,
      "learning_rate": 6.504920502629187e-08,
      "loss": 0.1463,
      "step": 13283
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5141921611143105,
      "learning_rate": 6.486308843343558e-08,
      "loss": 0.1935,
      "step": 13284
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3318243542536914,
      "learning_rate": 6.467723673912185e-08,
      "loss": 0.1732,
      "step": 13285
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3120647066274833,
      "learning_rate": 6.449164995332768e-08,
      "loss": 0.1641,
      "step": 13286
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4009255292805334,
      "learning_rate": 6.4306328086014e-08,
      "loss": 0.1521,
      "step": 13287
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.48578473007008,
      "learning_rate": 6.41212711471284e-08,
      "loss": 0.1548,
      "step": 13288
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.358401053210439,
      "learning_rate": 6.393647914660294e-08,
      "loss": 0.1492,
      "step": 13289
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3250508821516083,
      "learning_rate": 6.375195209435803e-08,
      "loss": 0.1689,
      "step": 13290
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4656157093677809,
      "learning_rate": 6.356769000029739e-08,
      "loss": 0.1903,
      "step": 13291
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3818822044955037,
      "learning_rate": 6.3383692874312e-08,
      "loss": 0.2122,
      "step": 13292
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4730473651680245,
      "learning_rate": 6.319996072627732e-08,
      "loss": 0.1918,
      "step": 13293
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2674267147016414,
      "learning_rate": 6.301649356605654e-08,
      "loss": 0.157,
      "step": 13294
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1830084310265918,
      "learning_rate": 6.283329140349626e-08,
      "loss": 0.1293,
      "step": 13295
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4218508529157183,
      "learning_rate": 6.26503542484308e-08,
      "loss": 0.164,
      "step": 13296
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.222018035693397,
      "learning_rate": 6.2467682110679e-08,
      "loss": 0.5788,
      "step": 13297
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2132260342523824,
      "learning_rate": 6.228527500004689e-08,
      "loss": 0.145,
      "step": 13298
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2614549166426727,
      "learning_rate": 6.210313292632386e-08,
      "loss": 0.1696,
      "step": 13299
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5315587375111008,
      "learning_rate": 6.192125589928821e-08,
      "loss": 0.1832,
      "step": 13300
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.401038314734205,
      "learning_rate": 6.173964392870102e-08,
      "loss": 0.1512,
      "step": 13301
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.21109833931684,
      "learning_rate": 6.15582970243117e-08,
      "loss": 0.1685,
      "step": 13302
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.297054989176573,
      "learning_rate": 6.137721519585304e-08,
      "loss": 0.192,
      "step": 13303
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.395650862574453,
      "learning_rate": 6.119639845304559e-08,
      "loss": 0.579,
      "step": 13304
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.147773363851346,
      "learning_rate": 6.101584680559435e-08,
      "loss": 0.4924,
      "step": 13305
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3199385825542405,
      "learning_rate": 6.08355602631916e-08,
      "loss": 0.1552,
      "step": 13306
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5480018070497346,
      "learning_rate": 6.065553883551289e-08,
      "loss": 0.1963,
      "step": 13307
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1936634452362769,
      "learning_rate": 6.047578253222274e-08,
      "loss": 0.1656,
      "step": 13308
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3050289600060894,
      "learning_rate": 6.029629136296844e-08,
      "loss": 0.1844,
      "step": 13309
    },
    {
      "epoch": 0.95,
      "grad_norm": 10.498520030951294,
      "learning_rate": 6.011706533738448e-08,
      "loss": 0.4027,
      "step": 13310
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4907994128497006,
      "learning_rate": 5.993810446509152e-08,
      "loss": 0.1937,
      "step": 13311
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.439007954762212,
      "learning_rate": 5.97594087556952e-08,
      "loss": 0.6615,
      "step": 13312
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.433832491403932,
      "learning_rate": 5.958097821878783e-08,
      "loss": 0.1299,
      "step": 13313
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.750590433012544,
      "learning_rate": 5.940281286394622e-08,
      "loss": 0.2214,
      "step": 13314
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.326044406318544,
      "learning_rate": 5.922491270073327e-08,
      "loss": 0.2027,
      "step": 13315
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2576458229959064,
      "learning_rate": 5.904727773869856e-08,
      "loss": 0.1369,
      "step": 13316
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4481887572809524,
      "learning_rate": 5.886990798737668e-08,
      "loss": 0.1642,
      "step": 13317
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.331185790914141,
      "learning_rate": 5.86928034562878e-08,
      "loss": 0.1453,
      "step": 13318
    },
    {
      "epoch": 0.95,
      "grad_norm": 9.856244889250446,
      "learning_rate": 5.8515964154938765e-08,
      "loss": 0.6249,
      "step": 13319
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3697926286046078,
      "learning_rate": 5.833939009282086e-08,
      "loss": 0.1739,
      "step": 13320
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.121631552967087,
      "learning_rate": 5.816308127941317e-08,
      "loss": 0.5855,
      "step": 13321
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.380012654949571,
      "learning_rate": 5.798703772417758e-08,
      "loss": 0.1774,
      "step": 13322
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5785816694601777,
      "learning_rate": 5.781125943656485e-08,
      "loss": 0.1981,
      "step": 13323
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.251334898223548,
      "learning_rate": 5.7635746426008556e-08,
      "loss": 0.1487,
      "step": 13324
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3370060453185597,
      "learning_rate": 5.7460498701931154e-08,
      "loss": 0.1528,
      "step": 13325
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0664712674479397,
      "learning_rate": 5.728551627373846e-08,
      "loss": 0.1421,
      "step": 13326
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.256741311145682,
      "learning_rate": 5.71107991508224e-08,
      "loss": 0.1626,
      "step": 13327
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2580714492004215,
      "learning_rate": 5.693634734256215e-08,
      "loss": 0.1495,
      "step": 13328
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4378931754320057,
      "learning_rate": 5.676216085832076e-08,
      "loss": 0.1779,
      "step": 13329
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2078462543445339,
      "learning_rate": 5.6588239707447976e-08,
      "loss": 0.1348,
      "step": 13330
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3533986303394667,
      "learning_rate": 5.6414583899279675e-08,
      "loss": 0.1555,
      "step": 13331
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2627942575878692,
      "learning_rate": 5.624119344313672e-08,
      "loss": 0.1382,
      "step": 13332
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4373704264141196,
      "learning_rate": 5.606806834832612e-08,
      "loss": 0.1825,
      "step": 13333
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3456351830861386,
      "learning_rate": 5.5895208624139885e-08,
      "loss": 0.1834,
      "step": 13334
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3485527747737822,
      "learning_rate": 5.5722614279857255e-08,
      "loss": 0.1686,
      "step": 13335
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2105391216883623,
      "learning_rate": 5.5550285324741936e-08,
      "loss": 0.157,
      "step": 13336
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2652996262811467,
      "learning_rate": 5.53782217680443e-08,
      "loss": 0.1647,
      "step": 13337
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3463600148166348,
      "learning_rate": 5.520642361899975e-08,
      "loss": 0.1849,
      "step": 13338
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2036475762872985,
      "learning_rate": 5.5034890886829785e-08,
      "loss": 0.1313,
      "step": 13339
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3399224061386756,
      "learning_rate": 5.486362358074093e-08,
      "loss": 0.1678,
      "step": 13340
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2092158749392885,
      "learning_rate": 5.469262170992751e-08,
      "loss": 0.1283,
      "step": 13341
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3136108378510918,
      "learning_rate": 5.4521885283567165e-08,
      "loss": 0.1535,
      "step": 13342
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3558511872621661,
      "learning_rate": 5.43514143108248e-08,
      "loss": 0.1786,
      "step": 13343
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2285151000405505,
      "learning_rate": 5.4181208800850316e-08,
      "loss": 0.1393,
      "step": 13344
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2208502706003217,
      "learning_rate": 5.4011268762780286e-08,
      "loss": 0.1778,
      "step": 13345
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2159366049961524,
      "learning_rate": 5.3841594205735756e-08,
      "loss": 0.1491,
      "step": 13346
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.52271175688257,
      "learning_rate": 5.367218513882444e-08,
      "loss": 0.2022,
      "step": 13347
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5162390871495097,
      "learning_rate": 5.3503041571139615e-08,
      "loss": 0.1809,
      "step": 13348
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3098647538696273,
      "learning_rate": 5.3334163511759574e-08,
      "loss": 0.1349,
      "step": 13349
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.357042284406413,
      "learning_rate": 5.31655509697504e-08,
      "loss": 0.1542,
      "step": 13350
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.37228549287707,
      "learning_rate": 5.299720395416097e-08,
      "loss": 0.1965,
      "step": 13351
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.073265668477661,
      "learning_rate": 5.282912247402849e-08,
      "loss": 0.4443,
      "step": 13352
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3773165249115564,
      "learning_rate": 5.2661306538374647e-08,
      "loss": 0.1505,
      "step": 13353
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.9919602453498335,
      "learning_rate": 5.2493756156207224e-08,
      "loss": 0.5429,
      "step": 13354
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.105664864594872,
      "learning_rate": 5.232647133651958e-08,
      "loss": 0.5636,
      "step": 13355
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3579666109855832,
      "learning_rate": 5.2159452088290654e-08,
      "loss": 0.1699,
      "step": 13356
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3171739091207284,
      "learning_rate": 5.1992698420486045e-08,
      "loss": 0.1551,
      "step": 13357
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3148577970301574,
      "learning_rate": 5.182621034205582e-08,
      "loss": 0.1669,
      "step": 13358
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3866526739992016,
      "learning_rate": 5.1659987861936154e-08,
      "loss": 0.1748,
      "step": 13359
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2428152545721247,
      "learning_rate": 5.1494030989049926e-08,
      "loss": 0.1368,
      "step": 13360
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2402516349793913,
      "learning_rate": 5.1328339732304446e-08,
      "loss": 0.1662,
      "step": 13361
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4311569486596438,
      "learning_rate": 5.116291410059426e-08,
      "loss": 0.1553,
      "step": 13362
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2792495997952833,
      "learning_rate": 5.0997754102797836e-08,
      "loss": 0.1589,
      "step": 13363
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3390687671844013,
      "learning_rate": 5.083285974778085e-08,
      "loss": 0.181,
      "step": 13364
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2775621884755708,
      "learning_rate": 5.0668231044394e-08,
      "loss": 0.1754,
      "step": 13365
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3059576246723856,
      "learning_rate": 5.050386800147356e-08,
      "loss": 0.1812,
      "step": 13366
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.310051122957858,
      "learning_rate": 5.033977062784245e-08,
      "loss": 0.1567,
      "step": 13367
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2069005617167312,
      "learning_rate": 5.01759389323081e-08,
      "loss": 0.1601,
      "step": 13368
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2388821618341201,
      "learning_rate": 5.0012372923664565e-08,
      "loss": 0.1657,
      "step": 13369
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.383510345278049,
      "learning_rate": 4.9849072610692605e-08,
      "loss": 0.1436,
      "step": 13370
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.3380576104167705,
      "learning_rate": 4.968603800215577e-08,
      "loss": 0.6283,
      "step": 13371
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.690474436780941,
      "learning_rate": 4.952326910680594e-08,
      "loss": 0.555,
      "step": 13372
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3003011457676212,
      "learning_rate": 4.936076593338002e-08,
      "loss": 0.1753,
      "step": 13373
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4804120316538376,
      "learning_rate": 4.919852849059992e-08,
      "loss": 0.177,
      "step": 13374
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2709296202328781,
      "learning_rate": 4.9036556787174785e-08,
      "loss": 0.1533,
      "step": 13375
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3585750777153645,
      "learning_rate": 4.887485083179766e-08,
      "loss": 0.1773,
      "step": 13376
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4202582383668876,
      "learning_rate": 4.871341063314883e-08,
      "loss": 0.1625,
      "step": 13377
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3141249428230466,
      "learning_rate": 4.855223619989413e-08,
      "loss": 0.1478,
      "step": 13378
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5275567644878913,
      "learning_rate": 4.839132754068443e-08,
      "loss": 0.2305,
      "step": 13379
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4363399500901421,
      "learning_rate": 4.823068466415615e-08,
      "loss": 0.1512,
      "step": 13380
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3632251100023647,
      "learning_rate": 4.807030757893238e-08,
      "loss": 0.1442,
      "step": 13381
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2103501576223712,
      "learning_rate": 4.7910196293621814e-08,
      "loss": 0.1329,
      "step": 13382
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5332606217612996,
      "learning_rate": 4.775035081681812e-08,
      "loss": 0.1703,
      "step": 13383
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.340850258744859,
      "learning_rate": 4.759077115710109e-08,
      "loss": 0.1649,
      "step": 13384
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3767552298241532,
      "learning_rate": 4.743145732303722e-08,
      "loss": 0.1717,
      "step": 13385
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2609956949348624,
      "learning_rate": 4.7272409323176894e-08,
      "loss": 0.1476,
      "step": 13386
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3904608078504561,
      "learning_rate": 4.711362716605716e-08,
      "loss": 0.1699,
      "step": 13387
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.552395597100296,
      "learning_rate": 4.6955110860201214e-08,
      "loss": 0.1706,
      "step": 13388
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3650201046552604,
      "learning_rate": 4.679686041411835e-08,
      "loss": 0.1568,
      "step": 13389
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4027235500671005,
      "learning_rate": 4.663887583630122e-08,
      "loss": 0.1782,
      "step": 13390
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.145779485786868,
      "learning_rate": 4.6481157135230827e-08,
      "loss": 0.1399,
      "step": 13391
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3665753861588434,
      "learning_rate": 4.632370431937261e-08,
      "loss": 0.1983,
      "step": 13392
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.257564542521957,
      "learning_rate": 4.6166517397178144e-08,
      "loss": 0.1528,
      "step": 13393
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.472840982699897,
      "learning_rate": 4.6009596377084023e-08,
      "loss": 0.157,
      "step": 13394
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1384737356892514,
      "learning_rate": 4.58529412675135e-08,
      "loss": 0.1554,
      "step": 13395
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.32307034310547,
      "learning_rate": 4.5696552076875954e-08,
      "loss": 0.1938,
      "step": 13396
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.136596042085612,
      "learning_rate": 4.5540428813564684e-08,
      "loss": 0.1231,
      "step": 13397
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.264315099096542,
      "learning_rate": 4.5384571485960204e-08,
      "loss": 0.1546,
      "step": 13398
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5468919884633898,
      "learning_rate": 4.52289801024286e-08,
      "loss": 0.2084,
      "step": 13399
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2726377104857365,
      "learning_rate": 4.5073654671320965e-08,
      "loss": 0.158,
      "step": 13400
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.964969621632054,
      "learning_rate": 4.491859520097452e-08,
      "loss": 0.589,
      "step": 13401
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.19618936197148,
      "learning_rate": 4.476380169971262e-08,
      "loss": 0.1455,
      "step": 13402
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1565798383005856,
      "learning_rate": 4.46092741758436e-08,
      "loss": 0.1488,
      "step": 13403
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3095333949218393,
      "learning_rate": 4.445501263766194e-08,
      "loss": 0.1803,
      "step": 13404
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3758516805632601,
      "learning_rate": 4.4301017093447695e-08,
      "loss": 0.1738,
      "step": 13405
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3167378257075941,
      "learning_rate": 4.414728755146702e-08,
      "loss": 0.1422,
      "step": 13406
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5680713710979421,
      "learning_rate": 4.3993824019971654e-08,
      "loss": 0.1826,
      "step": 13407
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4868944384571994,
      "learning_rate": 4.3840626507198334e-08,
      "loss": 0.168,
      "step": 13408
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2962314810006972,
      "learning_rate": 4.368769502137049e-08,
      "loss": 0.1582,
      "step": 13409
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4072354396182396,
      "learning_rate": 4.353502957069711e-08,
      "loss": 0.1316,
      "step": 13410
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.156987034701498,
      "learning_rate": 4.338263016337219e-08,
      "loss": 0.1236,
      "step": 13411
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.132235860517753,
      "learning_rate": 4.323049680757585e-08,
      "loss": 0.3685,
      "step": 13412
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3470456979430232,
      "learning_rate": 4.307862951147434e-08,
      "loss": 0.1579,
      "step": 13413
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4361559686875618,
      "learning_rate": 4.292702828321948e-08,
      "loss": 0.1252,
      "step": 13414
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4610914571945541,
      "learning_rate": 4.2775693130948094e-08,
      "loss": 0.1861,
      "step": 13415
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2594491982909732,
      "learning_rate": 4.262462406278367e-08,
      "loss": 0.1574,
      "step": 13416
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4964065089956797,
      "learning_rate": 4.247382108683418e-08,
      "loss": 0.1889,
      "step": 13417
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4175671685548346,
      "learning_rate": 4.232328421119591e-08,
      "loss": 0.171,
      "step": 13418
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4713696020562725,
      "learning_rate": 4.2173013443946866e-08,
      "loss": 0.1568,
      "step": 13419
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3120353953799364,
      "learning_rate": 4.202300879315446e-08,
      "loss": 0.1384,
      "step": 13420
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1979885422368324,
      "learning_rate": 4.1873270266870046e-08,
      "loss": 0.1246,
      "step": 13421
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2084865364089519,
      "learning_rate": 4.1723797873130524e-08,
      "loss": 0.1445,
      "step": 13422
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.106882124440699,
      "learning_rate": 4.1574591619959473e-08,
      "loss": 0.4468,
      "step": 13423
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3132370954953507,
      "learning_rate": 4.142565151536604e-08,
      "loss": 0.163,
      "step": 13424
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3737600751147108,
      "learning_rate": 4.1276977567343835e-08,
      "loss": 0.1542,
      "step": 13425
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4480994235103994,
      "learning_rate": 4.112856978387369e-08,
      "loss": 0.1643,
      "step": 13426
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3671886661168962,
      "learning_rate": 4.098042817292147e-08,
      "loss": 0.182,
      "step": 13427
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3451970784005587,
      "learning_rate": 4.083255274243858e-08,
      "loss": 0.1605,
      "step": 13428
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3075391320870409,
      "learning_rate": 4.0684943500362004e-08,
      "loss": 0.1493,
      "step": 13429
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4190269636146942,
      "learning_rate": 4.0537600454615964e-08,
      "loss": 0.1766,
      "step": 13430
    },
    {
      "epoch": 0.96,
      "grad_norm": 9.455965064516791,
      "learning_rate": 4.039052361310858e-08,
      "loss": 0.5239,
      "step": 13431
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3762837000499397,
      "learning_rate": 4.024371298373464e-08,
      "loss": 0.1615,
      "step": 13432
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.311405070834158,
      "learning_rate": 4.0097168574373404e-08,
      "loss": 0.1694,
      "step": 13433
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.647325873059729,
      "learning_rate": 3.9950890392892463e-08,
      "loss": 0.4556,
      "step": 13434
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.195812367605104,
      "learning_rate": 3.980487844714165e-08,
      "loss": 0.1372,
      "step": 13435
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.21639358457569,
      "learning_rate": 3.965913274495914e-08,
      "loss": 0.1578,
      "step": 13436
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5475300240448313,
      "learning_rate": 3.951365329416868e-08,
      "loss": 0.2478,
      "step": 13437
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3909560837744872,
      "learning_rate": 3.9368440102577344e-08,
      "loss": 0.1491,
      "step": 13438
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3906920920374972,
      "learning_rate": 3.922349317798169e-08,
      "loss": 0.1446,
      "step": 13439
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.407279423094399,
      "learning_rate": 3.907881252816048e-08,
      "loss": 0.623,
      "step": 13440
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2883523489273276,
      "learning_rate": 3.893439816087974e-08,
      "loss": 0.1564,
      "step": 13441
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.325358919754688,
      "learning_rate": 3.8790250083891035e-08,
      "loss": 0.2085,
      "step": 13442
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4168206996971229,
      "learning_rate": 3.864636830493263e-08,
      "loss": 0.1819,
      "step": 13443
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.297821226690224,
      "learning_rate": 3.850275283172611e-08,
      "loss": 0.1557,
      "step": 13444
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3453988818866625,
      "learning_rate": 3.8359403671980876e-08,
      "loss": 0.1489,
      "step": 13445
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.310748126225704,
      "learning_rate": 3.821632083339133e-08,
      "loss": 0.1756,
      "step": 13446
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3953551702796914,
      "learning_rate": 3.8073504323637434e-08,
      "loss": 0.1544,
      "step": 13447
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.894984344402886,
      "learning_rate": 3.793095415038528e-08,
      "loss": 0.5836,
      "step": 13448
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2750278313666839,
      "learning_rate": 3.778867032128653e-08,
      "loss": 0.1588,
      "step": 13449
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6094053246130275,
      "learning_rate": 3.764665284397784e-08,
      "loss": 0.1746,
      "step": 13450
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.238658067704647,
      "learning_rate": 3.750490172608257e-08,
      "loss": 0.1464,
      "step": 13451
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4034514801998699,
      "learning_rate": 3.7363416975209065e-08,
      "loss": 0.1544,
      "step": 13452
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4911247207287885,
      "learning_rate": 3.722219859895182e-08,
      "loss": 0.1969,
      "step": 13453
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3525103632612157,
      "learning_rate": 3.7081246604890876e-08,
      "loss": 0.1652,
      "step": 13454
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4092494223929755,
      "learning_rate": 3.69405610005924e-08,
      "loss": 0.1874,
      "step": 13455
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1959723883173319,
      "learning_rate": 3.6800141793607026e-08,
      "loss": 0.1484,
      "step": 13456
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.17583171042287,
      "learning_rate": 3.665998899147261e-08,
      "loss": 0.1321,
      "step": 13457
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3453310996721675,
      "learning_rate": 3.652010260171146e-08,
      "loss": 0.1308,
      "step": 13458
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5268250521353204,
      "learning_rate": 3.638048263183258e-08,
      "loss": 0.1989,
      "step": 13459
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5151274014800573,
      "learning_rate": 3.6241129089329416e-08,
      "loss": 0.1882,
      "step": 13460
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4312937639909997,
      "learning_rate": 3.6102041981682656e-08,
      "loss": 0.1766,
      "step": 13461
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3018182090407715,
      "learning_rate": 3.596322131635799e-08,
      "loss": 0.1452,
      "step": 13462
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.633747961327465,
      "learning_rate": 3.5824667100806676e-08,
      "loss": 0.5477,
      "step": 13463
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3934875907470676,
      "learning_rate": 3.5686379342464994e-08,
      "loss": 0.1936,
      "step": 13464
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3088132671013497,
      "learning_rate": 3.5548358048757007e-08,
      "loss": 0.1549,
      "step": 13465
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2805761724438518,
      "learning_rate": 3.541060322709011e-08,
      "loss": 0.156,
      "step": 13466
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3045256029709005,
      "learning_rate": 3.527311488485896e-08,
      "loss": 0.2026,
      "step": 13467
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.386591364130476,
      "learning_rate": 3.513589302944265e-08,
      "loss": 0.1849,
      "step": 13468
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.512860731630957,
      "learning_rate": 3.49989376682075e-08,
      "loss": 0.594,
      "step": 13469
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3971922711846156,
      "learning_rate": 3.486224880850431e-08,
      "loss": 0.1768,
      "step": 13470
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3534205229345293,
      "learning_rate": 3.472582645767053e-08,
      "loss": 0.1699,
      "step": 13471
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.306108318680653,
      "learning_rate": 3.458967062302865e-08,
      "loss": 0.1426,
      "step": 13472
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1931731613072083,
      "learning_rate": 3.4453781311886145e-08,
      "loss": 0.1331,
      "step": 13473
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1795034432627136,
      "learning_rate": 3.4318158531537746e-08,
      "loss": 0.1205,
      "step": 13474
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4211815443932077,
      "learning_rate": 3.4182802289263186e-08,
      "loss": 0.177,
      "step": 13475
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2465314056057581,
      "learning_rate": 3.4047712592327753e-08,
      "loss": 0.1556,
      "step": 13476
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3795035586194502,
      "learning_rate": 3.3912889447981765e-08,
      "loss": 0.1764,
      "step": 13477
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.520335697353306,
      "learning_rate": 3.377833286346333e-08,
      "loss": 0.2082,
      "step": 13478
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.349788380273607,
      "learning_rate": 3.3644042845993876e-08,
      "loss": 0.1825,
      "step": 13479
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6004288411105598,
      "learning_rate": 3.351001940278209e-08,
      "loss": 0.1877,
      "step": 13480
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.169855199066051,
      "learning_rate": 3.3376262541021667e-08,
      "loss": 0.1493,
      "step": 13481
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.441778544750508,
      "learning_rate": 3.3242772267891853e-08,
      "loss": 0.5192,
      "step": 13482
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2088909576113247,
      "learning_rate": 3.3109548590558595e-08,
      "loss": 0.1252,
      "step": 13483
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.7969975915167264,
      "learning_rate": 3.297659151617172e-08,
      "loss": 0.5821,
      "step": 13484
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3614781721951619,
      "learning_rate": 3.2843901051868854e-08,
      "loss": 0.1933,
      "step": 13485
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2482815065413153,
      "learning_rate": 3.271147720477208e-08,
      "loss": 0.1232,
      "step": 13486
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.355364373719001,
      "learning_rate": 3.257931998198904e-08,
      "loss": 0.1911,
      "step": 13487
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.4889812160759455,
      "learning_rate": 3.244742939061352e-08,
      "loss": 0.507,
      "step": 13488
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4028676514759573,
      "learning_rate": 3.231580543772483e-08,
      "loss": 0.1781,
      "step": 13489
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3188738616019953,
      "learning_rate": 3.2184448130388456e-08,
      "loss": 0.1356,
      "step": 13490
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3738877569673011,
      "learning_rate": 3.205335747565486e-08,
      "loss": 0.1908,
      "step": 13491
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5249259141448719,
      "learning_rate": 3.192253348056062e-08,
      "loss": 0.1746,
      "step": 13492
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5052533731646958,
      "learning_rate": 3.1791976152126816e-08,
      "loss": 0.1891,
      "step": 13493
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.238036928738311,
      "learning_rate": 3.1661685497362816e-08,
      "loss": 0.1458,
      "step": 13494
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.409437953025684,
      "learning_rate": 3.153166152326137e-08,
      "loss": 0.6771,
      "step": 13495
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3269830434880143,
      "learning_rate": 3.1401904236801336e-08,
      "loss": 0.1431,
      "step": 13496
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.375706950459523,
      "learning_rate": 3.127241364494771e-08,
      "loss": 0.1382,
      "step": 13497
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2645535490570792,
      "learning_rate": 3.1143189754651584e-08,
      "loss": 0.1605,
      "step": 13498
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2799245748736396,
      "learning_rate": 3.1014232572848525e-08,
      "loss": 0.1312,
      "step": 13499
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.136469198607651,
      "learning_rate": 3.088554210646133e-08,
      "loss": 0.1442,
      "step": 13500
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4319355551291426,
      "learning_rate": 3.0757118362396144e-08,
      "loss": 0.1684,
      "step": 13501
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6479174243125727,
      "learning_rate": 3.0628961347547445e-08,
      "loss": 0.2019,
      "step": 13502
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2836780399883283,
      "learning_rate": 3.0501071068794185e-08,
      "loss": 0.1845,
      "step": 13503
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.285343074122367,
      "learning_rate": 3.037344753300031e-08,
      "loss": 0.1475,
      "step": 13504
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5010741709669615,
      "learning_rate": 3.0246090747016455e-08,
      "loss": 0.1653,
      "step": 13505
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2568853881008153,
      "learning_rate": 3.0119000717678814e-08,
      "loss": 0.1399,
      "step": 13506
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.393717375410588,
      "learning_rate": 2.9992177451809156e-08,
      "loss": 0.1682,
      "step": 13507
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3677880404379226,
      "learning_rate": 2.986562095621426e-08,
      "loss": 0.1907,
      "step": 13508
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4056186342977315,
      "learning_rate": 2.9739331237687575e-08,
      "loss": 0.1635,
      "step": 13509
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3079993999593347,
      "learning_rate": 2.9613308303007572e-08,
      "loss": 0.1399,
      "step": 13510
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2760156072200008,
      "learning_rate": 2.948755215893939e-08,
      "loss": 0.1691,
      "step": 13511
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.373478363788751,
      "learning_rate": 2.936206281223264e-08,
      "loss": 0.484,
      "step": 13512
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4733086704891554,
      "learning_rate": 2.923684026962248e-08,
      "loss": 0.1784,
      "step": 13513
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4061030306317441,
      "learning_rate": 2.9111884537831315e-08,
      "loss": 0.2032,
      "step": 13514
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3760417016050845,
      "learning_rate": 2.8987195623566e-08,
      "loss": 0.1589,
      "step": 13515
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2306461682771985,
      "learning_rate": 2.886277353351896e-08,
      "loss": 0.1757,
      "step": 13516
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5113202700567616,
      "learning_rate": 2.8738618274369302e-08,
      "loss": 0.1755,
      "step": 13517
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3946485198098404,
      "learning_rate": 2.861472985278002e-08,
      "loss": 0.2106,
      "step": 13518
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.423962663122846,
      "learning_rate": 2.8491108275403023e-08,
      "loss": 0.1372,
      "step": 13519
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2973631005391462,
      "learning_rate": 2.8367753548871335e-08,
      "loss": 0.1771,
      "step": 13520
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3442053530083253,
      "learning_rate": 2.8244665679807992e-08,
      "loss": 0.1863,
      "step": 13521
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.052238576094346,
      "learning_rate": 2.8121844674819375e-08,
      "loss": 0.5306,
      "step": 13522
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2926721786385134,
      "learning_rate": 2.799929054049799e-08,
      "loss": 0.1715,
      "step": 13523
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2991845001196516,
      "learning_rate": 2.7877003283421356e-08,
      "loss": 0.1665,
      "step": 13524
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1588570258818658,
      "learning_rate": 2.775498291015477e-08,
      "loss": 0.142,
      "step": 13525
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4322364013974507,
      "learning_rate": 2.7633229427246334e-08,
      "loss": 0.1483,
      "step": 13526
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.294714964995783,
      "learning_rate": 2.751174284123248e-08,
      "loss": 0.1055,
      "step": 13527
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.39709608110624,
      "learning_rate": 2.7390523158633552e-08,
      "loss": 0.1364,
      "step": 13528
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3136065560964818,
      "learning_rate": 2.7269570385956012e-08,
      "loss": 0.1412,
      "step": 13529
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3340868632646825,
      "learning_rate": 2.7148884529692444e-08,
      "loss": 0.1773,
      "step": 13530
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.36558477984767,
      "learning_rate": 2.702846559632044e-08,
      "loss": 0.1767,
      "step": 13531
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.393137723140857,
      "learning_rate": 2.690831359230428e-08,
      "loss": 0.139,
      "step": 13532
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.503710022547652,
      "learning_rate": 2.6788428524092136e-08,
      "loss": 0.1695,
      "step": 13533
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.493751394526797,
      "learning_rate": 2.6668810398120524e-08,
      "loss": 0.1842,
      "step": 13534
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.386651815626172,
      "learning_rate": 2.6549459220808204e-08,
      "loss": 0.1685,
      "step": 13535
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3463293067683428,
      "learning_rate": 2.643037499856338e-08,
      "loss": 0.1578,
      "step": 13536
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3135651399226986,
      "learning_rate": 2.6311557737776495e-08,
      "loss": 0.178,
      "step": 13537
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1243493549468184,
      "learning_rate": 2.619300744482578e-08,
      "loss": 0.1297,
      "step": 13538
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.269470891552126,
      "learning_rate": 2.607472412607448e-08,
      "loss": 0.1522,
      "step": 13539
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3015713698427231,
      "learning_rate": 2.595670778787196e-08,
      "loss": 0.1689,
      "step": 13540
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3901035013019694,
      "learning_rate": 2.5838958436552043e-08,
      "loss": 0.1863,
      "step": 13541
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3181246927638073,
      "learning_rate": 2.5721476078436335e-08,
      "loss": 0.1708,
      "step": 13542
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.307648757868234,
      "learning_rate": 2.5604260719829244e-08,
      "loss": 0.1572,
      "step": 13543
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2966709292091905,
      "learning_rate": 2.5487312367024063e-08,
      "loss": 0.1583,
      "step": 13544
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2317757922946628,
      "learning_rate": 2.5370631026296334e-08,
      "loss": 0.1846,
      "step": 13545
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.447115643116738,
      "learning_rate": 2.5254216703910483e-08,
      "loss": 0.1747,
      "step": 13546
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2540977062694127,
      "learning_rate": 2.5138069406114296e-08,
      "loss": 0.1591,
      "step": 13547
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.140631595495422,
      "learning_rate": 2.5022189139143338e-08,
      "loss": 0.1201,
      "step": 13548
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.737416834232494,
      "learning_rate": 2.4906575909215968e-08,
      "loss": 0.5465,
      "step": 13549
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.380948013559843,
      "learning_rate": 2.479122972253889e-08,
      "loss": 0.1572,
      "step": 13550
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4641710164753734,
      "learning_rate": 2.4676150585302705e-08,
      "loss": 0.1943,
      "step": 13551
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.211314833726081,
      "learning_rate": 2.456133850368525e-08,
      "loss": 0.1467,
      "step": 13552
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3583144272661467,
      "learning_rate": 2.4446793483848818e-08,
      "loss": 0.1889,
      "step": 13553
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2773645151015598,
      "learning_rate": 2.433251553194127e-08,
      "loss": 0.142,
      "step": 13554
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3066702279865525,
      "learning_rate": 2.4218504654097696e-08,
      "loss": 0.1842,
      "step": 13555
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2953732597589545,
      "learning_rate": 2.4104760856436538e-08,
      "loss": 0.1492,
      "step": 13556
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.353893000896847,
      "learning_rate": 2.3991284145063465e-08,
      "loss": 0.1551,
      "step": 13557
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.40933615701645,
      "learning_rate": 2.387807452607027e-08,
      "loss": 0.1677,
      "step": 13558
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1648837223117543,
      "learning_rate": 2.3765132005532654e-08,
      "loss": 0.1336,
      "step": 13559
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2435376472134816,
      "learning_rate": 2.3652456589512983e-08,
      "loss": 0.174,
      "step": 13560
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.527629800209174,
      "learning_rate": 2.3540048284059758e-08,
      "loss": 0.1792,
      "step": 13561
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3682203102662622,
      "learning_rate": 2.3427907095205927e-08,
      "loss": 0.1337,
      "step": 13562
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4065334497810493,
      "learning_rate": 2.3316033028971672e-08,
      "loss": 0.1517,
      "step": 13563
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.641842457643189,
      "learning_rate": 2.3204426091361087e-08,
      "loss": 0.5222,
      "step": 13564
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6858020574352042,
      "learning_rate": 2.3093086288365486e-08,
      "loss": 0.2067,
      "step": 13565
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5102376833356161,
      "learning_rate": 2.298201362596064e-08,
      "loss": 0.1865,
      "step": 13566
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3377594375223474,
      "learning_rate": 2.2871208110108457e-08,
      "loss": 0.1749,
      "step": 13567
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.323008478608899,
      "learning_rate": 2.2760669746756947e-08,
      "loss": 0.142,
      "step": 13568
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2851556554603194,
      "learning_rate": 2.2650398541839147e-08,
      "loss": 0.1452,
      "step": 13569
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5506223029508786,
      "learning_rate": 2.254039450127421e-08,
      "loss": 0.2062,
      "step": 13570
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.462420555545724,
      "learning_rate": 2.24306576309663e-08,
      "loss": 0.1888,
      "step": 13571
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4176010585166643,
      "learning_rate": 2.232118793680571e-08,
      "loss": 0.1737,
      "step": 13572
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.370509904235416,
      "learning_rate": 2.2211985424668292e-08,
      "loss": 0.2076,
      "step": 13573
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2507334598061663,
      "learning_rate": 2.2103050100416023e-08,
      "loss": 0.1228,
      "step": 13574
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2975733931000102,
      "learning_rate": 2.199438196989645e-08,
      "loss": 0.1411,
      "step": 13575
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3106961880159813,
      "learning_rate": 2.188598103894102e-08,
      "loss": 0.1624,
      "step": 13576
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.275428201541623,
      "learning_rate": 2.1777847313370072e-08,
      "loss": 0.1385,
      "step": 13577
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.021685072078932,
      "learning_rate": 2.166998079898619e-08,
      "loss": 0.6221,
      "step": 13578
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6505374795838146,
      "learning_rate": 2.1562381501579742e-08,
      "loss": 0.231,
      "step": 13579
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2723782252385625,
      "learning_rate": 2.1455049426926666e-08,
      "loss": 0.1548,
      "step": 13580
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4079686129393176,
      "learning_rate": 2.1347984580787905e-08,
      "loss": 0.176,
      "step": 13581
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4057700186481081,
      "learning_rate": 2.1241186968909978e-08,
      "loss": 0.1799,
      "step": 13582
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5915269908275451,
      "learning_rate": 2.1134656597025515e-08,
      "loss": 0.1894,
      "step": 13583
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.322932801834802,
      "learning_rate": 2.1028393470852726e-08,
      "loss": 0.4568,
      "step": 13584
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.115610777201104,
      "learning_rate": 2.0922397596095934e-08,
      "loss": 0.12,
      "step": 13585
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4340244667987418,
      "learning_rate": 2.0816668978443366e-08,
      "loss": 0.1749,
      "step": 13586
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2748364596144908,
      "learning_rate": 2.071120762357104e-08,
      "loss": 0.1472,
      "step": 13587
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4396492205043323,
      "learning_rate": 2.0606013537139426e-08,
      "loss": 0.2077,
      "step": 13588
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3522635400638698,
      "learning_rate": 2.0501086724794562e-08,
      "loss": 0.1609,
      "step": 13589
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5862277454772007,
      "learning_rate": 2.0396427192169167e-08,
      "loss": 0.2063,
      "step": 13590
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1371611264516623,
      "learning_rate": 2.0292034944880967e-08,
      "loss": 0.1396,
      "step": 13591
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2887817169502267,
      "learning_rate": 2.018790998853215e-08,
      "loss": 0.1591,
      "step": 13592
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4724216636395278,
      "learning_rate": 2.0084052328713244e-08,
      "loss": 0.1678,
      "step": 13593
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.166303051009998,
      "learning_rate": 1.998046197099812e-08,
      "loss": 0.1328,
      "step": 13594
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3836734178556176,
      "learning_rate": 1.9877138920947336e-08,
      "loss": 0.1661,
      "step": 13595
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3333959953656356,
      "learning_rate": 1.977408318410645e-08,
      "loss": 0.1908,
      "step": 13596
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4519152017831924,
      "learning_rate": 1.967129476600771e-08,
      "loss": 0.1864,
      "step": 13597
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2563387112455275,
      "learning_rate": 1.9568773672167808e-08,
      "loss": 0.1298,
      "step": 13598
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4986368972098423,
      "learning_rate": 1.946651990809012e-08,
      "loss": 0.1667,
      "step": 13599
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6021681789516407,
      "learning_rate": 1.9364533479263036e-08,
      "loss": 0.186,
      "step": 13600
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2442081121357813,
      "learning_rate": 1.9262814391160512e-08,
      "loss": 0.1705,
      "step": 13601
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2591588490206833,
      "learning_rate": 1.9161362649243175e-08,
      "loss": 0.1817,
      "step": 13602
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.34088666759668,
      "learning_rate": 1.9060178258955565e-08,
      "loss": 0.1498,
      "step": 13603
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3504884940156416,
      "learning_rate": 1.8959261225729997e-08,
      "loss": 0.1376,
      "step": 13604
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4763908506576335,
      "learning_rate": 1.8858611554982143e-08,
      "loss": 0.1876,
      "step": 13605
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4522179504093689,
      "learning_rate": 1.8758229252114902e-08,
      "loss": 0.2006,
      "step": 13606
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3870830387503175,
      "learning_rate": 1.8658114322516742e-08,
      "loss": 0.154,
      "step": 13607
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2696722845007449,
      "learning_rate": 1.855826677156114e-08,
      "loss": 0.1837,
      "step": 13608
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4218807358969392,
      "learning_rate": 1.84586866046077e-08,
      "loss": 0.1683,
      "step": 13609
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3104689302336376,
      "learning_rate": 1.8359373827001037e-08,
      "loss": 0.1515,
      "step": 13610
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2619490552277752,
      "learning_rate": 1.826032844407244e-08,
      "loss": 0.1357,
      "step": 13611
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2678489252363436,
      "learning_rate": 1.8161550461137657e-08,
      "loss": 0.1521,
      "step": 13612
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3738406959803933,
      "learning_rate": 1.8063039883499668e-08,
      "loss": 0.1877,
      "step": 13613
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2103895848046795,
      "learning_rate": 1.79647967164448e-08,
      "loss": 0.1513,
      "step": 13614
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3599412761436713,
      "learning_rate": 1.7866820965247168e-08,
      "loss": 0.1392,
      "step": 13615
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6249835013724747,
      "learning_rate": 1.7769112635166453e-08,
      "loss": 0.1704,
      "step": 13616
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.957354925186498,
      "learning_rate": 1.7671671731445682e-08,
      "loss": 0.6573,
      "step": 13617
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2429284440111148,
      "learning_rate": 1.757449825931623e-08,
      "loss": 0.1696,
      "step": 13618
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4260422732158284,
      "learning_rate": 1.7477592223992812e-08,
      "loss": 0.1883,
      "step": 13619
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3387779045222323,
      "learning_rate": 1.7380953630678488e-08,
      "loss": 0.1432,
      "step": 13620
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.092783678754347,
      "learning_rate": 1.7284582484559665e-08,
      "loss": 0.5289,
      "step": 13621
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2509832713388866,
      "learning_rate": 1.718847879080887e-08,
      "loss": 0.1478,
      "step": 13622
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3551402677435755,
      "learning_rate": 1.709264255458476e-08,
      "loss": 0.1565,
      "step": 13623
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3422694794441525,
      "learning_rate": 1.6997073781032104e-08,
      "loss": 0.1787,
      "step": 13624
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.607756781975114,
      "learning_rate": 1.6901772475279575e-08,
      "loss": 0.6996,
      "step": 13625
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.398580887468457,
      "learning_rate": 1.680673864244309e-08,
      "loss": 0.1527,
      "step": 13626
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.277015634211775,
      "learning_rate": 1.671197228762411e-08,
      "loss": 0.1448,
      "step": 13627
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.482700690368156,
      "learning_rate": 1.661747341590858e-08,
      "loss": 0.194,
      "step": 13628
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5557842591039206,
      "learning_rate": 1.6523242032369658e-08,
      "loss": 0.1923,
      "step": 13629
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.70891279574617,
      "learning_rate": 1.6429278142063853e-08,
      "loss": 0.1571,
      "step": 13630
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3731601703717145,
      "learning_rate": 1.6335581750036578e-08,
      "loss": 0.1778,
      "step": 13631
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3757931445094258,
      "learning_rate": 1.624215286131603e-08,
      "loss": 0.1448,
      "step": 13632
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2359218706321575,
      "learning_rate": 1.6148991480917086e-08,
      "loss": 0.1522,
      "step": 13633
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.726410027981581,
      "learning_rate": 1.605609761384075e-08,
      "loss": 0.5658,
      "step": 13634
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5818416633369334,
      "learning_rate": 1.5963471265072472e-08,
      "loss": 0.1914,
      "step": 13635
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4887617997817009,
      "learning_rate": 1.5871112439584946e-08,
      "loss": 0.188,
      "step": 13636
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5783026153143933,
      "learning_rate": 1.5779021142335316e-08,
      "loss": 0.2248,
      "step": 13637
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.28243782167815,
      "learning_rate": 1.5687197378266296e-08,
      "loss": 0.1547,
      "step": 13638
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.491263138746027,
      "learning_rate": 1.559564115230672e-08,
      "loss": 0.2329,
      "step": 13639
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3412353556003431,
      "learning_rate": 1.5504352469371543e-08,
      "loss": 0.1698,
      "step": 13640
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.01429030309989,
      "learning_rate": 1.541333133436018e-08,
      "loss": 0.5715,
      "step": 13641
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.831156879732491,
      "learning_rate": 1.5322577752158175e-08,
      "loss": 0.5884,
      "step": 13642
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2958416136460544,
      "learning_rate": 1.5232091727637734e-08,
      "loss": 0.1613,
      "step": 13643
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2502241942707604,
      "learning_rate": 1.5141873265654973e-08,
      "loss": 0.1576,
      "step": 13644
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3086153672252985,
      "learning_rate": 1.5051922371052685e-08,
      "loss": 0.1669,
      "step": 13645
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4330177714146035,
      "learning_rate": 1.4962239048658677e-08,
      "loss": 0.1604,
      "step": 13646
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4866202884983841,
      "learning_rate": 1.4872823303287431e-08,
      "loss": 0.1837,
      "step": 13647
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4817413513364268,
      "learning_rate": 1.478367513973844e-08,
      "loss": 0.1731,
      "step": 13648
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3986636947498032,
      "learning_rate": 1.4694794562796211e-08,
      "loss": 0.181,
      "step": 13649
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4873203935920392,
      "learning_rate": 1.4606181577231927e-08,
      "loss": 0.1482,
      "step": 13650
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.284585597745238,
      "learning_rate": 1.4517836187802337e-08,
      "loss": 0.1417,
      "step": 13651
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2932031321709383,
      "learning_rate": 1.4429758399248094e-08,
      "loss": 0.1441,
      "step": 13652
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2561672503447745,
      "learning_rate": 1.434194821629875e-08,
      "loss": 0.1676,
      "step": 13653
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3874813212949326,
      "learning_rate": 1.4254405643666092e-08,
      "loss": 0.2054,
      "step": 13654
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8281917446366627,
      "learning_rate": 1.416713068604969e-08,
      "loss": 0.1868,
      "step": 13655
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2495077626406412,
      "learning_rate": 1.4080123348134133e-08,
      "loss": 0.1484,
      "step": 13656
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3787498720016917,
      "learning_rate": 1.3993383634589575e-08,
      "loss": 0.1544,
      "step": 13657
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3549737262384072,
      "learning_rate": 1.3906911550071734e-08,
      "loss": 0.1648,
      "step": 13658
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2818011836952403,
      "learning_rate": 1.3820707099222453e-08,
      "loss": 0.1702,
      "step": 13659
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4021808740955757,
      "learning_rate": 1.373477028666803e-08,
      "loss": 0.1556,
      "step": 13660
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.171152481828228,
      "learning_rate": 1.3649101117022002e-08,
      "loss": 0.1322,
      "step": 13661
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.392952745961978,
      "learning_rate": 1.3563699594882907e-08,
      "loss": 0.5976,
      "step": 13662
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1817304071105879,
      "learning_rate": 1.3478565724833747e-08,
      "loss": 0.1381,
      "step": 13663
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.802152022118856,
      "learning_rate": 1.3393699511445314e-08,
      "loss": 0.5644,
      "step": 13664
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2840154929253358,
      "learning_rate": 1.3309100959271737e-08,
      "loss": 0.1852,
      "step": 13665
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4299286498283428,
      "learning_rate": 1.3224770072854942e-08,
      "loss": 0.1513,
      "step": 13666
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5649447232021354,
      "learning_rate": 1.3140706856721307e-08,
      "loss": 0.2265,
      "step": 13667
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2365933082157086,
      "learning_rate": 1.3056911315382226e-08,
      "loss": 0.1586,
      "step": 13668
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2718908973097662,
      "learning_rate": 1.297338345333632e-08,
      "loss": 0.1768,
      "step": 13669
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2849420592441336,
      "learning_rate": 1.2890123275067224e-08,
      "loss": 0.1337,
      "step": 13670
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4040937369768771,
      "learning_rate": 1.2807130785043031e-08,
      "loss": 0.1943,
      "step": 13671
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2769576609827307,
      "learning_rate": 1.2724405987719069e-08,
      "loss": 0.1788,
      "step": 13672
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4783892473952724,
      "learning_rate": 1.264194888753567e-08,
      "loss": 0.173,
      "step": 13673
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4050780206677376,
      "learning_rate": 1.2559759488918743e-08,
      "loss": 0.1594,
      "step": 13674
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3470650374524797,
      "learning_rate": 1.2477837796280312e-08,
      "loss": 0.1628,
      "step": 13675
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4469025340298471,
      "learning_rate": 1.239618381401686e-08,
      "loss": 0.1834,
      "step": 13676
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.439013780601163,
      "learning_rate": 1.2314797546510992e-08,
      "loss": 0.4704,
      "step": 13677
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4365031854155996,
      "learning_rate": 1.2233678998132548e-08,
      "loss": 0.1742,
      "step": 13678
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2454433950419583,
      "learning_rate": 1.2152828173234155e-08,
      "loss": 0.1527,
      "step": 13679
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.30664595558892,
      "learning_rate": 1.2072245076156786e-08,
      "loss": 0.1675,
      "step": 13680
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3265562309851793,
      "learning_rate": 1.1991929711225314e-08,
      "loss": 0.1087,
      "step": 13681
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.419309967487841,
      "learning_rate": 1.1911882082750736e-08,
      "loss": 0.1896,
      "step": 13682
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.253065982253021,
      "learning_rate": 1.1832102195029615e-08,
      "loss": 0.1417,
      "step": 13683
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4424089615964648,
      "learning_rate": 1.1752590052344081e-08,
      "loss": 0.1847,
      "step": 13684
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4326565573057186,
      "learning_rate": 1.1673345658962388e-08,
      "loss": 0.1571,
      "step": 13685
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.488183356945678,
      "learning_rate": 1.1594369019138352e-08,
      "loss": 0.154,
      "step": 13686
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4508986156085968,
      "learning_rate": 1.1515660137110252e-08,
      "loss": 0.1662,
      "step": 13687
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4102147912612923,
      "learning_rate": 1.1437219017103596e-08,
      "loss": 0.1705,
      "step": 13688
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2619362175231899,
      "learning_rate": 1.135904566332835e-08,
      "loss": 0.1623,
      "step": 13689
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1532689969027141,
      "learning_rate": 1.1281140079980602e-08,
      "loss": 0.1479,
      "step": 13690
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.359846341065935,
      "learning_rate": 1.1203502271242006e-08,
      "loss": 0.1484,
      "step": 13691
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3963050551267666,
      "learning_rate": 1.1126132241280342e-08,
      "loss": 0.1922,
      "step": 13692
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.538547415771429,
      "learning_rate": 1.1049029994248395e-08,
      "loss": 0.7705,
      "step": 13693
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.412139762557859,
      "learning_rate": 1.097219553428397e-08,
      "loss": 0.1855,
      "step": 13694
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.174411274526148,
      "learning_rate": 1.0895628865512099e-08,
      "loss": 0.6033,
      "step": 13695
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.941993593021459,
      "learning_rate": 1.0819329992041715e-08,
      "loss": 0.5794,
      "step": 13696
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.665516705063931,
      "learning_rate": 1.0743298917968992e-08,
      "loss": 0.6437,
      "step": 13697
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3171863462035416,
      "learning_rate": 1.066753564737455e-08,
      "loss": 0.1633,
      "step": 13698
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.488037106473797,
      "learning_rate": 1.0592040184325691e-08,
      "loss": 0.2197,
      "step": 13699
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3146266555087547,
      "learning_rate": 1.0516812532873622e-08,
      "loss": 0.1828,
      "step": 13700
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8944955460657231,
      "learning_rate": 1.0441852697057331e-08,
      "loss": 0.2096,
      "step": 13701
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1950643756463852,
      "learning_rate": 1.0367160680899712e-08,
      "loss": 0.1191,
      "step": 13702
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0103163851092223,
      "learning_rate": 1.0292736488410338e-08,
      "loss": 0.1091,
      "step": 13703
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2977024662123242,
      "learning_rate": 1.0218580123583788e-08,
      "loss": 0.1816,
      "step": 13704
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.333465346668352,
      "learning_rate": 1.0144691590400768e-08,
      "loss": 0.16,
      "step": 13705
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4560829074450061,
      "learning_rate": 1.007107089282644e-08,
      "loss": 0.1832,
      "step": 13706
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.580717367412487,
      "learning_rate": 9.997718034813197e-09,
      "loss": 0.4536,
      "step": 13707
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6792400841646018,
      "learning_rate": 9.924633020298446e-09,
      "loss": 0.1925,
      "step": 13708
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1682522489313207,
      "learning_rate": 9.85181585320516e-09,
      "loss": 0.1474,
      "step": 13709
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.275167987637848,
      "learning_rate": 9.779266537441323e-09,
      "loss": 0.1605,
      "step": 13710
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1430458150342195,
      "learning_rate": 9.706985076901043e-09,
      "loss": 0.1225,
      "step": 13711
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3836941751382688,
      "learning_rate": 9.63497147546455e-09,
      "loss": 0.1672,
      "step": 13712
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5474968334224275,
      "learning_rate": 9.563225736997638e-09,
      "loss": 0.2042,
      "step": 13713
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2862071068011836,
      "learning_rate": 9.491747865350564e-09,
      "loss": 0.1637,
      "step": 13714
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3945164789340625,
      "learning_rate": 9.420537864359703e-09,
      "loss": 0.1689,
      "step": 13715
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2973386921974353,
      "learning_rate": 9.349595737848105e-09,
      "loss": 0.1466,
      "step": 13716
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1836265285983647,
      "learning_rate": 9.278921489623838e-09,
      "loss": 0.1158,
      "step": 13717
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.311131786681849,
      "learning_rate": 9.208515123479978e-09,
      "loss": 0.1411,
      "step": 13718
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1869810888984487,
      "learning_rate": 9.138376643195168e-09,
      "loss": 0.1644,
      "step": 13719
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4224385635842964,
      "learning_rate": 9.068506052534732e-09,
      "loss": 0.1632,
      "step": 13720
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.393752525969081,
      "learning_rate": 8.998903355249556e-09,
      "loss": 0.1678,
      "step": 13721
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4300052794041918,
      "learning_rate": 8.92956855507443e-09,
      "loss": 0.1542,
      "step": 13722
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.548638313269367,
      "learning_rate": 8.860501655731379e-09,
      "loss": 0.476,
      "step": 13723
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1525192944754574,
      "learning_rate": 8.791702660928547e-09,
      "loss": 0.1448,
      "step": 13724
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4760506281235386,
      "learning_rate": 8.723171574357426e-09,
      "loss": 0.1626,
      "step": 13725
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.255571970163762,
      "learning_rate": 8.654908399697848e-09,
      "loss": 0.5562,
      "step": 13726
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3349079459426851,
      "learning_rate": 8.586913140612996e-09,
      "loss": 0.1928,
      "step": 13727
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3700897179093627,
      "learning_rate": 8.519185800752727e-09,
      "loss": 0.1699,
      "step": 13728
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3145531214857635,
      "learning_rate": 8.451726383752468e-09,
      "loss": 0.1523,
      "step": 13729
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3461762697053308,
      "learning_rate": 8.384534893233209e-09,
      "loss": 0.1554,
      "step": 13730
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2291045194725614,
      "learning_rate": 8.317611332801512e-09,
      "loss": 0.1517,
      "step": 13731
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3675255427154311,
      "learning_rate": 8.250955706049502e-09,
      "loss": 0.1602,
      "step": 13732
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.289106167270459,
      "learning_rate": 8.184568016554872e-09,
      "loss": 0.1596,
      "step": 13733
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2813556293207777,
      "learning_rate": 8.11844826788144e-09,
      "loss": 0.166,
      "step": 13734
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2719586337953566,
      "learning_rate": 8.052596463577477e-09,
      "loss": 0.1513,
      "step": 13735
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.227281161319752,
      "learning_rate": 7.987012607179046e-09,
      "loss": 0.396,
      "step": 13736
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2950775116329474,
      "learning_rate": 7.921696702204995e-09,
      "loss": 0.152,
      "step": 13737
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0928852801904914,
      "learning_rate": 7.856648752161966e-09,
      "loss": 0.1413,
      "step": 13738
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5318661272977012,
      "learning_rate": 7.791868760541055e-09,
      "loss": 0.1874,
      "step": 13739
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3761890682488607,
      "learning_rate": 7.727356730820035e-09,
      "loss": 0.1852,
      "step": 13740
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4822769714589838,
      "learning_rate": 7.66311266646058e-09,
      "loss": 0.2208,
      "step": 13741
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2489083369739473,
      "learning_rate": 7.599136570912713e-09,
      "loss": 0.1471,
      "step": 13742
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3381313753926616,
      "learning_rate": 7.535428447608683e-09,
      "loss": 0.156,
      "step": 13743
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.596558459246474,
      "learning_rate": 7.47198829996909e-09,
      "loss": 0.2048,
      "step": 13744
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4135431346394731,
      "learning_rate": 7.408816131398988e-09,
      "loss": 0.1843,
      "step": 13745
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3830020049533036,
      "learning_rate": 7.345911945288997e-09,
      "loss": 0.1783,
      "step": 13746
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5241591561959147,
      "learning_rate": 7.283275745015861e-09,
      "loss": 0.1955,
      "step": 13747
    },
    {
      "epoch": 0.98,
      "grad_norm": 12.828411825211115,
      "learning_rate": 7.220907533941335e-09,
      "loss": 0.5722,
      "step": 13748
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2432314711460515,
      "learning_rate": 7.158807315412741e-09,
      "loss": 0.1618,
      "step": 13749
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2423568703940722,
      "learning_rate": 7.096975092764636e-09,
      "loss": 0.1339,
      "step": 13750
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.245637178702311,
      "learning_rate": 7.035410869314918e-09,
      "loss": 0.1441,
      "step": 13751
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3596460615185115,
      "learning_rate": 6.974114648368169e-09,
      "loss": 0.1551,
      "step": 13752
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.632297140446915,
      "learning_rate": 6.913086433215088e-09,
      "loss": 0.5309,
      "step": 13753
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3802793238812316,
      "learning_rate": 6.852326227130835e-09,
      "loss": 0.1928,
      "step": 13754
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3730332427384633,
      "learning_rate": 6.791834033376688e-09,
      "loss": 0.167,
      "step": 13755
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3396676914975802,
      "learning_rate": 6.7316098552000495e-09,
      "loss": 0.1711,
      "step": 13756
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.335828160071708,
      "learning_rate": 6.6716536958338905e-09,
      "loss": 0.1537,
      "step": 13757
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3572897822043613,
      "learning_rate": 6.611965558495637e-09,
      "loss": 0.1557,
      "step": 13758
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3616984237363876,
      "learning_rate": 6.552545446389391e-09,
      "loss": 0.131,
      "step": 13759
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.339915040423464,
      "learning_rate": 6.49339336270427e-09,
      "loss": 0.1804,
      "step": 13760
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.488336141687242,
      "learning_rate": 6.434509310616066e-09,
      "loss": 0.5782,
      "step": 13761
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2834270414653823,
      "learning_rate": 6.375893293284474e-09,
      "loss": 0.1824,
      "step": 13762
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4529077352286064,
      "learning_rate": 6.3175453138569765e-09,
      "loss": 0.1757,
      "step": 13763
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0778138032486184,
      "learning_rate": 6.259465375464402e-09,
      "loss": 0.1414,
      "step": 13764
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2504729875014426,
      "learning_rate": 6.201653481224812e-09,
      "loss": 0.1778,
      "step": 13765
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.524317093784593,
      "learning_rate": 6.14410963424128e-09,
      "loss": 0.1727,
      "step": 13766
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3002314894107676,
      "learning_rate": 6.086833837601891e-09,
      "loss": 0.1652,
      "step": 13767
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.412010258639999,
      "learning_rate": 6.029826094381963e-09,
      "loss": 0.1682,
      "step": 13768
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.604111773345135,
      "learning_rate": 5.9730864076407155e-09,
      "loss": 0.1827,
      "step": 13769
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.441446528046009,
      "learning_rate": 5.916614780424046e-09,
      "loss": 0.1989,
      "step": 13770
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4225681161413641,
      "learning_rate": 5.8604112157634176e-09,
      "loss": 0.1664,
      "step": 13771
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.257400145508779,
      "learning_rate": 5.8044757166747515e-09,
      "loss": 0.1355,
      "step": 13772
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5911716864761534,
      "learning_rate": 5.7488082861612005e-09,
      "loss": 0.1467,
      "step": 13773
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.357925181368716,
      "learning_rate": 5.693408927210376e-09,
      "loss": 0.1566,
      "step": 13774
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3351715466857323,
      "learning_rate": 5.6382776427960085e-09,
      "loss": 0.1731,
      "step": 13775
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3736141821497354,
      "learning_rate": 5.583414435877399e-09,
      "loss": 0.1774,
      "step": 13776
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4220336920638375,
      "learning_rate": 5.5288193093999685e-09,
      "loss": 0.2062,
      "step": 13777
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2236400141963666,
      "learning_rate": 5.474492266292486e-09,
      "loss": 0.1441,
      "step": 13778
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7857313572670095,
      "learning_rate": 5.420433309472617e-09,
      "loss": 0.1771,
      "step": 13779
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.469384947483464,
      "learning_rate": 5.366642441841374e-09,
      "loss": 0.1514,
      "step": 13780
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3329658960211284,
      "learning_rate": 5.313119666286448e-09,
      "loss": 0.1746,
      "step": 13781
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3244398817176808,
      "learning_rate": 5.259864985679985e-09,
      "loss": 0.1579,
      "step": 13782
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4639015529504087,
      "learning_rate": 5.20687840288081e-09,
      "loss": 0.179,
      "step": 13783
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.452927532342113,
      "learning_rate": 5.15415992073387e-09,
      "loss": 0.1764,
      "step": 13784
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1497458903228337,
      "learning_rate": 5.101709542067457e-09,
      "loss": 0.1445,
      "step": 13785
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2946875451041968,
      "learning_rate": 5.0495272696976515e-09,
      "loss": 0.1703,
      "step": 13786
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.341664595397723,
      "learning_rate": 4.9976131064255475e-09,
      "loss": 0.1771,
      "step": 13787
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5134442351925002,
      "learning_rate": 4.945967055037803e-09,
      "loss": 0.1581,
      "step": 13788
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3872128808359643,
      "learning_rate": 4.8945891183055374e-09,
      "loss": 0.1718,
      "step": 13789
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4643336406867975,
      "learning_rate": 4.843479298987652e-09,
      "loss": 0.1742,
      "step": 13790
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4236404793807012,
      "learning_rate": 4.792637599826954e-09,
      "loss": 0.2016,
      "step": 13791
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5668141598943903,
      "learning_rate": 4.742064023552928e-09,
      "loss": 0.1795,
      "step": 13792
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2328018099672282,
      "learning_rate": 4.691758572879512e-09,
      "loss": 0.1688,
      "step": 13793
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2962102341211879,
      "learning_rate": 4.64172125050677e-09,
      "loss": 0.1638,
      "step": 13794
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2411754179759507,
      "learning_rate": 4.591952059121441e-09,
      "loss": 0.1416,
      "step": 13795
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2895239430957919,
      "learning_rate": 4.542451001394721e-09,
      "loss": 0.1468,
      "step": 13796
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.4972244028484365,
      "learning_rate": 4.49321807998282e-09,
      "loss": 0.3848,
      "step": 13797
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.394042904006588,
      "learning_rate": 4.444253297528622e-09,
      "loss": 0.1618,
      "step": 13798
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3155024722609006,
      "learning_rate": 4.395556656661138e-09,
      "loss": 0.1686,
      "step": 13799
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5044844397560435,
      "learning_rate": 4.347128159993829e-09,
      "loss": 0.1547,
      "step": 13800
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3640647844849467,
      "learning_rate": 4.298967810125176e-09,
      "loss": 0.1709,
      "step": 13801
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3016381806809318,
      "learning_rate": 4.251075609641997e-09,
      "loss": 0.1808,
      "step": 13802
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2635510496716096,
      "learning_rate": 4.203451561113347e-09,
      "loss": 0.1338,
      "step": 13803
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3457264783451994,
      "learning_rate": 4.156095667096627e-09,
      "loss": 0.1609,
      "step": 13804
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3491570976060996,
      "learning_rate": 4.109007930133136e-09,
      "loss": 0.1322,
      "step": 13805
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.276806054389065,
      "learning_rate": 4.062188352750296e-09,
      "loss": 0.4151,
      "step": 13806
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4233562060078815,
      "learning_rate": 4.015636937461654e-09,
      "loss": 0.1451,
      "step": 13807
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4262538640117528,
      "learning_rate": 3.9693536867652096e-09,
      "loss": 0.1628,
      "step": 13808
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3219953544239822,
      "learning_rate": 3.923338603146198e-09,
      "loss": 0.1525,
      "step": 13809
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.239379911562932,
      "learning_rate": 3.877591689073756e-09,
      "loss": 0.1789,
      "step": 13810
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2982096725138397,
      "learning_rate": 3.832112947003697e-09,
      "loss": 0.1922,
      "step": 13811
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4406857720172193,
      "learning_rate": 3.786902379376844e-09,
      "loss": 0.2045,
      "step": 13812
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3810198627663803,
      "learning_rate": 3.741959988620147e-09,
      "loss": 0.187,
      "step": 13813
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.565326891081775,
      "learning_rate": 3.697285777146675e-09,
      "loss": 0.5557,
      "step": 13814
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3634287249816766,
      "learning_rate": 3.6528797473528444e-09,
      "loss": 0.1538,
      "step": 13815
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.180420957823594,
      "learning_rate": 3.6087419016234138e-09,
      "loss": 0.1623,
      "step": 13816
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3549995143170401,
      "learning_rate": 3.5648722423270443e-09,
      "loss": 0.1849,
      "step": 13817
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.419438213368951,
      "learning_rate": 3.521270771818519e-09,
      "loss": 0.1618,
      "step": 13818
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3854194785045055,
      "learning_rate": 3.477937492438188e-09,
      "loss": 0.2154,
      "step": 13819
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.857489542091022,
      "learning_rate": 3.4348724065119687e-09,
      "loss": 0.5328,
      "step": 13820
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4054028283122966,
      "learning_rate": 3.3920755163513452e-09,
      "loss": 0.2015,
      "step": 13821
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4328396991498686,
      "learning_rate": 3.349546824253924e-09,
      "loss": 0.1644,
      "step": 13822
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6048744891724689,
      "learning_rate": 3.307286332502324e-09,
      "loss": 0.1818,
      "step": 13823
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9169040545962999,
      "learning_rate": 3.2652940433647307e-09,
      "loss": 0.204,
      "step": 13824
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2049123804937716,
      "learning_rate": 3.2235699590948966e-09,
      "loss": 0.169,
      "step": 13825
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2746606210826756,
      "learning_rate": 3.1821140819326966e-09,
      "loss": 0.1965,
      "step": 13826
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3407719254261528,
      "learning_rate": 3.1409264141035733e-09,
      "loss": 0.1721,
      "step": 13827
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4506271364821373,
      "learning_rate": 3.1000069578179803e-09,
      "loss": 0.1884,
      "step": 13828
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4971809970222851,
      "learning_rate": 3.059355715272494e-09,
      "loss": 0.1954,
      "step": 13829
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.978356842968884,
      "learning_rate": 3.0189726886487026e-09,
      "loss": 0.6222,
      "step": 13830
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4012262409401515,
      "learning_rate": 2.9788578801148717e-09,
      "loss": 0.1834,
      "step": 13831
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3882680803367355,
      "learning_rate": 2.9390112918231684e-09,
      "loss": 0.1603,
      "step": 13832
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.440990790482852,
      "learning_rate": 2.8994329259135477e-09,
      "loss": 0.1478,
      "step": 13833
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.304334709826213,
      "learning_rate": 2.8601227845098667e-09,
      "loss": 0.188,
      "step": 13834
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.415511835647568,
      "learning_rate": 2.8210808697226586e-09,
      "loss": 0.2173,
      "step": 13835
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.395973490321563,
      "learning_rate": 2.78230718364636e-09,
      "loss": 0.1599,
      "step": 13836
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2888119754190626,
      "learning_rate": 2.7438017283631936e-09,
      "loss": 0.15,
      "step": 13837
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.630092657364098,
      "learning_rate": 2.7055645059398393e-09,
      "loss": 0.5132,
      "step": 13838
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3054035283267547,
      "learning_rate": 2.667595518428545e-09,
      "loss": 0.143,
      "step": 13839
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3884604685213746,
      "learning_rate": 2.62989476786768e-09,
      "loss": 0.1692,
      "step": 13840
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5262629410272037,
      "learning_rate": 2.5924622562800705e-09,
      "loss": 0.2212,
      "step": 13841
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3329465370982787,
      "learning_rate": 2.55529798567522e-09,
      "loss": 0.1485,
      "step": 13842
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.562009305947579,
      "learning_rate": 2.518401958048755e-09,
      "loss": 0.4823,
      "step": 13843
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.376370030540331,
      "learning_rate": 2.4817741753802027e-09,
      "loss": 0.2137,
      "step": 13844
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4222291116963806,
      "learning_rate": 2.445414639636323e-09,
      "loss": 0.2027,
      "step": 13845
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.384628795485498,
      "learning_rate": 2.4093233527677785e-09,
      "loss": 0.1685,
      "step": 13846
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1898567976599435,
      "learning_rate": 2.373500316713018e-09,
      "loss": 0.1485,
      "step": 13847
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.512665913106766,
      "learning_rate": 2.337945533393837e-09,
      "loss": 0.1757,
      "step": 13848
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4958055241781885,
      "learning_rate": 2.3026590047192652e-09,
      "loss": 0.1892,
      "step": 13849
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.383501439427282,
      "learning_rate": 2.2676407325827875e-09,
      "loss": 0.6196,
      "step": 13850
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.138526099515865,
      "learning_rate": 2.2328907188645664e-09,
      "loss": 0.111,
      "step": 13851
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.122043276348766,
      "learning_rate": 2.198408965429777e-09,
      "loss": 0.4725,
      "step": 13852
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3747909989192764,
      "learning_rate": 2.1641954741297157e-09,
      "loss": 0.2071,
      "step": 13853
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3359711194513268,
      "learning_rate": 2.130250246799581e-09,
      "loss": 0.1808,
      "step": 13854
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3684408717022054,
      "learning_rate": 2.0965732852623598e-09,
      "loss": 0.1762,
      "step": 13855
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3177776065431566,
      "learning_rate": 2.063164591325495e-09,
      "loss": 0.1614,
      "step": 13856
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2239040329795832,
      "learning_rate": 2.0300241667819963e-09,
      "loss": 0.1577,
      "step": 13857
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.411373262074015,
      "learning_rate": 1.9971520134109966e-09,
      "loss": 0.1701,
      "step": 13858
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3451912997335598,
      "learning_rate": 1.9645481329771954e-09,
      "loss": 0.1549,
      "step": 13859
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4628715625094626,
      "learning_rate": 1.9322125272297488e-09,
      "loss": 0.2052,
      "step": 13860
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3529642814259122,
      "learning_rate": 1.9001451979056006e-09,
      "loss": 0.1591,
      "step": 13861
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4468314423681135,
      "learning_rate": 1.868346146724487e-09,
      "loss": 0.1791,
      "step": 13862
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3347701353413681,
      "learning_rate": 1.8368153753944852e-09,
      "loss": 0.1665,
      "step": 13863
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.418342642271985,
      "learning_rate": 1.8055528856075755e-09,
      "loss": 0.2077,
      "step": 13864
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.401456031810564,
      "learning_rate": 1.77455867904186e-09,
      "loss": 0.2072,
      "step": 13865
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2801068683967731,
      "learning_rate": 1.743832757360453e-09,
      "loss": 0.1537,
      "step": 13866
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.730572935216498,
      "learning_rate": 1.7133751222137007e-09,
      "loss": 0.4997,
      "step": 13867
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4688276040561234,
      "learning_rate": 1.6831857752352964e-09,
      "loss": 0.2097,
      "step": 13868
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3153144200890292,
      "learning_rate": 1.653264718046721e-09,
      "loss": 0.1587,
      "step": 13869
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3668286885821832,
      "learning_rate": 1.6236119522533566e-09,
      "loss": 0.1698,
      "step": 13870
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2400394192461697,
      "learning_rate": 1.5942274794467083e-09,
      "loss": 0.1433,
      "step": 13871
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.445513603297033,
      "learning_rate": 1.5651113012044029e-09,
      "loss": 0.1596,
      "step": 13872
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3453218681076602,
      "learning_rate": 1.5362634190896341e-09,
      "loss": 0.1506,
      "step": 13873
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5851901722524955,
      "learning_rate": 1.5076838346500533e-09,
      "loss": 0.1384,
      "step": 13874
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.438759733122165,
      "learning_rate": 1.4793725494199884e-09,
      "loss": 0.1684,
      "step": 13875
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4078784053755165,
      "learning_rate": 1.4513295649193348e-09,
      "loss": 0.161,
      "step": 13876
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1438857995315674,
      "learning_rate": 1.423554882653e-09,
      "loss": 0.1333,
      "step": 13877
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2479457009002595,
      "learning_rate": 1.3960485041120131e-09,
      "loss": 0.1593,
      "step": 13878
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.014545653290866,
      "learning_rate": 1.3688104307729711e-09,
      "loss": 0.4985,
      "step": 13879
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2082669911508943,
      "learning_rate": 1.3418406640969272e-09,
      "loss": 0.1534,
      "step": 13880
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.173697454548059,
      "learning_rate": 1.315139205533278e-09,
      "loss": 0.1485,
      "step": 13881
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2608928752143274,
      "learning_rate": 1.2887060565136556e-09,
      "loss": 0.1598,
      "step": 13882
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3193664366532913,
      "learning_rate": 1.2625412184574803e-09,
      "loss": 0.1723,
      "step": 13883
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.302599573484244,
      "learning_rate": 1.2366446927691844e-09,
      "loss": 0.1784,
      "step": 13884
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4277753765424852,
      "learning_rate": 1.211016480838767e-09,
      "loss": 0.1668,
      "step": 13885
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3813149938073364,
      "learning_rate": 1.1856565840417944e-09,
      "loss": 0.1471,
      "step": 13886
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3540424568857683,
      "learning_rate": 1.1605650037394002e-09,
      "loss": 0.1735,
      "step": 13887
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.295677276055178,
      "learning_rate": 1.1357417412788396e-09,
      "loss": 0.1827,
      "step": 13888
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4712042296682597,
      "learning_rate": 1.1111867979923807e-09,
      "loss": 0.1677,
      "step": 13889
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3207283262469796,
      "learning_rate": 1.086900175197303e-09,
      "loss": 0.1943,
      "step": 13890
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3408684139390483,
      "learning_rate": 1.0628818741981184e-09,
      "loss": 0.1387,
      "step": 13891
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.378357771708658,
      "learning_rate": 1.0391318962837959e-09,
      "loss": 0.2034,
      "step": 13892
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3880528326760135,
      "learning_rate": 1.0156502427288717e-09,
      "loss": 0.1437,
      "step": 13893
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2072290797790888,
      "learning_rate": 9.924369147945589e-10,
      "loss": 0.1455,
      "step": 13894
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1302197174429163,
      "learning_rate": 9.694919137254177e-10,
      "loss": 0.1391,
      "step": 13895
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4240799404606188,
      "learning_rate": 9.468152407543508e-10,
      "loss": 0.1603,
      "step": 13896
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.506016209679943,
      "learning_rate": 9.244068970976072e-10,
      "loss": 0.4413,
      "step": 13897
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.126116456405039,
      "learning_rate": 9.022668839586691e-10,
      "loss": 0.6294,
      "step": 13898
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.6041474639039635,
      "learning_rate": 8.803952025260299e-10,
      "loss": 0.5389,
      "step": 13899
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4050807767214724,
      "learning_rate": 8.587918539726403e-10,
      "loss": 0.1506,
      "step": 13900
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4972876465146017,
      "learning_rate": 8.374568394592386e-10,
      "loss": 0.1827,
      "step": 13901
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2036911774184678,
      "learning_rate": 8.163901601304647e-10,
      "loss": 0.1414,
      "step": 13902
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4557849792409847,
      "learning_rate": 7.955918171170806e-10,
      "loss": 0.2152,
      "step": 13903
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4505874176139495,
      "learning_rate": 7.750618115354159e-10,
      "loss": 0.1862,
      "step": 13904
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2189394103247244,
      "learning_rate": 7.548001444879216e-10,
      "loss": 0.1643,
      "step": 13905
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4786841351789568,
      "learning_rate": 7.348068170615064e-10,
      "loss": 0.2165,
      "step": 13906
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.271113039067953,
      "learning_rate": 7.150818303297557e-10,
      "loss": 0.1695,
      "step": 13907
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2875075035786803,
      "learning_rate": 6.956251853512675e-10,
      "loss": 0.1775,
      "step": 13908
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3157150797300077,
      "learning_rate": 6.764368831707613e-10,
      "loss": 0.1924,
      "step": 13909
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4205952036871892,
      "learning_rate": 6.57516924817414e-10,
      "loss": 0.1931,
      "step": 13910
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.511636514646889,
      "learning_rate": 6.388653113076348e-10,
      "loss": 0.164,
      "step": 13911
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2997966283205473,
      "learning_rate": 6.204820436417347e-10,
      "loss": 0.1731,
      "step": 13912
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2816180364063396,
      "learning_rate": 6.02367122807257e-10,
      "loss": 0.1468,
      "step": 13913
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4587143702967096,
      "learning_rate": 5.845205497756467e-10,
      "loss": 0.1793,
      "step": 13914
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2829631467376925,
      "learning_rate": 5.669423255055817e-10,
      "loss": 0.1791,
      "step": 13915
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3280606742170686,
      "learning_rate": 5.496324509407513e-10,
      "loss": 0.129,
      "step": 13916
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4183881083256866,
      "learning_rate": 5.325909270093022e-10,
      "loss": 0.1633,
      "step": 13917
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.766115601224815,
      "learning_rate": 5.158177546271681e-10,
      "loss": 0.4822,
      "step": 13918
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3647217835434147,
      "learning_rate": 4.993129346936298e-10,
      "loss": 0.1768,
      "step": 13919
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.188324187214796,
      "learning_rate": 4.830764680946453e-10,
      "loss": 0.5553,
      "step": 13920
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3739583052308526,
      "learning_rate": 4.6710835570285e-10,
      "loss": 0.1926,
      "step": 13921
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.230569548273155,
      "learning_rate": 4.514085983742256e-10,
      "loss": 0.5354,
      "step": 13922
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.346658279551643,
      "learning_rate": 4.359771969519866e-10,
      "loss": 0.1643,
      "step": 13923
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2420010751890922,
      "learning_rate": 4.2081415226380427e-10,
      "loss": 0.1814,
      "step": 13924
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4550993678193778,
      "learning_rate": 4.0591946512458237e-10,
      "loss": 0.1689,
      "step": 13925
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.330650214688605,
      "learning_rate": 3.9129313633312626e-10,
      "loss": 0.1792,
      "step": 13926
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4145277310451065,
      "learning_rate": 3.769351666743637e-10,
      "loss": 0.1701,
      "step": 13927
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4874394205383876,
      "learning_rate": 3.6284555691989964e-10,
      "loss": 0.1481,
      "step": 13928
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4165259524333382,
      "learning_rate": 3.4902430782468577e-10,
      "loss": 0.1655,
      "step": 13929
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.936843726406026,
      "learning_rate": 3.354714201314613e-10,
      "loss": 0.6275,
      "step": 13930
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.283585215521246,
      "learning_rate": 3.2218689456797734e-10,
      "loss": 0.1462,
      "step": 13931
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.30010841347557,
      "learning_rate": 3.091707318464421e-10,
      "loss": 0.6043,
      "step": 13932
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3424364324041096,
      "learning_rate": 2.9642293266574087e-10,
      "loss": 0.1836,
      "step": 13933
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.348948289457645,
      "learning_rate": 2.8394349771088124e-10,
      "loss": 0.1765,
      "step": 13934
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.221247548853169,
      "learning_rate": 2.717324276507727e-10,
      "loss": 0.1421,
      "step": 13935
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.283334635554951,
      "learning_rate": 2.597897231410018e-10,
      "loss": 0.1518,
      "step": 13936
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.457703061820408,
      "learning_rate": 2.4811538482272246e-10,
      "loss": 0.1922,
      "step": 13937
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5056527336136298,
      "learning_rate": 2.3670941332321063e-10,
      "loss": 0.1894,
      "step": 13938
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.309650636904002,
      "learning_rate": 2.2557180925364408e-10,
      "loss": 0.1535,
      "step": 13939
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.635154066566935,
      "learning_rate": 2.1470257321298815e-10,
      "loss": 0.2301,
      "step": 13940
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2666979728499836,
      "learning_rate": 2.041017057829997e-10,
      "loss": 0.1621,
      "step": 13941
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3344421000365319,
      "learning_rate": 1.9376920753433338e-10,
      "loss": 0.147,
      "step": 13942
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1543823983652604,
      "learning_rate": 1.8370507902099045e-10,
      "loss": 0.1352,
      "step": 13943
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.445876798883753,
      "learning_rate": 1.739093207830944e-10,
      "loss": 0.5656,
      "step": 13944
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3932824054848654,
      "learning_rate": 1.643819333463359e-10,
      "loss": 0.1458,
      "step": 13945
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.44860105327731,
      "learning_rate": 1.5512291722252772e-10,
      "loss": 0.1824,
      "step": 13946
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.431555380354713,
      "learning_rate": 1.4613227290793953e-10,
      "loss": 0.1711,
      "step": 13947
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2979265552282082,
      "learning_rate": 1.374100008860735e-10,
      "loss": 0.146,
      "step": 13948
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2939053540419243,
      "learning_rate": 1.2895610162433347e-10,
      "loss": 0.1474,
      "step": 13949
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3326453605207067,
      "learning_rate": 1.207705755768007e-10,
      "loss": 0.1396,
      "step": 13950
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2714279412649456,
      "learning_rate": 1.1285342318312354e-10,
      "loss": 0.1828,
      "step": 13951
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5078356970200613,
      "learning_rate": 1.0520464486796222e-10,
      "loss": 0.1994,
      "step": 13952
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3478369737524984,
      "learning_rate": 9.782424104154419e-11,
      "loss": 0.1361,
      "step": 13953
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2866285511869955,
      "learning_rate": 9.071221210021908e-11,
      "loss": 0.1673,
      "step": 13954
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3372848353363207,
      "learning_rate": 8.386855842645869e-11,
      "loss": 0.1735,
      "step": 13955
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1863449345978516,
      "learning_rate": 7.729328038663663e-11,
      "loss": 0.1565,
      "step": 13956
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.238621027931929,
      "learning_rate": 7.098637833435895e-11,
      "loss": 0.1389,
      "step": 13957
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3421719137941692,
      "learning_rate": 6.494785260768854e-11,
      "loss": 0.2053,
      "step": 13958
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2767878594519209,
      "learning_rate": 5.917770353081053e-11,
      "loss": 0.1573,
      "step": 13959
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.514873896403063,
      "learning_rate": 5.3675931413477156e-11,
      "loss": 0.2187,
      "step": 13960
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.250040612724496,
      "learning_rate": 4.8442536551562837e-11,
      "loss": 0.1793,
      "step": 13961
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5591863753386381,
      "learning_rate": 4.347751922539889e-11,
      "loss": 0.1735,
      "step": 13962
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5153531766594353,
      "learning_rate": 3.878087970143884e-11,
      "loss": 0.1795,
      "step": 13963
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4087817224091694,
      "learning_rate": 3.435261823170333e-11,
      "loss": 0.1428,
      "step": 13964
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.547560325445944,
      "learning_rate": 3.019273505433518e-11,
      "loss": 0.4143,
      "step": 13965
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2956547619507537,
      "learning_rate": 2.6301230392489218e-11,
      "loss": 0.1594,
      "step": 13966
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.438946622684075,
      "learning_rate": 2.2678104454887383e-11,
      "loss": 0.1757,
      "step": 13967
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.35647642891667,
      "learning_rate": 1.93233574358187e-11,
      "loss": 0.1653,
      "step": 13968
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2487794628871496,
      "learning_rate": 1.6236989515694412e-11,
      "loss": 0.1473,
      "step": 13969
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.314917513225307,
      "learning_rate": 1.341900086049286e-11,
      "loss": 0.1733,
      "step": 13970
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3471196497156284,
      "learning_rate": 1.0869391620649262e-11,
      "loss": 0.1766,
      "step": 13971
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.234502324479324,
      "learning_rate": 8.588161933831273e-12,
      "loss": 0.1506,
      "step": 13972
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2963158923361942,
      "learning_rate": 6.5753119216083184e-12,
      "loss": 0.1415,
      "step": 13973
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3816031459170275,
      "learning_rate": 4.830841692782251e-12,
      "loss": 0.1777,
      "step": 13974
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5194552022525774,
      "learning_rate": 3.3547513406118058e-12,
      "loss": 0.1604,
      "step": 13975
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2988785591839587,
      "learning_rate": 2.1470409444779295e-12,
      "loss": 0.1909,
      "step": 13976
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3785665351758758,
      "learning_rate": 1.207710568773557e-12,
      "loss": 0.1628,
      "step": 13977
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2609014213578345,
      "learning_rate": 5.367602651240589e-13,
      "loss": 0.1749,
      "step": 13978
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3459893667371516,
      "learning_rate": 1.3419006794634925e-13,
      "loss": 0.1227,
      "step": 13979
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.5220905301420165,
      "learning_rate": 0.0,
      "loss": 0.3333,
      "step": 13980
    },
    {
      "epoch": 1.0,
      "step": 13980,
      "total_flos": 6714736945350656.0,
      "train_loss": 0.23951792627871463,
      "train_runtime": 45097.9201,
      "train_samples_per_second": 4.96,
      "train_steps_per_second": 0.31
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 13980,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "total_flos": 6714736945350656.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}