{
  "best_global_step": 420,
  "best_metric": 0.03601359203457832,
  "best_model_checkpoint": "/content/models/gemma_qlora_lmh/checkpoint-420",
  "epoch": 1.6535433070866141,
  "eval_steps": 20,
  "global_step": 420,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "entropy": 2.0512761652469633,
      "epoch": 0.07874015748031496,
      "grad_norm": 2.0831573009490967,
      "learning_rate": 9.625984251968504e-06,
      "loss": 0.6806,
      "mean_token_accuracy": 0.8287491604685784,
      "num_tokens": 80574.0,
      "step": 20
    },
    {
      "epoch": 0.07874015748031496,
      "eval_entropy": 2.0480873808264732,
      "eval_loss": 0.16752119362354279,
      "eval_mean_token_accuracy": 0.9204465709626675,
      "eval_num_tokens": 80574.0,
      "eval_runtime": 14.4058,
      "eval_samples_per_second": 70.388,
      "eval_steps_per_second": 4.443,
      "step": 20
    },
    {
      "entropy": 2.043068042397499,
      "epoch": 0.15748031496062992,
      "grad_norm": 3.2172703742980957,
      "learning_rate": 9.23228346456693e-06,
      "loss": 0.1758,
      "mean_token_accuracy": 0.9078111112117767,
      "num_tokens": 163585.0,
      "step": 40
    },
    {
      "epoch": 0.15748031496062992,
      "eval_entropy": 1.998575333505869,
      "eval_loss": 0.13858385384082794,
      "eval_mean_token_accuracy": 0.9343411969020963,
      "eval_num_tokens": 163585.0,
      "eval_runtime": 14.4076,
      "eval_samples_per_second": 70.38,
      "eval_steps_per_second": 4.442,
      "step": 40
    },
    {
      "entropy": 2.0217554807662963,
      "epoch": 0.23622047244094488,
      "grad_norm": 6.096222877502441,
      "learning_rate": 8.838582677165355e-06,
      "loss": 0.1675,
      "mean_token_accuracy": 0.922796231508255,
      "num_tokens": 247496.0,
      "step": 60
    },
    {
      "epoch": 0.23622047244094488,
      "eval_entropy": 2.0012808348983526,
      "eval_loss": 0.17316989600658417,
      "eval_mean_token_accuracy": 0.9123187400400639,
      "eval_num_tokens": 247496.0,
      "eval_runtime": 14.4234,
      "eval_samples_per_second": 70.302,
      "eval_steps_per_second": 4.437,
      "step": 60
    },
    {
      "entropy": 2.01616033911705,
      "epoch": 0.31496062992125984,
      "grad_norm": 3.203735113143921,
      "learning_rate": 8.444881889763782e-06,
      "loss": 0.1077,
      "mean_token_accuracy": 0.9547338560223579,
      "num_tokens": 330519.0,
      "step": 80
    },
    {
      "epoch": 0.31496062992125984,
      "eval_entropy": 1.967957517132163,
      "eval_loss": 0.10772992670536041,
      "eval_mean_token_accuracy": 0.9575660079717636,
      "eval_num_tokens": 330519.0,
      "eval_runtime": 14.4579,
      "eval_samples_per_second": 70.135,
      "eval_steps_per_second": 4.427,
      "step": 80
    },
    {
      "entropy": 1.9572902411222457,
      "epoch": 0.3937007874015748,
      "grad_norm": 12.593894004821777,
      "learning_rate": 8.051181102362205e-06,
      "loss": 0.0977,
      "mean_token_accuracy": 0.9625995993614197,
      "num_tokens": 412053.0,
      "step": 100
    },
    {
      "epoch": 0.3937007874015748,
      "eval_entropy": 1.9561998695135117,
      "eval_loss": 0.10965846478939056,
      "eval_mean_token_accuracy": 0.9605293860659003,
      "eval_num_tokens": 412053.0,
      "eval_runtime": 14.4568,
      "eval_samples_per_second": 70.14,
      "eval_steps_per_second": 4.427,
      "step": 100
    },
    {
      "entropy": 1.9785019993782043,
      "epoch": 0.47244094488188976,
      "grad_norm": 5.114565849304199,
      "learning_rate": 7.65748031496063e-06,
      "loss": 0.0684,
      "mean_token_accuracy": 0.9732592537999153,
      "num_tokens": 492803.0,
      "step": 120
    },
    {
      "epoch": 0.47244094488188976,
      "eval_entropy": 1.9417847488075495,
      "eval_loss": 0.11676711589097977,
      "eval_mean_token_accuracy": 0.970998496748507,
      "eval_num_tokens": 492803.0,
      "eval_runtime": 15.073,
      "eval_samples_per_second": 67.273,
      "eval_steps_per_second": 4.246,
      "step": 120
    },
    {
      "entropy": 1.9924108654260635,
      "epoch": 0.5511811023622047,
      "grad_norm": 4.5052642822265625,
      "learning_rate": 7.263779527559056e-06,
      "loss": 0.0727,
      "mean_token_accuracy": 0.9757537111639977,
      "num_tokens": 574646.0,
      "step": 140
    },
    {
      "epoch": 0.5511811023622047,
      "eval_entropy": 1.9667269736528397,
      "eval_loss": 0.05474039167165756,
      "eval_mean_token_accuracy": 0.981005135923624,
      "eval_num_tokens": 574646.0,
      "eval_runtime": 14.4351,
      "eval_samples_per_second": 70.245,
      "eval_steps_per_second": 4.434,
      "step": 140
    },
    {
      "entropy": 1.9721805155277252,
      "epoch": 0.6299212598425197,
      "grad_norm": 3.2773079872131348,
      "learning_rate": 6.870078740157481e-06,
      "loss": 0.0417,
      "mean_token_accuracy": 0.9847830027341843,
      "num_tokens": 656089.0,
      "step": 160
    },
    {
      "epoch": 0.6299212598425197,
      "eval_entropy": 1.9357626736164093,
      "eval_loss": 0.05493154749274254,
      "eval_mean_token_accuracy": 0.9821033654734492,
      "eval_num_tokens": 656089.0,
      "eval_runtime": 14.7904,
      "eval_samples_per_second": 68.558,
      "eval_steps_per_second": 4.327,
      "step": 160
    },
    {
      "entropy": 1.980790689587593,
      "epoch": 0.7086614173228346,
      "grad_norm": 3.153142213821411,
      "learning_rate": 6.476377952755906e-06,
      "loss": 0.0717,
      "mean_token_accuracy": 0.9758390337228775,
      "num_tokens": 736840.0,
      "step": 180
    },
    {
      "epoch": 0.7086614173228346,
      "eval_entropy": 1.9667537324130535,
      "eval_loss": 0.05807032063603401,
      "eval_mean_token_accuracy": 0.9831889141350985,
      "eval_num_tokens": 736840.0,
      "eval_runtime": 14.7032,
      "eval_samples_per_second": 68.965,
      "eval_steps_per_second": 4.353,
      "step": 180
    },
    {
      "entropy": 1.990926167368889,
      "epoch": 0.7874015748031497,
      "grad_norm": 2.1746864318847656,
      "learning_rate": 6.082677165354331e-06,
      "loss": 0.0596,
      "mean_token_accuracy": 0.9820288747549057,
      "num_tokens": 818671.0,
      "step": 200
    },
    {
      "epoch": 0.7874015748031497,
      "eval_entropy": 1.9952750019729137,
      "eval_loss": 0.039442677050828934,
      "eval_mean_token_accuracy": 0.9868771303445101,
      "eval_num_tokens": 818671.0,
      "eval_runtime": 14.4324,
      "eval_samples_per_second": 70.259,
      "eval_steps_per_second": 4.434,
      "step": 200
    },
    {
      "entropy": 1.9992768168449402,
      "epoch": 0.8661417322834646,
      "grad_norm": 3.7657172679901123,
      "learning_rate": 5.6889763779527565e-06,
      "loss": 0.0434,
      "mean_token_accuracy": 0.9879351630806923,
      "num_tokens": 900585.0,
      "step": 220
    },
    {
      "epoch": 0.8661417322834646,
      "eval_entropy": 1.9681989178061485,
      "eval_loss": 0.06415939331054688,
      "eval_mean_token_accuracy": 0.9848531475290656,
      "eval_num_tokens": 900585.0,
      "eval_runtime": 14.4299,
      "eval_samples_per_second": 70.271,
      "eval_steps_per_second": 4.435,
      "step": 220
    },
    {
      "entropy": 1.9873531699180602,
      "epoch": 0.9448818897637795,
      "grad_norm": 1.2447177171707153,
      "learning_rate": 5.295275590551181e-06,
      "loss": 0.0495,
      "mean_token_accuracy": 0.9891579717397689,
      "num_tokens": 982502.0,
      "step": 240
    },
    {
      "epoch": 0.9448818897637795,
      "eval_entropy": 1.9764121137559414,
      "eval_loss": 0.05618462339043617,
      "eval_mean_token_accuracy": 0.984872730448842,
      "eval_num_tokens": 982502.0,
      "eval_runtime": 14.5073,
      "eval_samples_per_second": 69.896,
      "eval_steps_per_second": 4.412,
      "step": 240
    },
    {
      "entropy": 1.9799151510000228,
      "epoch": 1.0236220472440944,
      "grad_norm": 2.651578426361084,
      "learning_rate": 4.901574803149607e-06,
      "loss": 0.037,
      "mean_token_accuracy": 0.9879837512969971,
      "num_tokens": 1061407.0,
      "step": 260
    },
    {
      "epoch": 1.0236220472440944,
      "eval_entropy": 1.9760331977158785,
      "eval_loss": 0.04186183959245682,
      "eval_mean_token_accuracy": 0.9875151747837663,
      "eval_num_tokens": 1061407.0,
      "eval_runtime": 14.3962,
      "eval_samples_per_second": 70.435,
      "eval_steps_per_second": 4.446,
      "step": 260
    },
    {
      "entropy": 1.9939074516296387,
      "epoch": 1.1023622047244095,
      "grad_norm": 2.1040918827056885,
      "learning_rate": 4.507874015748032e-06,
      "loss": 0.0287,
      "mean_token_accuracy": 0.9914279609918595,
      "num_tokens": 1144556.0,
      "step": 280
    },
    {
      "epoch": 1.1023622047244095,
      "eval_entropy": 1.9513436201959848,
      "eval_loss": 0.03779410198330879,
      "eval_mean_token_accuracy": 0.9891293849796057,
      "eval_num_tokens": 1144556.0,
      "eval_runtime": 14.4692,
      "eval_samples_per_second": 70.08,
      "eval_steps_per_second": 4.423,
      "step": 280
    },
    {
      "entropy": 1.9635676503181458,
      "epoch": 1.1811023622047245,
      "grad_norm": 2.812749147415161,
      "learning_rate": 4.114173228346457e-06,
      "loss": 0.0257,
      "mean_token_accuracy": 0.9924521818757057,
      "num_tokens": 1228054.0,
      "step": 300
    },
    {
      "epoch": 1.1811023622047245,
      "eval_entropy": 1.9615301713347435,
      "eval_loss": 0.043686393648386,
      "eval_mean_token_accuracy": 0.9885605089366436,
      "eval_num_tokens": 1228054.0,
      "eval_runtime": 15.1051,
      "eval_samples_per_second": 67.13,
      "eval_steps_per_second": 4.237,
      "step": 300
    },
    {
      "entropy": 1.9595869064331055,
      "epoch": 1.2598425196850394,
      "grad_norm": 0.0898560956120491,
      "learning_rate": 3.7204724409448824e-06,
      "loss": 0.016,
      "mean_token_accuracy": 0.9946568146347999,
      "num_tokens": 1310062.0,
      "step": 320
    },
    {
      "epoch": 1.2598425196850394,
      "eval_entropy": 1.9571783430874348,
      "eval_loss": 0.04049532115459442,
      "eval_mean_token_accuracy": 0.9888405678793788,
      "eval_num_tokens": 1310062.0,
      "eval_runtime": 14.8004,
      "eval_samples_per_second": 68.512,
      "eval_steps_per_second": 4.324,
      "step": 320
    },
    {
      "entropy": 1.9704302370548248,
      "epoch": 1.3385826771653544,
      "grad_norm": 0.1551266759634018,
      "learning_rate": 3.3267716535433077e-06,
      "loss": 0.0272,
      "mean_token_accuracy": 0.991837514936924,
      "num_tokens": 1392015.0,
      "step": 340
    },
    {
      "epoch": 1.3385826771653544,
      "eval_entropy": 1.9510800130665302,
      "eval_loss": 0.040886349976062775,
      "eval_mean_token_accuracy": 0.9888559766113758,
      "eval_num_tokens": 1392015.0,
      "eval_runtime": 14.4788,
      "eval_samples_per_second": 70.033,
      "eval_steps_per_second": 4.42,
      "step": 340
    },
    {
      "entropy": 1.9628229856491088,
      "epoch": 1.4173228346456692,
      "grad_norm": 0.49692580103874207,
      "learning_rate": 2.9330708661417322e-06,
      "loss": 0.0242,
      "mean_token_accuracy": 0.99283407330513,
      "num_tokens": 1472981.0,
      "step": 360
    },
    {
      "epoch": 1.4173228346456692,
      "eval_entropy": 1.943079276010394,
      "eval_loss": 0.03727104514837265,
      "eval_mean_token_accuracy": 0.9910982735455036,
      "eval_num_tokens": 1472981.0,
      "eval_runtime": 14.4234,
      "eval_samples_per_second": 70.303,
      "eval_steps_per_second": 4.437,
      "step": 360
    },
    {
      "entropy": 1.957309964299202,
      "epoch": 1.4960629921259843,
      "grad_norm": 2.6952269077301025,
      "learning_rate": 2.5393700787401576e-06,
      "loss": 0.0304,
      "mean_token_accuracy": 0.9920013844966888,
      "num_tokens": 1554129.0,
      "step": 380
    },
    {
      "epoch": 1.4960629921259843,
      "eval_entropy": 1.942601079121232,
      "eval_loss": 0.03651127964258194,
      "eval_mean_token_accuracy": 0.9905505627393723,
      "eval_num_tokens": 1554129.0,
      "eval_runtime": 14.4247,
      "eval_samples_per_second": 70.296,
      "eval_steps_per_second": 4.437,
      "step": 380
    },
    {
      "entropy": 1.9591941982507706,
      "epoch": 1.574803149606299,
      "grad_norm": 4.7762556076049805,
      "learning_rate": 2.145669291338583e-06,
      "loss": 0.0251,
      "mean_token_accuracy": 0.9915260329842568,
      "num_tokens": 1636074.0,
      "step": 400
    },
    {
      "epoch": 1.574803149606299,
      "eval_entropy": 1.9433052614331245,
      "eval_loss": 0.039551958441734314,
      "eval_mean_token_accuracy": 0.9907209984958172,
      "eval_num_tokens": 1636074.0,
      "eval_runtime": 14.4728,
      "eval_samples_per_second": 70.063,
      "eval_steps_per_second": 4.422,
      "step": 400
    },
    {
      "entropy": 1.972610130906105,
      "epoch": 1.6535433070866141,
      "grad_norm": 6.626554012298584,
      "learning_rate": 1.7519685039370079e-06,
      "loss": 0.0172,
      "mean_token_accuracy": 0.9950673922896385,
      "num_tokens": 1718697.0,
      "step": 420
    },
    {
      "epoch": 1.6535433070866141,
      "eval_entropy": 1.9396098591387272,
      "eval_loss": 0.03601359203457832,
      "eval_mean_token_accuracy": 0.9912841068580747,
      "eval_num_tokens": 1718697.0,
      "eval_runtime": 15.1369,
      "eval_samples_per_second": 66.989,
      "eval_steps_per_second": 4.228,
      "step": 420
    }
  ],
  "logging_steps": 20,
  "max_steps": 508,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 20,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.5052021162399744e+16,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}