{
  "best_metric": 1.2860368490219116,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.837696335078534,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004188481675392671,
      "grad_norm": 3.926077127456665,
      "learning_rate": 1e-05,
      "loss": 2.0879,
      "step": 1
    },
    {
      "epoch": 0.004188481675392671,
      "eval_loss": 2.2211415767669678,
      "eval_runtime": 28.6548,
      "eval_samples_per_second": 14.064,
      "eval_steps_per_second": 3.525,
      "step": 1
    },
    {
      "epoch": 0.008376963350785341,
      "grad_norm": 5.116100788116455,
      "learning_rate": 2e-05,
      "loss": 2.1233,
      "step": 2
    },
    {
      "epoch": 0.012565445026178011,
      "grad_norm": 4.467835426330566,
      "learning_rate": 3e-05,
      "loss": 2.0685,
      "step": 3
    },
    {
      "epoch": 0.016753926701570682,
      "grad_norm": 3.099653482437134,
      "learning_rate": 4e-05,
      "loss": 2.0546,
      "step": 4
    },
    {
      "epoch": 0.020942408376963352,
      "grad_norm": 2.941957712173462,
      "learning_rate": 5e-05,
      "loss": 1.9169,
      "step": 5
    },
    {
      "epoch": 0.025130890052356022,
      "grad_norm": 1.8964356184005737,
      "learning_rate": 6e-05,
      "loss": 1.7972,
      "step": 6
    },
    {
      "epoch": 0.02931937172774869,
      "grad_norm": 0.822271466255188,
      "learning_rate": 7e-05,
      "loss": 1.6816,
      "step": 7
    },
    {
      "epoch": 0.033507853403141365,
      "grad_norm": 0.7898607850074768,
      "learning_rate": 8e-05,
      "loss": 1.692,
      "step": 8
    },
    {
      "epoch": 0.03769633507853403,
      "grad_norm": 0.8590221405029297,
      "learning_rate": 9e-05,
      "loss": 1.5978,
      "step": 9
    },
    {
      "epoch": 0.041884816753926704,
      "grad_norm": 0.6750182509422302,
      "learning_rate": 0.0001,
      "loss": 1.5605,
      "step": 10
    },
    {
      "epoch": 0.04607329842931937,
      "grad_norm": 0.5955801010131836,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.5164,
      "step": 11
    },
    {
      "epoch": 0.050261780104712044,
      "grad_norm": 0.41977617144584656,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.4814,
      "step": 12
    },
    {
      "epoch": 0.05445026178010471,
      "grad_norm": 0.6115280389785767,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.5194,
      "step": 13
    },
    {
      "epoch": 0.05863874345549738,
      "grad_norm": 0.458881676197052,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.5086,
      "step": 14
    },
    {
      "epoch": 0.06282722513089005,
      "grad_norm": 0.558876097202301,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.4856,
      "step": 15
    },
    {
      "epoch": 0.06701570680628273,
      "grad_norm": 0.45572343468666077,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.5626,
      "step": 16
    },
    {
      "epoch": 0.0712041884816754,
      "grad_norm": 0.5151314735412598,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.4417,
      "step": 17
    },
    {
      "epoch": 0.07539267015706806,
      "grad_norm": 0.4471381902694702,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.4668,
      "step": 18
    },
    {
      "epoch": 0.07958115183246073,
      "grad_norm": 0.4528507888317108,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.4522,
      "step": 19
    },
    {
      "epoch": 0.08376963350785341,
      "grad_norm": 0.4718516767024994,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.5029,
      "step": 20
    },
    {
      "epoch": 0.08795811518324607,
      "grad_norm": 0.41982200741767883,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.4611,
      "step": 21
    },
    {
      "epoch": 0.09214659685863874,
      "grad_norm": 0.3980315327644348,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.3786,
      "step": 22
    },
    {
      "epoch": 0.09633507853403141,
      "grad_norm": 0.4060964286327362,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.3963,
      "step": 23
    },
    {
      "epoch": 0.10052356020942409,
      "grad_norm": 0.3943502902984619,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.3841,
      "step": 24
    },
    {
      "epoch": 0.10471204188481675,
      "grad_norm": 0.40681180357933044,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.4104,
      "step": 25
    },
    {
      "epoch": 0.10890052356020942,
      "grad_norm": 0.4119963049888611,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.3985,
      "step": 26
    },
    {
      "epoch": 0.1130890052356021,
      "grad_norm": 0.42287012934684753,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.4265,
      "step": 27
    },
    {
      "epoch": 0.11727748691099477,
      "grad_norm": 0.4382915794849396,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.3904,
      "step": 28
    },
    {
      "epoch": 0.12146596858638743,
      "grad_norm": 0.45694679021835327,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.4149,
      "step": 29
    },
    {
      "epoch": 0.1256544502617801,
      "grad_norm": 0.48638594150543213,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.39,
      "step": 30
    },
    {
      "epoch": 0.12984293193717278,
      "grad_norm": 0.5300375819206238,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.3883,
      "step": 31
    },
    {
      "epoch": 0.13403141361256546,
      "grad_norm": 0.5210663676261902,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.4515,
      "step": 32
    },
    {
      "epoch": 0.1382198952879581,
      "grad_norm": 0.5227262377738953,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.3944,
      "step": 33
    },
    {
      "epoch": 0.1424083769633508,
      "grad_norm": 0.5467104315757751,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.4712,
      "step": 34
    },
    {
      "epoch": 0.14659685863874344,
      "grad_norm": 0.5446538329124451,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.3683,
      "step": 35
    },
    {
      "epoch": 0.15078534031413612,
      "grad_norm": 0.5889321565628052,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.3675,
      "step": 36
    },
    {
      "epoch": 0.1549738219895288,
      "grad_norm": 0.5657311677932739,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.3347,
      "step": 37
    },
    {
      "epoch": 0.15916230366492146,
      "grad_norm": 0.6621028780937195,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.3371,
      "step": 38
    },
    {
      "epoch": 0.16335078534031414,
      "grad_norm": 0.5807614326477051,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.3406,
      "step": 39
    },
    {
      "epoch": 0.16753926701570682,
      "grad_norm": 0.5891103744506836,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.3347,
      "step": 40
    },
    {
      "epoch": 0.17172774869109947,
      "grad_norm": 0.6490166187286377,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.3791,
      "step": 41
    },
    {
      "epoch": 0.17591623036649215,
      "grad_norm": 0.670409083366394,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.3146,
      "step": 42
    },
    {
      "epoch": 0.18010471204188483,
      "grad_norm": 0.6172012686729431,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.3258,
      "step": 43
    },
    {
      "epoch": 0.18429319371727748,
      "grad_norm": 0.731728196144104,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.3535,
      "step": 44
    },
    {
      "epoch": 0.18848167539267016,
      "grad_norm": 0.6701034307479858,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.3067,
      "step": 45
    },
    {
      "epoch": 0.19267015706806281,
      "grad_norm": 0.7045776844024658,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.2787,
      "step": 46
    },
    {
      "epoch": 0.1968586387434555,
      "grad_norm": 0.7145847082138062,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.3167,
      "step": 47
    },
    {
      "epoch": 0.20104712041884817,
      "grad_norm": 0.7663804292678833,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.2806,
      "step": 48
    },
    {
      "epoch": 0.20523560209424083,
      "grad_norm": 0.9702144265174866,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.3609,
      "step": 49
    },
    {
      "epoch": 0.2094240837696335,
      "grad_norm": 1.0464814901351929,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.2305,
      "step": 50
    },
    {
      "epoch": 0.2094240837696335,
      "eval_loss": 1.4609953165054321,
      "eval_runtime": 29.0147,
      "eval_samples_per_second": 13.889,
      "eval_steps_per_second": 3.481,
      "step": 50
    },
    {
      "epoch": 0.2136125654450262,
      "grad_norm": 0.7326249480247498,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.5822,
      "step": 51
    },
    {
      "epoch": 0.21780104712041884,
      "grad_norm": 0.6222273111343384,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.4945,
      "step": 52
    },
    {
      "epoch": 0.22198952879581152,
      "grad_norm": 0.4404517710208893,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.4369,
      "step": 53
    },
    {
      "epoch": 0.2261780104712042,
      "grad_norm": 0.3408048450946808,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.3665,
      "step": 54
    },
    {
      "epoch": 0.23036649214659685,
      "grad_norm": 0.3527311384677887,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.4797,
      "step": 55
    },
    {
      "epoch": 0.23455497382198953,
      "grad_norm": 0.34772562980651855,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.3664,
      "step": 56
    },
    {
      "epoch": 0.2387434554973822,
      "grad_norm": 0.31533706188201904,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.358,
      "step": 57
    },
    {
      "epoch": 0.24293193717277486,
      "grad_norm": 0.28406214714050293,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.3421,
      "step": 58
    },
    {
      "epoch": 0.24712041884816754,
      "grad_norm": 0.297507107257843,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.3763,
      "step": 59
    },
    {
      "epoch": 0.2513089005235602,
      "grad_norm": 0.32910609245300293,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.3574,
      "step": 60
    },
    {
      "epoch": 0.2554973821989529,
      "grad_norm": 0.3502568304538727,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.4294,
      "step": 61
    },
    {
      "epoch": 0.25968586387434556,
      "grad_norm": 0.31924673914909363,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.3464,
      "step": 62
    },
    {
      "epoch": 0.2638743455497382,
      "grad_norm": 0.3037525415420532,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.2925,
      "step": 63
    },
    {
      "epoch": 0.2680628272251309,
      "grad_norm": 0.30458396673202515,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.3565,
      "step": 64
    },
    {
      "epoch": 0.27225130890052357,
      "grad_norm": 0.31747856736183167,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.3534,
      "step": 65
    },
    {
      "epoch": 0.2764397905759162,
      "grad_norm": 0.3151717782020569,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.3482,
      "step": 66
    },
    {
      "epoch": 0.2806282722513089,
      "grad_norm": 0.3036843538284302,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.3459,
      "step": 67
    },
    {
      "epoch": 0.2848167539267016,
      "grad_norm": 0.3110540211200714,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.3271,
      "step": 68
    },
    {
      "epoch": 0.28900523560209423,
      "grad_norm": 0.3139282763004303,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.3885,
      "step": 69
    },
    {
      "epoch": 0.2931937172774869,
      "grad_norm": 0.32266420125961304,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.3249,
      "step": 70
    },
    {
      "epoch": 0.2973821989528796,
      "grad_norm": 0.3372081518173218,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.2639,
      "step": 71
    },
    {
      "epoch": 0.30157068062827225,
      "grad_norm": 0.3430722951889038,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.3682,
      "step": 72
    },
    {
      "epoch": 0.3057591623036649,
      "grad_norm": 0.35122257471084595,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.3369,
      "step": 73
    },
    {
      "epoch": 0.3099476439790576,
      "grad_norm": 0.3526090085506439,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.3259,
      "step": 74
    },
    {
      "epoch": 0.31413612565445026,
      "grad_norm": 0.37421196699142456,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.3914,
      "step": 75
    },
    {
      "epoch": 0.3183246073298429,
      "grad_norm": 0.37143388390541077,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.4003,
      "step": 76
    },
    {
      "epoch": 0.3225130890052356,
      "grad_norm": 0.3665562868118286,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.2909,
      "step": 77
    },
    {
      "epoch": 0.3267015706806283,
      "grad_norm": 0.3665645122528076,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.3176,
      "step": 78
    },
    {
      "epoch": 0.3308900523560209,
      "grad_norm": 0.3843286335468292,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.2935,
      "step": 79
    },
    {
      "epoch": 0.33507853403141363,
      "grad_norm": 0.42542752623558044,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.3432,
      "step": 80
    },
    {
      "epoch": 0.3392670157068063,
      "grad_norm": 0.4372069239616394,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.3206,
      "step": 81
    },
    {
      "epoch": 0.34345549738219894,
      "grad_norm": 0.46938851475715637,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.3445,
      "step": 82
    },
    {
      "epoch": 0.34764397905759165,
      "grad_norm": 0.46583279967308044,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.3522,
      "step": 83
    },
    {
      "epoch": 0.3518324607329843,
      "grad_norm": 0.5285698771476746,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.3371,
      "step": 84
    },
    {
      "epoch": 0.35602094240837695,
      "grad_norm": 0.5335991382598877,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.3201,
      "step": 85
    },
    {
      "epoch": 0.36020942408376966,
      "grad_norm": 0.48621413111686707,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.2709,
      "step": 86
    },
    {
      "epoch": 0.3643979057591623,
      "grad_norm": 0.5229520201683044,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.2761,
      "step": 87
    },
    {
      "epoch": 0.36858638743455496,
      "grad_norm": 0.5092804431915283,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.3199,
      "step": 88
    },
    {
      "epoch": 0.37277486910994767,
      "grad_norm": 0.5382634401321411,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.3828,
      "step": 89
    },
    {
      "epoch": 0.3769633507853403,
      "grad_norm": 0.5918077230453491,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.2516,
      "step": 90
    },
    {
      "epoch": 0.381151832460733,
      "grad_norm": 0.5466107726097107,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.2599,
      "step": 91
    },
    {
      "epoch": 0.38534031413612563,
      "grad_norm": 0.61199951171875,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.3207,
      "step": 92
    },
    {
      "epoch": 0.38952879581151834,
      "grad_norm": 0.6159107685089111,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.2706,
      "step": 93
    },
    {
      "epoch": 0.393717277486911,
      "grad_norm": 0.6025269627571106,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.3188,
      "step": 94
    },
    {
      "epoch": 0.39790575916230364,
      "grad_norm": 0.6132339239120483,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.233,
      "step": 95
    },
    {
      "epoch": 0.40209424083769635,
      "grad_norm": 0.6684319972991943,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.3105,
      "step": 96
    },
    {
      "epoch": 0.406282722513089,
      "grad_norm": 0.7070481181144714,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.2007,
      "step": 97
    },
    {
      "epoch": 0.41047120418848165,
      "grad_norm": 0.7131744027137756,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.1659,
      "step": 98
    },
    {
      "epoch": 0.41465968586387436,
      "grad_norm": 0.7557907104492188,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.2086,
      "step": 99
    },
    {
      "epoch": 0.418848167539267,
      "grad_norm": 1.0365667343139648,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.0911,
      "step": 100
    },
    {
      "epoch": 0.418848167539267,
      "eval_loss": 1.351357102394104,
      "eval_runtime": 28.9204,
      "eval_samples_per_second": 13.935,
      "eval_steps_per_second": 3.492,
      "step": 100
    },
    {
      "epoch": 0.42303664921465967,
      "grad_norm": 0.5068424940109253,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.4539,
      "step": 101
    },
    {
      "epoch": 0.4272251308900524,
      "grad_norm": 0.38696298003196716,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.3384,
      "step": 102
    },
    {
      "epoch": 0.431413612565445,
      "grad_norm": 0.3579586148262024,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.3704,
      "step": 103
    },
    {
      "epoch": 0.4356020942408377,
      "grad_norm": 0.3296695947647095,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.406,
      "step": 104
    },
    {
      "epoch": 0.4397905759162304,
      "grad_norm": 0.317214697599411,
      "learning_rate": 5e-05,
      "loss": 1.3961,
      "step": 105
    },
    {
      "epoch": 0.44397905759162304,
      "grad_norm": 0.304355263710022,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.3664,
      "step": 106
    },
    {
      "epoch": 0.4481675392670157,
      "grad_norm": 0.2801881730556488,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.338,
      "step": 107
    },
    {
      "epoch": 0.4523560209424084,
      "grad_norm": 0.2879219949245453,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.3735,
      "step": 108
    },
    {
      "epoch": 0.45654450261780105,
      "grad_norm": 0.28712937235832214,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.3113,
      "step": 109
    },
    {
      "epoch": 0.4607329842931937,
      "grad_norm": 0.2836691737174988,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.3791,
      "step": 110
    },
    {
      "epoch": 0.4649214659685864,
      "grad_norm": 0.3011457026004791,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.2896,
      "step": 111
    },
    {
      "epoch": 0.46910994764397906,
      "grad_norm": 0.303462952375412,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.3243,
      "step": 112
    },
    {
      "epoch": 0.4732984293193717,
      "grad_norm": 0.31702524423599243,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.2916,
      "step": 113
    },
    {
      "epoch": 0.4774869109947644,
      "grad_norm": 0.3053862154483795,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.2521,
      "step": 114
    },
    {
      "epoch": 0.4816753926701571,
      "grad_norm": 0.31785744428634644,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.3485,
      "step": 115
    },
    {
      "epoch": 0.48586387434554973,
      "grad_norm": 0.32555049657821655,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.3108,
      "step": 116
    },
    {
      "epoch": 0.4900523560209424,
      "grad_norm": 0.3192926347255707,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.3012,
      "step": 117
    },
    {
      "epoch": 0.4942408376963351,
      "grad_norm": 0.3216697871685028,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.3056,
      "step": 118
    },
    {
      "epoch": 0.49842931937172774,
      "grad_norm": 0.3411600887775421,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.3146,
      "step": 119
    },
    {
      "epoch": 0.5026178010471204,
      "grad_norm": 0.3299861550331116,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.2334,
      "step": 120
    },
    {
      "epoch": 0.506806282722513,
      "grad_norm": 0.33841753005981445,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.3606,
      "step": 121
    },
    {
      "epoch": 0.5109947643979058,
      "grad_norm": 0.34549593925476074,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.2903,
      "step": 122
    },
    {
      "epoch": 0.5151832460732985,
      "grad_norm": 0.34298786520957947,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.334,
      "step": 123
    },
    {
      "epoch": 0.5193717277486911,
      "grad_norm": 0.3591674268245697,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.3375,
      "step": 124
    },
    {
      "epoch": 0.5235602094240838,
      "grad_norm": 0.365177720785141,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.2256,
      "step": 125
    },
    {
      "epoch": 0.5277486910994764,
      "grad_norm": 0.3893211781978607,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.3821,
      "step": 126
    },
    {
      "epoch": 0.5319371727748691,
      "grad_norm": 0.3771529197692871,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.2985,
      "step": 127
    },
    {
      "epoch": 0.5361256544502618,
      "grad_norm": 0.3897879123687744,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.2558,
      "step": 128
    },
    {
      "epoch": 0.5403141361256545,
      "grad_norm": 0.4145451784133911,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.3072,
      "step": 129
    },
    {
      "epoch": 0.5445026178010471,
      "grad_norm": 0.4503288269042969,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.302,
      "step": 130
    },
    {
      "epoch": 0.5486910994764398,
      "grad_norm": 0.4462653696537018,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.2795,
      "step": 131
    },
    {
      "epoch": 0.5528795811518324,
      "grad_norm": 0.4874064028263092,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.2657,
      "step": 132
    },
    {
      "epoch": 0.5570680628272251,
      "grad_norm": 0.466003954410553,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.2944,
      "step": 133
    },
    {
      "epoch": 0.5612565445026177,
      "grad_norm": 0.49740466475486755,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.2326,
      "step": 134
    },
    {
      "epoch": 0.5654450261780105,
      "grad_norm": 0.49777016043663025,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.2254,
      "step": 135
    },
    {
      "epoch": 0.5696335078534032,
      "grad_norm": 0.49835050106048584,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.3042,
      "step": 136
    },
    {
      "epoch": 0.5738219895287958,
      "grad_norm": 0.5383933782577515,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.2733,
      "step": 137
    },
    {
      "epoch": 0.5780104712041885,
      "grad_norm": 0.5482964515686035,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.2498,
      "step": 138
    },
    {
      "epoch": 0.5821989528795811,
      "grad_norm": 0.5489109754562378,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.178,
      "step": 139
    },
    {
      "epoch": 0.5863874345549738,
      "grad_norm": 0.5807023644447327,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.264,
      "step": 140
    },
    {
      "epoch": 0.5905759162303665,
      "grad_norm": 0.5718259811401367,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.1922,
      "step": 141
    },
    {
      "epoch": 0.5947643979057592,
      "grad_norm": 0.5809677243232727,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.1855,
      "step": 142
    },
    {
      "epoch": 0.5989528795811518,
      "grad_norm": 0.609746515750885,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.2812,
      "step": 143
    },
    {
      "epoch": 0.6031413612565445,
      "grad_norm": 0.6022087931632996,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.182,
      "step": 144
    },
    {
      "epoch": 0.6073298429319371,
      "grad_norm": 0.5959063768386841,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.1689,
      "step": 145
    },
    {
      "epoch": 0.6115183246073298,
      "grad_norm": 0.6428928971290588,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.2292,
      "step": 146
    },
    {
      "epoch": 0.6157068062827226,
      "grad_norm": 0.7082703709602356,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.1795,
      "step": 147
    },
    {
      "epoch": 0.6198952879581152,
      "grad_norm": 0.7620861530303955,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.1711,
      "step": 148
    },
    {
      "epoch": 0.6240837696335079,
      "grad_norm": 0.744573175907135,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.1737,
      "step": 149
    },
    {
      "epoch": 0.6282722513089005,
      "grad_norm": 0.8863393068313599,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.9523,
      "step": 150
    },
    {
      "epoch": 0.6282722513089005,
      "eval_loss": 1.3025740385055542,
      "eval_runtime": 28.9037,
      "eval_samples_per_second": 13.943,
      "eval_steps_per_second": 3.494,
      "step": 150
    },
    {
      "epoch": 0.6324607329842932,
      "grad_norm": 0.36740434169769287,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.3556,
      "step": 151
    },
    {
      "epoch": 0.6366492146596858,
      "grad_norm": 0.31871405243873596,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.3485,
      "step": 152
    },
    {
      "epoch": 0.6408376963350786,
      "grad_norm": 0.3007407486438751,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.395,
      "step": 153
    },
    {
      "epoch": 0.6450261780104712,
      "grad_norm": 0.3127298653125763,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.3471,
      "step": 154
    },
    {
      "epoch": 0.6492146596858639,
      "grad_norm": 0.29624295234680176,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.3299,
      "step": 155
    },
    {
      "epoch": 0.6534031413612565,
      "grad_norm": 0.303844690322876,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.3282,
      "step": 156
    },
    {
      "epoch": 0.6575916230366492,
      "grad_norm": 0.31024369597435,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.3696,
      "step": 157
    },
    {
      "epoch": 0.6617801047120419,
      "grad_norm": 0.29527774453163147,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.3166,
      "step": 158
    },
    {
      "epoch": 0.6659685863874345,
      "grad_norm": 0.29768288135528564,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.3244,
      "step": 159
    },
    {
      "epoch": 0.6701570680628273,
      "grad_norm": 0.3053095042705536,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.2835,
      "step": 160
    },
    {
      "epoch": 0.6743455497382199,
      "grad_norm": 0.2892249524593353,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.2986,
      "step": 161
    },
    {
      "epoch": 0.6785340314136126,
      "grad_norm": 0.3029867112636566,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.2818,
      "step": 162
    },
    {
      "epoch": 0.6827225130890052,
      "grad_norm": 0.3176477253437042,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.2575,
      "step": 163
    },
    {
      "epoch": 0.6869109947643979,
      "grad_norm": 0.3086993098258972,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.244,
      "step": 164
    },
    {
      "epoch": 0.6910994764397905,
      "grad_norm": 0.3168316185474396,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.2918,
      "step": 165
    },
    {
      "epoch": 0.6952879581151833,
      "grad_norm": 0.31938818097114563,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.2826,
      "step": 166
    },
    {
      "epoch": 0.6994764397905759,
      "grad_norm": 0.31262513995170593,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.2324,
      "step": 167
    },
    {
      "epoch": 0.7036649214659686,
      "grad_norm": 0.33030468225479126,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.3704,
      "step": 168
    },
    {
      "epoch": 0.7078534031413612,
      "grad_norm": 0.3215632736682892,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.3044,
      "step": 169
    },
    {
      "epoch": 0.7120418848167539,
      "grad_norm": 0.3407300114631653,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.2733,
      "step": 170
    },
    {
      "epoch": 0.7162303664921466,
      "grad_norm": 0.35280951857566833,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.319,
      "step": 171
    },
    {
      "epoch": 0.7204188481675393,
      "grad_norm": 0.3457036316394806,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.2758,
      "step": 172
    },
    {
      "epoch": 0.724607329842932,
      "grad_norm": 0.3777696490287781,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.2739,
      "step": 173
    },
    {
      "epoch": 0.7287958115183246,
      "grad_norm": 0.36770495772361755,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.2434,
      "step": 174
    },
    {
      "epoch": 0.7329842931937173,
      "grad_norm": 0.3864113688468933,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.2795,
      "step": 175
    },
    {
      "epoch": 0.7371727748691099,
      "grad_norm": 0.39327654242515564,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.3229,
      "step": 176
    },
    {
      "epoch": 0.7413612565445026,
      "grad_norm": 0.3867397904396057,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.2473,
      "step": 177
    },
    {
      "epoch": 0.7455497382198953,
      "grad_norm": 0.40516242384910583,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.2815,
      "step": 178
    },
    {
      "epoch": 0.749738219895288,
      "grad_norm": 0.4050538241863251,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.2551,
      "step": 179
    },
    {
      "epoch": 0.7539267015706806,
      "grad_norm": 0.4257807731628418,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.3629,
      "step": 180
    },
    {
      "epoch": 0.7581151832460733,
      "grad_norm": 0.4621107280254364,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.4015,
      "step": 181
    },
    {
      "epoch": 0.762303664921466,
      "grad_norm": 0.49346381425857544,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.2686,
      "step": 182
    },
    {
      "epoch": 0.7664921465968586,
      "grad_norm": 0.5245895385742188,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.2762,
      "step": 183
    },
    {
      "epoch": 0.7706806282722513,
      "grad_norm": 0.5255253314971924,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.2838,
      "step": 184
    },
    {
      "epoch": 0.774869109947644,
      "grad_norm": 0.5585934519767761,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.2086,
      "step": 185
    },
    {
      "epoch": 0.7790575916230367,
      "grad_norm": 0.5777465105056763,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.1868,
      "step": 186
    },
    {
      "epoch": 0.7832460732984293,
      "grad_norm": 0.5840594172477722,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.2256,
      "step": 187
    },
    {
      "epoch": 0.787434554973822,
      "grad_norm": 0.6179370880126953,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.2439,
      "step": 188
    },
    {
      "epoch": 0.7916230366492146,
      "grad_norm": 0.5706912875175476,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.2189,
      "step": 189
    },
    {
      "epoch": 0.7958115183246073,
      "grad_norm": 0.6214474439620972,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.2372,
      "step": 190
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6089590787887573,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.1911,
      "step": 191
    },
    {
      "epoch": 0.8041884816753927,
      "grad_norm": 0.6615989804267883,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.1566,
      "step": 192
    },
    {
      "epoch": 0.8083769633507853,
      "grad_norm": 0.6192319989204407,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.2414,
      "step": 193
    },
    {
      "epoch": 0.812565445026178,
      "grad_norm": 0.6306219696998596,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.1748,
      "step": 194
    },
    {
      "epoch": 0.8167539267015707,
      "grad_norm": 0.6606384515762329,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.2203,
      "step": 195
    },
    {
      "epoch": 0.8209424083769633,
      "grad_norm": 0.6851112842559814,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.2559,
      "step": 196
    },
    {
      "epoch": 0.8251308900523561,
      "grad_norm": 0.6846991777420044,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.2011,
      "step": 197
    },
    {
      "epoch": 0.8293193717277487,
      "grad_norm": 0.7207460403442383,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.0395,
      "step": 198
    },
    {
      "epoch": 0.8335078534031414,
      "grad_norm": 0.7653758525848389,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.0862,
      "step": 199
    },
    {
      "epoch": 0.837696335078534,
      "grad_norm": 0.904840350151062,
      "learning_rate": 0.0,
      "loss": 0.9664,
      "step": 200
    },
    {
      "epoch": 0.837696335078534,
      "eval_loss": 1.2860368490219116,
      "eval_runtime": 29.217,
      "eval_samples_per_second": 13.793,
      "eval_steps_per_second": 3.457,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.684081292705792e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}