{
  "best_metric": 1.339070439338684,
  "best_model_checkpoint": "miner_id_24/checkpoint-100",
  "epoch": 2.9197080291970803,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.014598540145985401,
      "grad_norm": 0.3314630389213562,
      "learning_rate": 1e-05,
      "loss": 1.181,
      "step": 1
    },
    {
      "epoch": 0.014598540145985401,
      "eval_loss": 1.5133534669876099,
      "eval_runtime": 9.2058,
      "eval_samples_per_second": 12.601,
      "eval_steps_per_second": 3.15,
      "step": 1
    },
    {
      "epoch": 0.029197080291970802,
      "grad_norm": 0.22670935094356537,
      "learning_rate": 2e-05,
      "loss": 1.4452,
      "step": 2
    },
    {
      "epoch": 0.043795620437956206,
      "grad_norm": 0.3789464831352234,
      "learning_rate": 3e-05,
      "loss": 1.4232,
      "step": 3
    },
    {
      "epoch": 0.058394160583941604,
      "grad_norm": 0.2621401250362396,
      "learning_rate": 4e-05,
      "loss": 1.4073,
      "step": 4
    },
    {
      "epoch": 0.072992700729927,
      "grad_norm": 0.25601255893707275,
      "learning_rate": 5e-05,
      "loss": 1.344,
      "step": 5
    },
    {
      "epoch": 0.08759124087591241,
      "grad_norm": 0.2619723677635193,
      "learning_rate": 6e-05,
      "loss": 1.3306,
      "step": 6
    },
    {
      "epoch": 0.10218978102189781,
      "grad_norm": 0.2971075177192688,
      "learning_rate": 7e-05,
      "loss": 1.4307,
      "step": 7
    },
    {
      "epoch": 0.11678832116788321,
      "grad_norm": 0.3079770803451538,
      "learning_rate": 8e-05,
      "loss": 1.5267,
      "step": 8
    },
    {
      "epoch": 0.13138686131386862,
      "grad_norm": 0.3286444842815399,
      "learning_rate": 9e-05,
      "loss": 1.5575,
      "step": 9
    },
    {
      "epoch": 0.145985401459854,
      "grad_norm": 0.31285712122917175,
      "learning_rate": 0.0001,
      "loss": 1.5352,
      "step": 10
    },
    {
      "epoch": 0.16058394160583941,
      "grad_norm": 0.34175583720207214,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.3777,
      "step": 11
    },
    {
      "epoch": 0.17518248175182483,
      "grad_norm": 0.3486191928386688,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.5109,
      "step": 12
    },
    {
      "epoch": 0.1897810218978102,
      "grad_norm": 0.3993035852909088,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.6024,
      "step": 13
    },
    {
      "epoch": 0.20437956204379562,
      "grad_norm": 0.4641835391521454,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.6562,
      "step": 14
    },
    {
      "epoch": 0.21897810218978103,
      "grad_norm": 0.5170352458953857,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.5123,
      "step": 15
    },
    {
      "epoch": 0.23357664233576642,
      "grad_norm": 0.7671294212341309,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.5528,
      "step": 16
    },
    {
      "epoch": 0.24817518248175183,
      "grad_norm": 1.0964692831039429,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.5627,
      "step": 17
    },
    {
      "epoch": 0.26277372262773724,
      "grad_norm": 0.22408375144004822,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.1167,
      "step": 18
    },
    {
      "epoch": 0.2773722627737226,
      "grad_norm": 0.27535220980644226,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.3162,
      "step": 19
    },
    {
      "epoch": 0.291970802919708,
      "grad_norm": 0.32067593932151794,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.4931,
      "step": 20
    },
    {
      "epoch": 0.30656934306569344,
      "grad_norm": 0.2732066810131073,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.4165,
      "step": 21
    },
    {
      "epoch": 0.32116788321167883,
      "grad_norm": 0.2587907314300537,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.3312,
      "step": 22
    },
    {
      "epoch": 0.3357664233576642,
      "grad_norm": 0.27388066053390503,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.2406,
      "step": 23
    },
    {
      "epoch": 0.35036496350364965,
      "grad_norm": 0.2729540467262268,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.2964,
      "step": 24
    },
    {
      "epoch": 0.36496350364963503,
      "grad_norm": 0.2680898904800415,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.3259,
      "step": 25
    },
    {
      "epoch": 0.3795620437956204,
      "grad_norm": 0.27648648619651794,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.5216,
      "step": 26
    },
    {
      "epoch": 0.39416058394160586,
      "grad_norm": 0.2837010324001312,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.4691,
      "step": 27
    },
    {
      "epoch": 0.40875912408759124,
      "grad_norm": 0.30292704701423645,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.4608,
      "step": 28
    },
    {
      "epoch": 0.4233576642335766,
      "grad_norm": 0.302458792924881,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.2839,
      "step": 29
    },
    {
      "epoch": 0.43795620437956206,
      "grad_norm": 0.3292553424835205,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.5101,
      "step": 30
    },
    {
      "epoch": 0.45255474452554745,
      "grad_norm": 0.4009181559085846,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.5497,
      "step": 31
    },
    {
      "epoch": 0.46715328467153283,
      "grad_norm": 0.42407190799713135,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.3821,
      "step": 32
    },
    {
      "epoch": 0.48175182481751827,
      "grad_norm": 0.595664918422699,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.2652,
      "step": 33
    },
    {
      "epoch": 0.49635036496350365,
      "grad_norm": 1.1629176139831543,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.7652,
      "step": 34
    },
    {
      "epoch": 0.5109489051094891,
      "grad_norm": 0.23664163053035736,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.0537,
      "step": 35
    },
    {
      "epoch": 0.5255474452554745,
      "grad_norm": 0.29486626386642456,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.2419,
      "step": 36
    },
    {
      "epoch": 0.5401459854014599,
      "grad_norm": 0.3429765999317169,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.3914,
      "step": 37
    },
    {
      "epoch": 0.5547445255474452,
      "grad_norm": 0.31401655077934265,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.3065,
      "step": 38
    },
    {
      "epoch": 0.5693430656934306,
      "grad_norm": 0.2617841064929962,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.4523,
      "step": 39
    },
    {
      "epoch": 0.583941605839416,
      "grad_norm": 0.30883294343948364,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.4265,
      "step": 40
    },
    {
      "epoch": 0.5985401459854015,
      "grad_norm": 0.24459517002105713,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.2564,
      "step": 41
    },
    {
      "epoch": 0.6131386861313869,
      "grad_norm": 0.2322646975517273,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.2948,
      "step": 42
    },
    {
      "epoch": 0.6277372262773723,
      "grad_norm": 0.2812507152557373,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.6696,
      "step": 43
    },
    {
      "epoch": 0.6423357664233577,
      "grad_norm": 0.24903182685375214,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.5108,
      "step": 44
    },
    {
      "epoch": 0.656934306569343,
      "grad_norm": 0.2567322254180908,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.5481,
      "step": 45
    },
    {
      "epoch": 0.6715328467153284,
      "grad_norm": 0.2882676124572754,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.5023,
      "step": 46
    },
    {
      "epoch": 0.6861313868613139,
      "grad_norm": 0.307873010635376,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.6597,
      "step": 47
    },
    {
      "epoch": 0.7007299270072993,
      "grad_norm": 0.32634541392326355,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.3647,
      "step": 48
    },
    {
      "epoch": 0.7153284671532847,
      "grad_norm": 0.4027782678604126,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.4353,
      "step": 49
    },
    {
      "epoch": 0.7299270072992701,
      "grad_norm": 0.5115939378738403,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.544,
      "step": 50
    },
    {
      "epoch": 0.7299270072992701,
      "eval_loss": 1.349143624305725,
      "eval_runtime": 9.4009,
      "eval_samples_per_second": 12.339,
      "eval_steps_per_second": 3.085,
      "step": 50
    },
    {
      "epoch": 0.7445255474452555,
      "grad_norm": 0.8445609211921692,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.8464,
      "step": 51
    },
    {
      "epoch": 0.7591240875912408,
      "grad_norm": 0.1988578885793686,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.2327,
      "step": 52
    },
    {
      "epoch": 0.7737226277372263,
      "grad_norm": 0.19616658985614777,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.1412,
      "step": 53
    },
    {
      "epoch": 0.7883211678832117,
      "grad_norm": 0.2642453908920288,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.2773,
      "step": 54
    },
    {
      "epoch": 0.8029197080291971,
      "grad_norm": 0.24618254601955414,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.3942,
      "step": 55
    },
    {
      "epoch": 0.8175182481751825,
      "grad_norm": 0.2514287233352661,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.3295,
      "step": 56
    },
    {
      "epoch": 0.8321167883211679,
      "grad_norm": 0.24099355936050415,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.2815,
      "step": 57
    },
    {
      "epoch": 0.8467153284671532,
      "grad_norm": 0.24348944425582886,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.3757,
      "step": 58
    },
    {
      "epoch": 0.8613138686131386,
      "grad_norm": 0.30838096141815186,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.5061,
      "step": 59
    },
    {
      "epoch": 0.8759124087591241,
      "grad_norm": 0.30461886525154114,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.5202,
      "step": 60
    },
    {
      "epoch": 0.8905109489051095,
      "grad_norm": 0.2882367670536041,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.3248,
      "step": 61
    },
    {
      "epoch": 0.9051094890510949,
      "grad_norm": 0.2931157946586609,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.3582,
      "step": 62
    },
    {
      "epoch": 0.9197080291970803,
      "grad_norm": 0.2998989522457123,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.4499,
      "step": 63
    },
    {
      "epoch": 0.9343065693430657,
      "grad_norm": 0.32732918858528137,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.6151,
      "step": 64
    },
    {
      "epoch": 0.948905109489051,
      "grad_norm": 0.38178375363349915,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.5141,
      "step": 65
    },
    {
      "epoch": 0.9635036496350365,
      "grad_norm": 0.45717567205429077,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.4771,
      "step": 66
    },
    {
      "epoch": 0.9781021897810219,
      "grad_norm": 0.6583555340766907,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.2932,
      "step": 67
    },
    {
      "epoch": 0.9927007299270073,
      "grad_norm": 0.8349634408950806,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.5006,
      "step": 68
    },
    {
      "epoch": 1.0072992700729928,
      "grad_norm": 0.3964296579360962,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.9244,
      "step": 69
    },
    {
      "epoch": 1.0218978102189782,
      "grad_norm": 0.2062792032957077,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.2561,
      "step": 70
    },
    {
      "epoch": 1.0364963503649636,
      "grad_norm": 0.18427956104278564,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.1998,
      "step": 71
    },
    {
      "epoch": 1.051094890510949,
      "grad_norm": 0.22568584978580475,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.3095,
      "step": 72
    },
    {
      "epoch": 1.0656934306569343,
      "grad_norm": 0.2334766983985901,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.2668,
      "step": 73
    },
    {
      "epoch": 1.0802919708029197,
      "grad_norm": 0.21018807590007782,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.2231,
      "step": 74
    },
    {
      "epoch": 1.094890510948905,
      "grad_norm": 0.20505350828170776,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.1482,
      "step": 75
    },
    {
      "epoch": 1.1094890510948905,
      "grad_norm": 0.22474557161331177,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.1606,
      "step": 76
    },
    {
      "epoch": 1.1240875912408759,
      "grad_norm": 0.24391663074493408,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.3634,
      "step": 77
    },
    {
      "epoch": 1.1386861313868613,
      "grad_norm": 0.25364410877227783,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.4746,
      "step": 78
    },
    {
      "epoch": 1.1532846715328466,
      "grad_norm": 0.27315691113471985,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.4394,
      "step": 79
    },
    {
      "epoch": 1.167883211678832,
      "grad_norm": 0.3008474111557007,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.3001,
      "step": 80
    },
    {
      "epoch": 1.1824817518248176,
      "grad_norm": 0.3116493225097656,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.2792,
      "step": 81
    },
    {
      "epoch": 1.197080291970803,
      "grad_norm": 0.3720718026161194,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.5076,
      "step": 82
    },
    {
      "epoch": 1.2116788321167884,
      "grad_norm": 0.4089587926864624,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.2173,
      "step": 83
    },
    {
      "epoch": 1.2262773722627738,
      "grad_norm": 0.4956599175930023,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.1261,
      "step": 84
    },
    {
      "epoch": 1.2408759124087592,
      "grad_norm": 0.6784999370574951,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.0259,
      "step": 85
    },
    {
      "epoch": 1.2554744525547445,
      "grad_norm": 0.5147650837898254,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.0484,
      "step": 86
    },
    {
      "epoch": 1.27007299270073,
      "grad_norm": 0.26920634508132935,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.1713,
      "step": 87
    },
    {
      "epoch": 1.2846715328467153,
      "grad_norm": 0.34564048051834106,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.3149,
      "step": 88
    },
    {
      "epoch": 1.2992700729927007,
      "grad_norm": 0.30313724279403687,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.2969,
      "step": 89
    },
    {
      "epoch": 1.313868613138686,
      "grad_norm": 0.29985570907592773,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.3962,
      "step": 90
    },
    {
      "epoch": 1.3284671532846715,
      "grad_norm": 0.2772767245769501,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.34,
      "step": 91
    },
    {
      "epoch": 1.343065693430657,
      "grad_norm": 0.2667594254016876,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.081,
      "step": 92
    },
    {
      "epoch": 1.3576642335766422,
      "grad_norm": 0.2955487370491028,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.298,
      "step": 93
    },
    {
      "epoch": 1.3722627737226278,
      "grad_norm": 0.34554988145828247,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.3997,
      "step": 94
    },
    {
      "epoch": 1.3868613138686132,
      "grad_norm": 0.358258455991745,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.2992,
      "step": 95
    },
    {
      "epoch": 1.4014598540145986,
      "grad_norm": 0.3398512303829193,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.2166,
      "step": 96
    },
    {
      "epoch": 1.416058394160584,
      "grad_norm": 0.38044995069503784,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.3533,
      "step": 97
    },
    {
      "epoch": 1.4306569343065694,
      "grad_norm": 0.42724108695983887,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.3527,
      "step": 98
    },
    {
      "epoch": 1.4452554744525548,
      "grad_norm": 0.4584716260433197,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.443,
      "step": 99
    },
    {
      "epoch": 1.4598540145985401,
      "grad_norm": 0.49747389554977417,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.2144,
      "step": 100
    },
    {
      "epoch": 1.4598540145985401,
      "eval_loss": 1.339070439338684,
      "eval_runtime": 9.3931,
      "eval_samples_per_second": 12.349,
      "eval_steps_per_second": 3.087,
      "step": 100
    },
    {
      "epoch": 1.4744525547445255,
      "grad_norm": 0.6011104583740234,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.0075,
      "step": 101
    },
    {
      "epoch": 1.489051094890511,
      "grad_norm": 0.872632622718811,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.2001,
      "step": 102
    },
    {
      "epoch": 1.5036496350364965,
      "grad_norm": 0.6876333355903625,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.0007,
      "step": 103
    },
    {
      "epoch": 1.5182481751824817,
      "grad_norm": 0.27193692326545715,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.1106,
      "step": 104
    },
    {
      "epoch": 1.5328467153284673,
      "grad_norm": 0.4119657576084137,
      "learning_rate": 5e-05,
      "loss": 1.336,
      "step": 105
    },
    {
      "epoch": 1.5474452554744524,
      "grad_norm": 0.34590259194374084,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.2776,
      "step": 106
    },
    {
      "epoch": 1.562043795620438,
      "grad_norm": 0.3446721136569977,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.2544,
      "step": 107
    },
    {
      "epoch": 1.5766423357664232,
      "grad_norm": 0.3119318187236786,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.248,
      "step": 108
    },
    {
      "epoch": 1.5912408759124088,
      "grad_norm": 0.3031798005104065,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.2072,
      "step": 109
    },
    {
      "epoch": 1.6058394160583942,
      "grad_norm": 0.28545770049095154,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.1897,
      "step": 110
    },
    {
      "epoch": 1.6204379562043796,
      "grad_norm": 0.3152903914451599,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.391,
      "step": 111
    },
    {
      "epoch": 1.635036496350365,
      "grad_norm": 0.35289424657821655,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.472,
      "step": 112
    },
    {
      "epoch": 1.6496350364963503,
      "grad_norm": 0.3506776690483093,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.2401,
      "step": 113
    },
    {
      "epoch": 1.6642335766423357,
      "grad_norm": 0.37549662590026855,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.2805,
      "step": 114
    },
    {
      "epoch": 1.6788321167883211,
      "grad_norm": 0.38670340180397034,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.4643,
      "step": 115
    },
    {
      "epoch": 1.6934306569343067,
      "grad_norm": 0.44386011362075806,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.4414,
      "step": 116
    },
    {
      "epoch": 1.7080291970802919,
      "grad_norm": 0.5334246754646301,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.2609,
      "step": 117
    },
    {
      "epoch": 1.7226277372262775,
      "grad_norm": 0.6283140778541565,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.1396,
      "step": 118
    },
    {
      "epoch": 1.7372262773722627,
      "grad_norm": 0.799329400062561,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.0874,
      "step": 119
    },
    {
      "epoch": 1.7518248175182483,
      "grad_norm": 0.48917075991630554,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.7536,
      "step": 120
    },
    {
      "epoch": 1.7664233576642334,
      "grad_norm": 0.47777023911476135,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.1768,
      "step": 121
    },
    {
      "epoch": 1.781021897810219,
      "grad_norm": 0.30405986309051514,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.2694,
      "step": 122
    },
    {
      "epoch": 1.7956204379562044,
      "grad_norm": 0.3249014914035797,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.3427,
      "step": 123
    },
    {
      "epoch": 1.8102189781021898,
      "grad_norm": 0.3003458082675934,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.2434,
      "step": 124
    },
    {
      "epoch": 1.8248175182481752,
      "grad_norm": 0.29160162806510925,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.1671,
      "step": 125
    },
    {
      "epoch": 1.8394160583941606,
      "grad_norm": 0.2894979417324066,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.2166,
      "step": 126
    },
    {
      "epoch": 1.854014598540146,
      "grad_norm": 0.31337809562683105,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.357,
      "step": 127
    },
    {
      "epoch": 1.8686131386861313,
      "grad_norm": 0.32146742939949036,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.5565,
      "step": 128
    },
    {
      "epoch": 1.883211678832117,
      "grad_norm": 0.350217342376709,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.4829,
      "step": 129
    },
    {
      "epoch": 1.897810218978102,
      "grad_norm": 0.3359857201576233,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.157,
      "step": 130
    },
    {
      "epoch": 1.9124087591240877,
      "grad_norm": 0.36404240131378174,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.2547,
      "step": 131
    },
    {
      "epoch": 1.9270072992700729,
      "grad_norm": 0.3899008631706238,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.3203,
      "step": 132
    },
    {
      "epoch": 1.9416058394160585,
      "grad_norm": 0.4493500888347626,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.339,
      "step": 133
    },
    {
      "epoch": 1.9562043795620438,
      "grad_norm": 0.5418854355812073,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.3316,
      "step": 134
    },
    {
      "epoch": 1.9708029197080292,
      "grad_norm": 0.6482955813407898,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.2831,
      "step": 135
    },
    {
      "epoch": 1.9854014598540146,
      "grad_norm": 0.9635832905769348,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.222,
      "step": 136
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.7864303588867188,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.8149,
      "step": 137
    },
    {
      "epoch": 2.0145985401459856,
      "grad_norm": 0.277055561542511,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.0349,
      "step": 138
    },
    {
      "epoch": 2.0291970802919708,
      "grad_norm": 0.2472752034664154,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.2393,
      "step": 139
    },
    {
      "epoch": 2.0437956204379564,
      "grad_norm": 0.2657463550567627,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.33,
      "step": 140
    },
    {
      "epoch": 2.0583941605839415,
      "grad_norm": 0.28191938996315,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.2391,
      "step": 141
    },
    {
      "epoch": 2.072992700729927,
      "grad_norm": 0.2934034466743469,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.1763,
      "step": 142
    },
    {
      "epoch": 2.0875912408759123,
      "grad_norm": 0.283013254404068,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.1281,
      "step": 143
    },
    {
      "epoch": 2.102189781021898,
      "grad_norm": 0.3024832010269165,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.2139,
      "step": 144
    },
    {
      "epoch": 2.116788321167883,
      "grad_norm": 0.3049968183040619,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.1669,
      "step": 145
    },
    {
      "epoch": 2.1313868613138687,
      "grad_norm": 0.31846362352371216,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.2401,
      "step": 146
    },
    {
      "epoch": 2.145985401459854,
      "grad_norm": 0.3461815416812897,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.2748,
      "step": 147
    },
    {
      "epoch": 2.1605839416058394,
      "grad_norm": 0.3440210819244385,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.0406,
      "step": 148
    },
    {
      "epoch": 2.1751824817518246,
      "grad_norm": 0.4017361104488373,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.3617,
      "step": 149
    },
    {
      "epoch": 2.18978102189781,
      "grad_norm": 0.4131558835506439,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.3052,
      "step": 150
    },
    {
      "epoch": 2.18978102189781,
      "eval_loss": 1.3495773077011108,
      "eval_runtime": 9.6569,
      "eval_samples_per_second": 12.012,
      "eval_steps_per_second": 3.003,
      "step": 150
    },
    {
      "epoch": 2.204379562043796,
      "grad_norm": 0.46815821528434753,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.3282,
      "step": 151
    },
    {
      "epoch": 2.218978102189781,
      "grad_norm": 0.6017105579376221,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.0545,
      "step": 152
    },
    {
      "epoch": 2.2335766423357666,
      "grad_norm": 0.6463600397109985,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.8,
      "step": 153
    },
    {
      "epoch": 2.2481751824817517,
      "grad_norm": 0.9423279166221619,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.7809,
      "step": 154
    },
    {
      "epoch": 2.2627737226277373,
      "grad_norm": 0.25780820846557617,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.1084,
      "step": 155
    },
    {
      "epoch": 2.2773722627737225,
      "grad_norm": 0.29834485054016113,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.2842,
      "step": 156
    },
    {
      "epoch": 2.291970802919708,
      "grad_norm": 0.3100908100605011,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.2667,
      "step": 157
    },
    {
      "epoch": 2.3065693430656933,
      "grad_norm": 0.3335644602775574,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.2981,
      "step": 158
    },
    {
      "epoch": 2.321167883211679,
      "grad_norm": 0.3528653085231781,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.2278,
      "step": 159
    },
    {
      "epoch": 2.335766423357664,
      "grad_norm": 0.3596493601799011,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.194,
      "step": 160
    },
    {
      "epoch": 2.3503649635036497,
      "grad_norm": 0.3201064169406891,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.076,
      "step": 161
    },
    {
      "epoch": 2.3649635036496353,
      "grad_norm": 0.34449872374534607,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.2571,
      "step": 162
    },
    {
      "epoch": 2.3795620437956204,
      "grad_norm": 0.36741119623184204,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.4119,
      "step": 163
    },
    {
      "epoch": 2.394160583941606,
      "grad_norm": 0.39679640531539917,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.1969,
      "step": 164
    },
    {
      "epoch": 2.408759124087591,
      "grad_norm": 0.40736815333366394,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.2261,
      "step": 165
    },
    {
      "epoch": 2.423357664233577,
      "grad_norm": 0.4197560250759125,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.3001,
      "step": 166
    },
    {
      "epoch": 2.437956204379562,
      "grad_norm": 0.45832544565200806,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.1555,
      "step": 167
    },
    {
      "epoch": 2.4525547445255476,
      "grad_norm": 0.5514608025550842,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.2566,
      "step": 168
    },
    {
      "epoch": 2.4671532846715327,
      "grad_norm": 0.5834535956382751,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.09,
      "step": 169
    },
    {
      "epoch": 2.4817518248175183,
      "grad_norm": 0.6819263100624084,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.761,
      "step": 170
    },
    {
      "epoch": 2.4963503649635035,
      "grad_norm": 0.9520904421806335,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.6747,
      "step": 171
    },
    {
      "epoch": 2.510948905109489,
      "grad_norm": 0.25262385606765747,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.1072,
      "step": 172
    },
    {
      "epoch": 2.5255474452554747,
      "grad_norm": 0.22888913750648499,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.9735,
      "step": 173
    },
    {
      "epoch": 2.54014598540146,
      "grad_norm": 0.27815407514572144,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.1549,
      "step": 174
    },
    {
      "epoch": 2.554744525547445,
      "grad_norm": 0.2952488362789154,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.1367,
      "step": 175
    },
    {
      "epoch": 2.5693430656934306,
      "grad_norm": 0.30662426352500916,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.1852,
      "step": 176
    },
    {
      "epoch": 2.5839416058394162,
      "grad_norm": 0.3116922676563263,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.125,
      "step": 177
    },
    {
      "epoch": 2.5985401459854014,
      "grad_norm": 0.3175325393676758,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.1855,
      "step": 178
    },
    {
      "epoch": 2.613138686131387,
      "grad_norm": 0.34450942277908325,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.218,
      "step": 179
    },
    {
      "epoch": 2.627737226277372,
      "grad_norm": 0.3555043041706085,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.4273,
      "step": 180
    },
    {
      "epoch": 2.6423357664233578,
      "grad_norm": 0.3660329282283783,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.2961,
      "step": 181
    },
    {
      "epoch": 2.656934306569343,
      "grad_norm": 0.38958269357681274,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.0501,
      "step": 182
    },
    {
      "epoch": 2.6715328467153285,
      "grad_norm": 0.4110100567340851,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.1618,
      "step": 183
    },
    {
      "epoch": 2.686131386861314,
      "grad_norm": 0.4916228652000427,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.1906,
      "step": 184
    },
    {
      "epoch": 2.7007299270072993,
      "grad_norm": 0.5581548810005188,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.0481,
      "step": 185
    },
    {
      "epoch": 2.7153284671532845,
      "grad_norm": 0.6251854300498962,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.8345,
      "step": 186
    },
    {
      "epoch": 2.72992700729927,
      "grad_norm": 0.7108532786369324,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.7292,
      "step": 187
    },
    {
      "epoch": 2.7445255474452557,
      "grad_norm": 0.9583161473274231,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.9962,
      "step": 188
    },
    {
      "epoch": 2.759124087591241,
      "grad_norm": 0.22289897501468658,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.9235,
      "step": 189
    },
    {
      "epoch": 2.7737226277372264,
      "grad_norm": 0.2782759368419647,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.2478,
      "step": 190
    },
    {
      "epoch": 2.7883211678832116,
      "grad_norm": 0.29329487681388855,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.2171,
      "step": 191
    },
    {
      "epoch": 2.802919708029197,
      "grad_norm": 0.3230476379394531,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.253,
      "step": 192
    },
    {
      "epoch": 2.8175182481751824,
      "grad_norm": 0.31199565529823303,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.1981,
      "step": 193
    },
    {
      "epoch": 2.832116788321168,
      "grad_norm": 0.3323785662651062,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.1482,
      "step": 194
    },
    {
      "epoch": 2.846715328467153,
      "grad_norm": 0.3421997129917145,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.246,
      "step": 195
    },
    {
      "epoch": 2.8613138686131387,
      "grad_norm": 0.3328395485877991,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.2957,
      "step": 196
    },
    {
      "epoch": 2.875912408759124,
      "grad_norm": 0.36648160219192505,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.3804,
      "step": 197
    },
    {
      "epoch": 2.8905109489051095,
      "grad_norm": 0.3710862100124359,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.2646,
      "step": 198
    },
    {
      "epoch": 2.905109489051095,
      "grad_norm": 0.3807961642742157,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.2366,
      "step": 199
    },
    {
      "epoch": 2.9197080291970803,
      "grad_norm": 0.40780678391456604,
      "learning_rate": 0.0,
      "loss": 1.1805,
      "step": 200
    },
    {
      "epoch": 2.9197080291970803,
      "eval_loss": 1.3487451076507568,
      "eval_runtime": 9.6553,
      "eval_samples_per_second": 12.014,
      "eval_steps_per_second": 3.004,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 2
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.105049416617165e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}