{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 8.605851979345955,
  "eval_steps": 500,
  "global_step": 5000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.17,
      "grad_norm": 1.356271743774414,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.4873,
      "step": 100
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4393625259399414,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.4091,
      "step": 200
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2397950887680054,
      "learning_rate": 1.2e-05,
      "loss": 0.3903,
      "step": 300
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5770806074142456,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.3758,
      "step": 400
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0589392185211182,
      "learning_rate": 2e-05,
      "loss": 0.3517,
      "step": 500
    },
    {
      "epoch": 0.86,
      "eval_loss": 0.28141337633132935,
      "eval_runtime": 64.5177,
      "eval_samples_per_second": 15.5,
      "eval_steps_per_second": 3.875,
      "step": 500
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.1658430099487305,
      "learning_rate": 1.9555555555555557e-05,
      "loss": 0.3359,
      "step": 600
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.1505637168884277,
      "learning_rate": 1.9111111111111113e-05,
      "loss": 0.2774,
      "step": 700
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1213314533233643,
      "learning_rate": 1.866666666666667e-05,
      "loss": 0.2753,
      "step": 800
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.256823182106018,
      "learning_rate": 1.8222222222222224e-05,
      "loss": 0.2868,
      "step": 900
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.22916579246521,
      "learning_rate": 1.7777777777777777e-05,
      "loss": 0.2625,
      "step": 1000
    },
    {
      "epoch": 1.72,
      "eval_loss": 0.22975237667560577,
      "eval_runtime": 64.5328,
      "eval_samples_per_second": 15.496,
      "eval_steps_per_second": 3.874,
      "step": 1000
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9721308946609497,
      "learning_rate": 1.7333333333333336e-05,
      "loss": 0.2547,
      "step": 1100
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9259058237075806,
      "learning_rate": 1.688888888888889e-05,
      "loss": 0.2175,
      "step": 1200
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3556184768676758,
      "learning_rate": 1.6444444444444444e-05,
      "loss": 0.1587,
      "step": 1300
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4878870248794556,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.1517,
      "step": 1400
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.0709530115127563,
      "learning_rate": 1.555555555555556e-05,
      "loss": 0.151,
      "step": 1500
    },
    {
      "epoch": 2.58,
      "eval_loss": 0.23017294704914093,
      "eval_runtime": 64.5369,
      "eval_samples_per_second": 15.495,
      "eval_steps_per_second": 3.874,
      "step": 1500
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.185964822769165,
      "learning_rate": 1.5111111111111112e-05,
      "loss": 0.1584,
      "step": 1600
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.543461561203003,
      "learning_rate": 1.4666666666666666e-05,
      "loss": 0.16,
      "step": 1700
    },
    {
      "epoch": 3.1,
      "grad_norm": 1.1960724592208862,
      "learning_rate": 1.4222222222222224e-05,
      "loss": 0.1081,
      "step": 1800
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.9588227868080139,
      "learning_rate": 1.377777777777778e-05,
      "loss": 0.0811,
      "step": 1900
    },
    {
      "epoch": 3.44,
      "grad_norm": 1.2663285732269287,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.0825,
      "step": 2000
    },
    {
      "epoch": 3.44,
      "eval_loss": 0.2604709267616272,
      "eval_runtime": 64.5026,
      "eval_samples_per_second": 15.503,
      "eval_steps_per_second": 3.876,
      "step": 2000
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.2963746786117554,
      "learning_rate": 1.288888888888889e-05,
      "loss": 0.0833,
      "step": 2100
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.282989263534546,
      "learning_rate": 1.2444444444444446e-05,
      "loss": 0.0811,
      "step": 2200
    },
    {
      "epoch": 3.96,
      "grad_norm": 1.1624544858932495,
      "learning_rate": 1.2e-05,
      "loss": 0.0786,
      "step": 2300
    },
    {
      "epoch": 4.13,
      "grad_norm": 1.0563055276870728,
      "learning_rate": 1.1555555555555556e-05,
      "loss": 0.0496,
      "step": 2400
    },
    {
      "epoch": 4.3,
      "grad_norm": 1.2742573022842407,
      "learning_rate": 1.1111111111111113e-05,
      "loss": 0.0398,
      "step": 2500
    },
    {
      "epoch": 4.3,
      "eval_loss": 0.3238039016723633,
      "eval_runtime": 64.5078,
      "eval_samples_per_second": 15.502,
      "eval_steps_per_second": 3.876,
      "step": 2500
    },
    {
      "epoch": 4.48,
      "grad_norm": 1.1727477312088013,
      "learning_rate": 1.0666666666666667e-05,
      "loss": 0.0401,
      "step": 2600
    },
    {
      "epoch": 4.65,
      "grad_norm": 1.1197234392166138,
      "learning_rate": 1.0222222222222223e-05,
      "loss": 0.042,
      "step": 2700
    },
    {
      "epoch": 4.82,
      "grad_norm": 0.9968566298484802,
      "learning_rate": 9.777777777777779e-06,
      "loss": 0.0431,
      "step": 2800
    },
    {
      "epoch": 4.99,
      "grad_norm": 0.8295079469680786,
      "learning_rate": 9.333333333333334e-06,
      "loss": 0.0422,
      "step": 2900
    },
    {
      "epoch": 5.16,
      "grad_norm": 0.4814624786376953,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.0191,
      "step": 3000
    },
    {
      "epoch": 5.16,
      "eval_loss": 0.3760123550891876,
      "eval_runtime": 64.4488,
      "eval_samples_per_second": 15.516,
      "eval_steps_per_second": 3.879,
      "step": 3000
    },
    {
      "epoch": 5.34,
      "grad_norm": 1.421600341796875,
      "learning_rate": 8.444444444444446e-06,
      "loss": 0.0189,
      "step": 3100
    },
    {
      "epoch": 5.51,
      "grad_norm": 0.989109218120575,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.0192,
      "step": 3200
    },
    {
      "epoch": 5.68,
      "grad_norm": 0.7597835063934326,
      "learning_rate": 7.555555555555556e-06,
      "loss": 0.02,
      "step": 3300
    },
    {
      "epoch": 5.85,
      "grad_norm": 0.9319747686386108,
      "learning_rate": 7.111111111111112e-06,
      "loss": 0.0212,
      "step": 3400
    },
    {
      "epoch": 6.02,
      "grad_norm": 0.8922848105430603,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.02,
      "step": 3500
    },
    {
      "epoch": 6.02,
      "eval_loss": 0.34943887591362,
      "eval_runtime": 64.4584,
      "eval_samples_per_second": 15.514,
      "eval_steps_per_second": 3.878,
      "step": 3500
    },
    {
      "epoch": 6.2,
      "grad_norm": 0.4450746178627014,
      "learning_rate": 6.222222222222223e-06,
      "loss": 0.0084,
      "step": 3600
    },
    {
      "epoch": 6.37,
      "grad_norm": 0.4923359751701355,
      "learning_rate": 5.777777777777778e-06,
      "loss": 0.0093,
      "step": 3700
    },
    {
      "epoch": 6.54,
      "grad_norm": 0.7262879014015198,
      "learning_rate": 5.333333333333334e-06,
      "loss": 0.0083,
      "step": 3800
    },
    {
      "epoch": 6.71,
      "grad_norm": 0.7225219011306763,
      "learning_rate": 4.888888888888889e-06,
      "loss": 0.0101,
      "step": 3900
    },
    {
      "epoch": 6.88,
      "grad_norm": 0.6946544647216797,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.0089,
      "step": 4000
    },
    {
      "epoch": 6.88,
      "eval_loss": 0.3759503960609436,
      "eval_runtime": 64.5688,
      "eval_samples_per_second": 15.487,
      "eval_steps_per_second": 3.872,
      "step": 4000
    },
    {
      "epoch": 7.06,
      "grad_norm": 0.4038643538951874,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.0066,
      "step": 4100
    },
    {
      "epoch": 7.23,
      "grad_norm": 0.6771142482757568,
      "learning_rate": 3.555555555555556e-06,
      "loss": 0.0047,
      "step": 4200
    },
    {
      "epoch": 7.4,
      "grad_norm": 0.3253248631954193,
      "learning_rate": 3.1111111111111116e-06,
      "loss": 0.0035,
      "step": 4300
    },
    {
      "epoch": 7.57,
      "grad_norm": 0.11476542800664902,
      "learning_rate": 2.666666666666667e-06,
      "loss": 0.0035,
      "step": 4400
    },
    {
      "epoch": 7.75,
      "grad_norm": 0.12515957653522491,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.0036,
      "step": 4500
    },
    {
      "epoch": 7.75,
      "eval_loss": 0.3938386142253876,
      "eval_runtime": 64.4918,
      "eval_samples_per_second": 15.506,
      "eval_steps_per_second": 3.876,
      "step": 4500
    },
    {
      "epoch": 7.92,
      "grad_norm": 0.31758418679237366,
      "learning_rate": 1.777777777777778e-06,
      "loss": 0.0042,
      "step": 4600
    },
    {
      "epoch": 8.09,
      "grad_norm": 0.08976474404335022,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 0.0035,
      "step": 4700
    },
    {
      "epoch": 8.26,
      "grad_norm": 0.1867213398218155,
      "learning_rate": 8.88888888888889e-07,
      "loss": 0.0026,
      "step": 4800
    },
    {
      "epoch": 8.43,
      "grad_norm": 0.10856834053993225,
      "learning_rate": 4.444444444444445e-07,
      "loss": 0.0018,
      "step": 4900
    },
    {
      "epoch": 8.61,
      "grad_norm": 0.15036863088607788,
      "learning_rate": 0.0,
      "loss": 0.0019,
      "step": 5000
    },
    {
      "epoch": 8.61,
      "eval_loss": 0.4249526262283325,
      "eval_runtime": 64.5498,
      "eval_samples_per_second": 15.492,
      "eval_steps_per_second": 3.873,
      "step": 5000
    }
  ],
  "logging_steps": 100,
  "max_steps": 5000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9,
  "save_steps": 500,
  "total_flos": 1.258569996863275e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}