{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 20.0,
  "eval_steps": 500,
  "global_step": 5000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.4,
      "grad_norm": 0.1675417125225067,
      "learning_rate": 3e-05,
      "loss": 1.5895,
      "step": 100
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1831749528646469,
      "learning_rate": 3e-05,
      "loss": 1.4742,
      "step": 200
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.6748151382823872,
      "eval_loss": 1.531341314315796,
      "eval_runtime": 9.1549,
      "eval_samples_per_second": 54.616,
      "eval_steps_per_second": 6.882,
      "step": 250
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.22951605916023254,
      "learning_rate": 3e-05,
      "loss": 1.4652,
      "step": 300
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.23647047579288483,
      "learning_rate": 3e-05,
      "loss": 1.4601,
      "step": 400
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.2248772829771042,
      "learning_rate": 3e-05,
      "loss": 1.45,
      "step": 500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.6756593886462882,
      "eval_loss": 1.5196012258529663,
      "eval_runtime": 9.0923,
      "eval_samples_per_second": 54.992,
      "eval_steps_per_second": 6.929,
      "step": 500
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.26751402020454407,
      "learning_rate": 3e-05,
      "loss": 1.4361,
      "step": 600
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.29390770196914673,
      "learning_rate": 3e-05,
      "loss": 1.4269,
      "step": 700
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.6760756914119359,
      "eval_loss": 1.5133788585662842,
      "eval_runtime": 9.1109,
      "eval_samples_per_second": 54.879,
      "eval_steps_per_second": 6.915,
      "step": 750
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.3322733938694,
      "learning_rate": 3e-05,
      "loss": 1.4184,
      "step": 800
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.3646068871021271,
      "learning_rate": 3e-05,
      "loss": 1.3973,
      "step": 900
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.4150474965572357,
      "learning_rate": 3e-05,
      "loss": 1.3999,
      "step": 1000
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.6762474526928676,
      "eval_loss": 1.511995792388916,
      "eval_runtime": 9.0922,
      "eval_samples_per_second": 54.992,
      "eval_steps_per_second": 6.929,
      "step": 1000
    },
    {
      "epoch": 4.4,
      "grad_norm": 0.44478657841682434,
      "learning_rate": 3e-05,
      "loss": 1.3624,
      "step": 1100
    },
    {
      "epoch": 4.8,
      "grad_norm": 0.48866602778434753,
      "learning_rate": 3e-05,
      "loss": 1.3614,
      "step": 1200
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.6759650655021834,
      "eval_loss": 1.5192290544509888,
      "eval_runtime": 9.1143,
      "eval_samples_per_second": 54.859,
      "eval_steps_per_second": 6.912,
      "step": 1250
    },
    {
      "epoch": 5.2,
      "grad_norm": 0.5195505023002625,
      "learning_rate": 3e-05,
      "loss": 1.3431,
      "step": 1300
    },
    {
      "epoch": 5.6,
      "grad_norm": 0.5769343972206116,
      "learning_rate": 3e-05,
      "loss": 1.3264,
      "step": 1400
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.5181849598884583,
      "learning_rate": 3e-05,
      "loss": 1.3303,
      "step": 1500
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.6755080058224163,
      "eval_loss": 1.5265752077102661,
      "eval_runtime": 9.1234,
      "eval_samples_per_second": 54.804,
      "eval_steps_per_second": 6.905,
      "step": 1500
    },
    {
      "epoch": 6.4,
      "grad_norm": 0.6372528076171875,
      "learning_rate": 3e-05,
      "loss": 1.2883,
      "step": 1600
    },
    {
      "epoch": 6.8,
      "grad_norm": 0.6501044034957886,
      "learning_rate": 3e-05,
      "loss": 1.2946,
      "step": 1700
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.6747045123726346,
      "eval_loss": 1.5446096658706665,
      "eval_runtime": 9.0797,
      "eval_samples_per_second": 55.068,
      "eval_steps_per_second": 6.939,
      "step": 1750
    },
    {
      "epoch": 7.2,
      "grad_norm": 0.7209816575050354,
      "learning_rate": 3e-05,
      "loss": 1.2705,
      "step": 1800
    },
    {
      "epoch": 7.6,
      "grad_norm": 0.7495877742767334,
      "learning_rate": 3e-05,
      "loss": 1.2498,
      "step": 1900
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.681526780128479,
      "learning_rate": 3e-05,
      "loss": 1.2518,
      "step": 2000
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.6745036390101893,
      "eval_loss": 1.5590205192565918,
      "eval_runtime": 8.0597,
      "eval_samples_per_second": 62.037,
      "eval_steps_per_second": 7.817,
      "step": 2000
    },
    {
      "epoch": 8.4,
      "grad_norm": 0.7470565438270569,
      "learning_rate": 3e-05,
      "loss": 1.2196,
      "step": 2100
    },
    {
      "epoch": 8.8,
      "grad_norm": 0.7745229005813599,
      "learning_rate": 3e-05,
      "loss": 1.2082,
      "step": 2200
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.6740349344978166,
      "eval_loss": 1.571682333946228,
      "eval_runtime": 9.0961,
      "eval_samples_per_second": 54.969,
      "eval_steps_per_second": 6.926,
      "step": 2250
    },
    {
      "epoch": 9.2,
      "grad_norm": 0.8478706479072571,
      "learning_rate": 3e-05,
      "loss": 1.2017,
      "step": 2300
    },
    {
      "epoch": 9.6,
      "grad_norm": 0.9340612292289734,
      "learning_rate": 3e-05,
      "loss": 1.1742,
      "step": 2400
    },
    {
      "epoch": 10.0,
      "grad_norm": 0.9207039475440979,
      "learning_rate": 3e-05,
      "loss": 1.19,
      "step": 2500
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.6727074235807861,
      "eval_loss": 1.6021865606307983,
      "eval_runtime": 9.0984,
      "eval_samples_per_second": 54.955,
      "eval_steps_per_second": 6.924,
      "step": 2500
    },
    {
      "epoch": 10.4,
      "grad_norm": 0.9244349598884583,
      "learning_rate": 3e-05,
      "loss": 1.1299,
      "step": 2600
    },
    {
      "epoch": 10.8,
      "grad_norm": 0.9110805988311768,
      "learning_rate": 3e-05,
      "loss": 1.1523,
      "step": 2700
    },
    {
      "epoch": 11.0,
      "eval_accuracy": 0.672608442503639,
      "eval_loss": 1.6098225116729736,
      "eval_runtime": 9.117,
      "eval_samples_per_second": 54.843,
      "eval_steps_per_second": 6.91,
      "step": 2750
    },
    {
      "epoch": 11.2,
      "grad_norm": 0.916287899017334,
      "learning_rate": 3e-05,
      "loss": 1.1278,
      "step": 2800
    },
    {
      "epoch": 11.6,
      "grad_norm": 1.0008416175842285,
      "learning_rate": 3e-05,
      "loss": 1.0981,
      "step": 2900
    },
    {
      "epoch": 12.0,
      "grad_norm": 0.9763438701629639,
      "learning_rate": 3e-05,
      "loss": 1.1193,
      "step": 3000
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.671589519650655,
      "eval_loss": 1.6344681978225708,
      "eval_runtime": 9.1836,
      "eval_samples_per_second": 54.445,
      "eval_steps_per_second": 6.86,
      "step": 3000
    },
    {
      "epoch": 12.4,
      "grad_norm": 1.0682412385940552,
      "learning_rate": 3e-05,
      "loss": 1.0604,
      "step": 3100
    },
    {
      "epoch": 12.8,
      "grad_norm": 1.1306017637252808,
      "learning_rate": 3e-05,
      "loss": 1.0736,
      "step": 3200
    },
    {
      "epoch": 13.0,
      "eval_accuracy": 0.6707016011644833,
      "eval_loss": 1.674833059310913,
      "eval_runtime": 8.098,
      "eval_samples_per_second": 61.744,
      "eval_steps_per_second": 7.78,
      "step": 3250
    },
    {
      "epoch": 13.2,
      "grad_norm": 1.154135823249817,
      "learning_rate": 3e-05,
      "loss": 1.054,
      "step": 3300
    },
    {
      "epoch": 13.6,
      "grad_norm": 1.1169854402542114,
      "learning_rate": 3e-05,
      "loss": 1.0253,
      "step": 3400
    },
    {
      "epoch": 14.0,
      "grad_norm": 1.0877137184143066,
      "learning_rate": 3e-05,
      "loss": 1.0414,
      "step": 3500
    },
    {
      "epoch": 14.0,
      "eval_accuracy": 0.6701280931586608,
      "eval_loss": 1.688016653060913,
      "eval_runtime": 9.1712,
      "eval_samples_per_second": 54.519,
      "eval_steps_per_second": 6.869,
      "step": 3500
    },
    {
      "epoch": 14.4,
      "grad_norm": 1.2245118618011475,
      "learning_rate": 3e-05,
      "loss": 0.9823,
      "step": 3600
    },
    {
      "epoch": 14.8,
      "grad_norm": 1.2784464359283447,
      "learning_rate": 3e-05,
      "loss": 1.0069,
      "step": 3700
    },
    {
      "epoch": 15.0,
      "eval_accuracy": 0.6693682678311499,
      "eval_loss": 1.7182435989379883,
      "eval_runtime": 9.1433,
      "eval_samples_per_second": 54.685,
      "eval_steps_per_second": 6.89,
      "step": 3750
    },
    {
      "epoch": 15.2,
      "grad_norm": 1.182626724243164,
      "learning_rate": 3e-05,
      "loss": 0.9834,
      "step": 3800
    },
    {
      "epoch": 15.6,
      "grad_norm": 1.3419315814971924,
      "learning_rate": 3e-05,
      "loss": 0.9608,
      "step": 3900
    },
    {
      "epoch": 16.0,
      "grad_norm": 1.2352997064590454,
      "learning_rate": 3e-05,
      "loss": 0.9654,
      "step": 4000
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.6685036390101893,
      "eval_loss": 1.7521902322769165,
      "eval_runtime": 9.1045,
      "eval_samples_per_second": 54.918,
      "eval_steps_per_second": 6.92,
      "step": 4000
    },
    {
      "epoch": 16.4,
      "grad_norm": 1.439382791519165,
      "learning_rate": 3e-05,
      "loss": 0.9236,
      "step": 4100
    },
    {
      "epoch": 16.8,
      "grad_norm": 1.3681950569152832,
      "learning_rate": 3e-05,
      "loss": 0.9337,
      "step": 4200
    },
    {
      "epoch": 17.0,
      "eval_accuracy": 0.6677409024745269,
      "eval_loss": 1.7825894355773926,
      "eval_runtime": 9.1188,
      "eval_samples_per_second": 54.832,
      "eval_steps_per_second": 6.909,
      "step": 4250
    },
    {
      "epoch": 17.2,
      "grad_norm": 1.417385458946228,
      "learning_rate": 3e-05,
      "loss": 0.9135,
      "step": 4300
    },
    {
      "epoch": 17.6,
      "grad_norm": 1.5673705339431763,
      "learning_rate": 3e-05,
      "loss": 0.8941,
      "step": 4400
    },
    {
      "epoch": 18.0,
      "grad_norm": 1.5578211545944214,
      "learning_rate": 3e-05,
      "loss": 0.9,
      "step": 4500
    },
    {
      "epoch": 18.0,
      "eval_accuracy": 0.6671615720524018,
      "eval_loss": 1.8080195188522339,
      "eval_runtime": 8.0689,
      "eval_samples_per_second": 61.967,
      "eval_steps_per_second": 7.808,
      "step": 4500
    },
    {
      "epoch": 18.4,
      "grad_norm": 1.671644687652588,
      "learning_rate": 3e-05,
      "loss": 0.8522,
      "step": 4600
    },
    {
      "epoch": 18.8,
      "grad_norm": 1.5266335010528564,
      "learning_rate": 3e-05,
      "loss": 0.8704,
      "step": 4700
    },
    {
      "epoch": 19.0,
      "eval_accuracy": 0.6663289665211063,
      "eval_loss": 1.8349848985671997,
      "eval_runtime": 9.1076,
      "eval_samples_per_second": 54.899,
      "eval_steps_per_second": 6.917,
      "step": 4750
    },
    {
      "epoch": 19.2,
      "grad_norm": 1.5969411134719849,
      "learning_rate": 3e-05,
      "loss": 0.8469,
      "step": 4800
    },
    {
      "epoch": 19.6,
      "grad_norm": 1.5390815734863281,
      "learning_rate": 3e-05,
      "loss": 0.8398,
      "step": 4900
    },
    {
      "epoch": 20.0,
      "grad_norm": 1.5651904344558716,
      "learning_rate": 3e-05,
      "loss": 0.8407,
      "step": 5000
    },
    {
      "epoch": 20.0,
      "eval_accuracy": 0.6657583697234353,
      "eval_loss": 1.8696147203445435,
      "eval_runtime": 9.1023,
      "eval_samples_per_second": 54.931,
      "eval_steps_per_second": 6.921,
      "step": 5000
    },
    {
      "epoch": 20.0,
      "step": 5000,
      "total_flos": 9.221411586147615e+17,
      "train_loss": 1.163707649230957,
      "train_runtime": 11749.7477,
      "train_samples_per_second": 13.617,
      "train_steps_per_second": 0.426
    }
  ],
  "logging_steps": 100,
  "max_steps": 5000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 20,
  "save_steps": 500,
  "total_flos": 9.221411586147615e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}