{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9987639060568603,
  "eval_steps": 500,
  "global_step": 505,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.009888751545117428,
      "grad_norm": 25.25,
      "learning_rate": 0.0002,
      "loss": 3.4401,
      "step": 5
    },
    {
      "epoch": 0.019777503090234856,
      "grad_norm": 11.125,
      "learning_rate": 0.0002,
      "loss": 2.2116,
      "step": 10
    },
    {
      "epoch": 0.029666254635352288,
      "grad_norm": 4.65625,
      "learning_rate": 0.0002,
      "loss": 1.6815,
      "step": 15
    },
    {
      "epoch": 0.03955500618046971,
      "grad_norm": 13.0,
      "learning_rate": 0.0002,
      "loss": 1.906,
      "step": 20
    },
    {
      "epoch": 0.049443757725587144,
      "grad_norm": 6.03125,
      "learning_rate": 0.0002,
      "loss": 1.8419,
      "step": 25
    },
    {
      "epoch": 0.059332509270704575,
      "grad_norm": 7.125,
      "learning_rate": 0.0002,
      "loss": 1.7867,
      "step": 30
    },
    {
      "epoch": 0.069221260815822,
      "grad_norm": 4.125,
      "learning_rate": 0.0002,
      "loss": 1.7177,
      "step": 35
    },
    {
      "epoch": 0.07911001236093942,
      "grad_norm": 4.375,
      "learning_rate": 0.0002,
      "loss": 1.4636,
      "step": 40
    },
    {
      "epoch": 0.08899876390605686,
      "grad_norm": 42.5,
      "learning_rate": 0.0002,
      "loss": 1.6758,
      "step": 45
    },
    {
      "epoch": 0.09888751545117429,
      "grad_norm": 3.71875,
      "learning_rate": 0.0002,
      "loss": 1.6556,
      "step": 50
    },
    {
      "epoch": 0.10877626699629171,
      "grad_norm": 3.953125,
      "learning_rate": 0.0002,
      "loss": 1.5998,
      "step": 55
    },
    {
      "epoch": 0.11866501854140915,
      "grad_norm": 3.21875,
      "learning_rate": 0.0002,
      "loss": 1.5765,
      "step": 60
    },
    {
      "epoch": 0.12855377008652658,
      "grad_norm": 3.71875,
      "learning_rate": 0.0002,
      "loss": 1.5169,
      "step": 65
    },
    {
      "epoch": 0.138442521631644,
      "grad_norm": 3.359375,
      "learning_rate": 0.0002,
      "loss": 1.582,
      "step": 70
    },
    {
      "epoch": 0.14833127317676142,
      "grad_norm": 4.96875,
      "learning_rate": 0.0002,
      "loss": 1.5751,
      "step": 75
    },
    {
      "epoch": 0.15822002472187885,
      "grad_norm": 3.03125,
      "learning_rate": 0.0002,
      "loss": 1.6266,
      "step": 80
    },
    {
      "epoch": 0.1681087762669963,
      "grad_norm": 3.640625,
      "learning_rate": 0.0002,
      "loss": 1.508,
      "step": 85
    },
    {
      "epoch": 0.17799752781211373,
      "grad_norm": 2.765625,
      "learning_rate": 0.0002,
      "loss": 1.5031,
      "step": 90
    },
    {
      "epoch": 0.18788627935723115,
      "grad_norm": 2.734375,
      "learning_rate": 0.0002,
      "loss": 1.5231,
      "step": 95
    },
    {
      "epoch": 0.19777503090234858,
      "grad_norm": 8.3125,
      "learning_rate": 0.0002,
      "loss": 1.6393,
      "step": 100
    },
    {
      "epoch": 0.207663782447466,
      "grad_norm": 145.0,
      "learning_rate": 0.0002,
      "loss": 1.7596,
      "step": 105
    },
    {
      "epoch": 0.21755253399258342,
      "grad_norm": 2.578125,
      "learning_rate": 0.0002,
      "loss": 1.5453,
      "step": 110
    },
    {
      "epoch": 0.22744128553770088,
      "grad_norm": 3.375,
      "learning_rate": 0.0002,
      "loss": 1.5337,
      "step": 115
    },
    {
      "epoch": 0.2373300370828183,
      "grad_norm": 15.375,
      "learning_rate": 0.0002,
      "loss": 1.6691,
      "step": 120
    },
    {
      "epoch": 0.24721878862793573,
      "grad_norm": 6.5625,
      "learning_rate": 0.0002,
      "loss": 1.5903,
      "step": 125
    },
    {
      "epoch": 0.25710754017305315,
      "grad_norm": 2.953125,
      "learning_rate": 0.0002,
      "loss": 1.4876,
      "step": 130
    },
    {
      "epoch": 0.2669962917181706,
      "grad_norm": 24.875,
      "learning_rate": 0.0002,
      "loss": 1.6292,
      "step": 135
    },
    {
      "epoch": 0.276885043263288,
      "grad_norm": 3.03125,
      "learning_rate": 0.0002,
      "loss": 1.5652,
      "step": 140
    },
    {
      "epoch": 0.2867737948084054,
      "grad_norm": 3.03125,
      "learning_rate": 0.0002,
      "loss": 1.5782,
      "step": 145
    },
    {
      "epoch": 0.29666254635352285,
      "grad_norm": 3.828125,
      "learning_rate": 0.0002,
      "loss": 1.5251,
      "step": 150
    },
    {
      "epoch": 0.3065512978986403,
      "grad_norm": 4.71875,
      "learning_rate": 0.0002,
      "loss": 1.4326,
      "step": 155
    },
    {
      "epoch": 0.3164400494437577,
      "grad_norm": 2.375,
      "learning_rate": 0.0002,
      "loss": 1.5085,
      "step": 160
    },
    {
      "epoch": 0.3263288009888752,
      "grad_norm": 3.84375,
      "learning_rate": 0.0002,
      "loss": 1.482,
      "step": 165
    },
    {
      "epoch": 0.3362175525339926,
      "grad_norm": 2.328125,
      "learning_rate": 0.0002,
      "loss": 1.516,
      "step": 170
    },
    {
      "epoch": 0.34610630407911,
      "grad_norm": 2.453125,
      "learning_rate": 0.0002,
      "loss": 1.5091,
      "step": 175
    },
    {
      "epoch": 0.35599505562422745,
      "grad_norm": 2.625,
      "learning_rate": 0.0002,
      "loss": 1.4517,
      "step": 180
    },
    {
      "epoch": 0.3658838071693449,
      "grad_norm": 2.84375,
      "learning_rate": 0.0002,
      "loss": 1.4647,
      "step": 185
    },
    {
      "epoch": 0.3757725587144623,
      "grad_norm": 2.734375,
      "learning_rate": 0.0002,
      "loss": 1.4764,
      "step": 190
    },
    {
      "epoch": 0.3856613102595797,
      "grad_norm": 2.15625,
      "learning_rate": 0.0002,
      "loss": 1.4254,
      "step": 195
    },
    {
      "epoch": 0.39555006180469715,
      "grad_norm": 2.34375,
      "learning_rate": 0.0002,
      "loss": 1.4868,
      "step": 200
    },
    {
      "epoch": 0.4054388133498146,
      "grad_norm": 2.609375,
      "learning_rate": 0.0002,
      "loss": 1.4534,
      "step": 205
    },
    {
      "epoch": 0.415327564894932,
      "grad_norm": 2.34375,
      "learning_rate": 0.0002,
      "loss": 1.447,
      "step": 210
    },
    {
      "epoch": 0.4252163164400494,
      "grad_norm": 2.3125,
      "learning_rate": 0.0002,
      "loss": 1.4514,
      "step": 215
    },
    {
      "epoch": 0.43510506798516685,
      "grad_norm": 2.109375,
      "learning_rate": 0.0002,
      "loss": 1.4372,
      "step": 220
    },
    {
      "epoch": 0.44499381953028433,
      "grad_norm": 2.859375,
      "learning_rate": 0.0002,
      "loss": 1.3961,
      "step": 225
    },
    {
      "epoch": 0.45488257107540175,
      "grad_norm": 3.0625,
      "learning_rate": 0.0002,
      "loss": 1.4363,
      "step": 230
    },
    {
      "epoch": 0.4647713226205192,
      "grad_norm": 2.703125,
      "learning_rate": 0.0002,
      "loss": 1.4355,
      "step": 235
    },
    {
      "epoch": 0.4746600741656366,
      "grad_norm": 2.8125,
      "learning_rate": 0.0002,
      "loss": 1.525,
      "step": 240
    },
    {
      "epoch": 0.484548825710754,
      "grad_norm": 2.46875,
      "learning_rate": 0.0002,
      "loss": 1.3801,
      "step": 245
    },
    {
      "epoch": 0.49443757725587145,
      "grad_norm": 2.40625,
      "learning_rate": 0.0002,
      "loss": 1.5133,
      "step": 250
    },
    {
      "epoch": 0.5043263288009888,
      "grad_norm": 2.328125,
      "learning_rate": 0.0002,
      "loss": 1.4364,
      "step": 255
    },
    {
      "epoch": 0.5142150803461063,
      "grad_norm": 2.3125,
      "learning_rate": 0.0002,
      "loss": 1.4363,
      "step": 260
    },
    {
      "epoch": 0.5241038318912238,
      "grad_norm": 2.75,
      "learning_rate": 0.0002,
      "loss": 1.521,
      "step": 265
    },
    {
      "epoch": 0.5339925834363412,
      "grad_norm": 2.4375,
      "learning_rate": 0.0002,
      "loss": 1.4559,
      "step": 270
    },
    {
      "epoch": 0.5438813349814586,
      "grad_norm": 2.5,
      "learning_rate": 0.0002,
      "loss": 1.4516,
      "step": 275
    },
    {
      "epoch": 0.553770086526576,
      "grad_norm": 2.125,
      "learning_rate": 0.0002,
      "loss": 1.3968,
      "step": 280
    },
    {
      "epoch": 0.5636588380716935,
      "grad_norm": 2.328125,
      "learning_rate": 0.0002,
      "loss": 1.4183,
      "step": 285
    },
    {
      "epoch": 0.5735475896168108,
      "grad_norm": 1.8046875,
      "learning_rate": 0.0002,
      "loss": 1.3653,
      "step": 290
    },
    {
      "epoch": 0.5834363411619283,
      "grad_norm": 2.0,
      "learning_rate": 0.0002,
      "loss": 1.388,
      "step": 295
    },
    {
      "epoch": 0.5933250927070457,
      "grad_norm": 1.953125,
      "learning_rate": 0.0002,
      "loss": 1.3131,
      "step": 300
    },
    {
      "epoch": 0.6032138442521632,
      "grad_norm": 2.390625,
      "learning_rate": 0.0002,
      "loss": 1.4784,
      "step": 305
    },
    {
      "epoch": 0.6131025957972805,
      "grad_norm": 2.34375,
      "learning_rate": 0.0002,
      "loss": 1.4339,
      "step": 310
    },
    {
      "epoch": 0.622991347342398,
      "grad_norm": 2.140625,
      "learning_rate": 0.0002,
      "loss": 1.4425,
      "step": 315
    },
    {
      "epoch": 0.6328800988875154,
      "grad_norm": 2.078125,
      "learning_rate": 0.0002,
      "loss": 1.3847,
      "step": 320
    },
    {
      "epoch": 0.6427688504326329,
      "grad_norm": 2.359375,
      "learning_rate": 0.0002,
      "loss": 1.4252,
      "step": 325
    },
    {
      "epoch": 0.6526576019777504,
      "grad_norm": 2.640625,
      "learning_rate": 0.0002,
      "loss": 1.4253,
      "step": 330
    },
    {
      "epoch": 0.6625463535228677,
      "grad_norm": 2.015625,
      "learning_rate": 0.0002,
      "loss": 1.3336,
      "step": 335
    },
    {
      "epoch": 0.6724351050679852,
      "grad_norm": 2.171875,
      "learning_rate": 0.0002,
      "loss": 1.4209,
      "step": 340
    },
    {
      "epoch": 0.6823238566131026,
      "grad_norm": 4.0625,
      "learning_rate": 0.0002,
      "loss": 1.4485,
      "step": 345
    },
    {
      "epoch": 0.69221260815822,
      "grad_norm": 2.1875,
      "learning_rate": 0.0002,
      "loss": 1.4261,
      "step": 350
    },
    {
      "epoch": 0.7021013597033374,
      "grad_norm": 2.515625,
      "learning_rate": 0.0002,
      "loss": 1.3869,
      "step": 355
    },
    {
      "epoch": 0.7119901112484549,
      "grad_norm": 2.1875,
      "learning_rate": 0.0002,
      "loss": 1.4681,
      "step": 360
    },
    {
      "epoch": 0.7218788627935723,
      "grad_norm": 3.609375,
      "learning_rate": 0.0002,
      "loss": 1.3513,
      "step": 365
    },
    {
      "epoch": 0.7317676143386898,
      "grad_norm": 2.296875,
      "learning_rate": 0.0002,
      "loss": 1.4029,
      "step": 370
    },
    {
      "epoch": 0.7416563658838071,
      "grad_norm": 2.15625,
      "learning_rate": 0.0002,
      "loss": 1.3716,
      "step": 375
    },
    {
      "epoch": 0.7515451174289246,
      "grad_norm": 3.75,
      "learning_rate": 0.0002,
      "loss": 1.3902,
      "step": 380
    },
    {
      "epoch": 0.761433868974042,
      "grad_norm": 2.0625,
      "learning_rate": 0.0002,
      "loss": 1.3469,
      "step": 385
    },
    {
      "epoch": 0.7713226205191595,
      "grad_norm": 1.9609375,
      "learning_rate": 0.0002,
      "loss": 1.3216,
      "step": 390
    },
    {
      "epoch": 0.7812113720642769,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0002,
      "loss": 1.4774,
      "step": 395
    },
    {
      "epoch": 0.7911001236093943,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0002,
      "loss": 1.372,
      "step": 400
    },
    {
      "epoch": 0.8009888751545118,
      "grad_norm": 1.921875,
      "learning_rate": 0.0002,
      "loss": 1.3574,
      "step": 405
    },
    {
      "epoch": 0.8108776266996292,
      "grad_norm": 2.1875,
      "learning_rate": 0.0002,
      "loss": 1.4444,
      "step": 410
    },
    {
      "epoch": 0.8207663782447466,
      "grad_norm": 2.296875,
      "learning_rate": 0.0002,
      "loss": 1.3743,
      "step": 415
    },
    {
      "epoch": 0.830655129789864,
      "grad_norm": 2.125,
      "learning_rate": 0.0002,
      "loss": 1.3589,
      "step": 420
    },
    {
      "epoch": 0.8405438813349815,
      "grad_norm": 2.046875,
      "learning_rate": 0.0002,
      "loss": 1.3336,
      "step": 425
    },
    {
      "epoch": 0.8504326328800988,
      "grad_norm": 3.84375,
      "learning_rate": 0.0002,
      "loss": 1.3584,
      "step": 430
    },
    {
      "epoch": 0.8603213844252163,
      "grad_norm": 2.125,
      "learning_rate": 0.0002,
      "loss": 1.332,
      "step": 435
    },
    {
      "epoch": 0.8702101359703337,
      "grad_norm": 1.8828125,
      "learning_rate": 0.0002,
      "loss": 1.3678,
      "step": 440
    },
    {
      "epoch": 0.8800988875154512,
      "grad_norm": 2.0625,
      "learning_rate": 0.0002,
      "loss": 1.2967,
      "step": 445
    },
    {
      "epoch": 0.8899876390605687,
      "grad_norm": 2.3125,
      "learning_rate": 0.0002,
      "loss": 1.3276,
      "step": 450
    },
    {
      "epoch": 0.899876390605686,
      "grad_norm": 1.953125,
      "learning_rate": 0.0002,
      "loss": 1.3175,
      "step": 455
    },
    {
      "epoch": 0.9097651421508035,
      "grad_norm": 1.7734375,
      "learning_rate": 0.0002,
      "loss": 1.366,
      "step": 460
    },
    {
      "epoch": 0.9196538936959209,
      "grad_norm": 2.46875,
      "learning_rate": 0.0002,
      "loss": 1.3105,
      "step": 465
    },
    {
      "epoch": 0.9295426452410384,
      "grad_norm": 1.84375,
      "learning_rate": 0.0002,
      "loss": 1.2856,
      "step": 470
    },
    {
      "epoch": 0.9394313967861557,
      "grad_norm": 1.7890625,
      "learning_rate": 0.0002,
      "loss": 1.3662,
      "step": 475
    },
    {
      "epoch": 0.9493201483312732,
      "grad_norm": 1.921875,
      "learning_rate": 0.0002,
      "loss": 1.3289,
      "step": 480
    },
    {
      "epoch": 0.9592088998763906,
      "grad_norm": 2.078125,
      "learning_rate": 0.0002,
      "loss": 1.3812,
      "step": 485
    },
    {
      "epoch": 0.969097651421508,
      "grad_norm": 2.703125,
      "learning_rate": 0.0002,
      "loss": 1.3344,
      "step": 490
    },
    {
      "epoch": 0.9789864029666254,
      "grad_norm": 1.9609375,
      "learning_rate": 0.0002,
      "loss": 1.345,
      "step": 495
    },
    {
      "epoch": 0.9888751545117429,
      "grad_norm": 2.125,
      "learning_rate": 0.0002,
      "loss": 1.3163,
      "step": 500
    },
    {
      "epoch": 0.9987639060568603,
      "grad_norm": 2.125,
      "learning_rate": 0.0002,
      "loss": 1.354,
      "step": 505
    },
    {
      "epoch": 0.9987639060568603,
      "step": 505,
      "total_flos": 9435624701952000.0,
      "train_loss": 1.491983689412032,
      "train_runtime": 449.9567,
      "train_samples_per_second": 17.977,
      "train_steps_per_second": 1.122
    }
  ],
  "logging_steps": 5,
  "max_steps": 505,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 9435624701952000.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}