{
  "best_global_step": 600,
  "best_metric": 1.2342418432235718,
  "best_model_checkpoint": "/scratch/multilingual_model_sft/checkpoint-600",
  "epoch": 2.4,
  "eval_steps": 100,
  "global_step": 600,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "entropy": 0.9756445910781621,
      "epoch": 0.08,
      "grad_norm": 3.6301491260528564,
      "learning_rate": 1e-05,
      "loss": 2.7480094909667967,
      "mean_token_accuracy": 0.5633710252121091,
      "num_tokens": 100509.0,
      "step": 20
    },
    {
      "entropy": 1.4356517121195793,
      "epoch": 0.16,
      "grad_norm": 1.0212956666946411,
      "learning_rate": 1.9999902656029183e-05,
      "loss": 1.8655733108520507,
      "mean_token_accuracy": 0.6183225071057677,
      "num_tokens": 194221.0,
      "step": 40
    },
    {
      "entropy": 1.4469021618366242,
      "epoch": 0.24,
      "grad_norm": 0.5234550833702087,
      "learning_rate": 1.995710194507367e-05,
      "loss": 1.4041429519653321,
      "mean_token_accuracy": 0.6799660105258226,
      "num_tokens": 292021.0,
      "step": 60
    },
    {
      "entropy": 1.3612655106931926,
      "epoch": 0.32,
      "grad_norm": 0.5519381165504456,
      "learning_rate": 1.9836810309184468e-05,
      "loss": 1.3517327308654785,
      "mean_token_accuracy": 0.6959601622074842,
      "num_tokens": 383296.0,
      "step": 80
    },
    {
      "entropy": 1.3319022830575704,
      "epoch": 0.4,
      "grad_norm": 0.6105360388755798,
      "learning_rate": 1.9639963915348547e-05,
      "loss": 1.309855842590332,
      "mean_token_accuracy": 0.6991457279771567,
      "num_tokens": 480575.0,
      "step": 100
    },
    {
      "epoch": 0.4,
      "eval_entropy": 1.3079482184516058,
      "eval_loss": 1.3204385042190552,
      "eval_mean_token_accuracy": 0.6943969362311893,
      "eval_num_tokens": 480575.0,
      "eval_runtime": 10.6851,
      "eval_samples_per_second": 26.673,
      "eval_steps_per_second": 3.369,
      "step": 100
    },
    {
      "entropy": 1.3555128309875726,
      "epoch": 0.48,
      "grad_norm": 0.6203790307044983,
      "learning_rate": 1.93680947162578e-05,
      "loss": 1.3396360397338867,
      "mean_token_accuracy": 0.7012265287339687,
      "num_tokens": 570090.0,
      "step": 120
    },
    {
      "entropy": 1.3061895180493592,
      "epoch": 0.56,
      "grad_norm": 0.4907386302947998,
      "learning_rate": 1.9023318527921213e-05,
      "loss": 1.2728803634643555,
      "mean_token_accuracy": 0.7048281095921993,
      "num_tokens": 667523.0,
      "step": 140
    },
    {
      "entropy": 1.3211671005934478,
      "epoch": 0.64,
      "grad_norm": 0.6228592395782471,
      "learning_rate": 1.8608318563373735e-05,
      "loss": 1.2911997795104981,
      "mean_token_accuracy": 0.6987178284674883,
      "num_tokens": 766038.0,
      "step": 160
    },
    {
      "entropy": 1.3343474190682172,
      "epoch": 0.72,
      "grad_norm": 0.5544903874397278,
      "learning_rate": 1.8126324550630364e-05,
      "loss": 1.318357276916504,
      "mean_token_accuracy": 0.7034054283052683,
      "num_tokens": 859651.0,
      "step": 180
    },
    {
      "entropy": 1.3386923231184482,
      "epoch": 0.8,
      "grad_norm": 0.6091942191123962,
      "learning_rate": 1.7581087597399556e-05,
      "loss": 1.3211461067199708,
      "mean_token_accuracy": 0.6983358260244131,
      "num_tokens": 954094.0,
      "step": 200
    },
    {
      "epoch": 0.8,
      "eval_entropy": 1.2773916522661846,
      "eval_loss": 1.2775527238845825,
      "eval_mean_token_accuracy": 0.701493599348598,
      "eval_num_tokens": 954094.0,
      "eval_runtime": 9.9476,
      "eval_samples_per_second": 28.65,
      "eval_steps_per_second": 3.619,
      "step": 200
    },
    {
      "entropy": 1.2788743745535611,
      "epoch": 0.88,
      "grad_norm": 0.6575713753700256,
      "learning_rate": 1.69768509981707e-05,
      "loss": 1.2579275131225587,
      "mean_token_accuracy": 0.7133958246558905,
      "num_tokens": 1043359.0,
      "step": 220
    },
    {
      "entropy": 1.2770573951303958,
      "epoch": 0.96,
      "grad_norm": 0.6401338577270508,
      "learning_rate": 1.6318317210868936e-05,
      "loss": 1.250250244140625,
      "mean_token_accuracy": 0.7097325529903173,
      "num_tokens": 1139550.0,
      "step": 240
    },
    {
      "entropy": 1.2663291092962026,
      "epoch": 1.04,
      "grad_norm": 0.6415250897407532,
      "learning_rate": 1.56106112600807e-05,
      "loss": 1.2160707473754884,
      "mean_token_accuracy": 0.7165820356458426,
      "num_tokens": 1237885.0,
      "step": 260
    },
    {
      "entropy": 1.2433586917817592,
      "epoch": 1.12,
      "grad_norm": 0.5600829124450684,
      "learning_rate": 1.485924085166359e-05,
      "loss": 1.2002551078796386,
      "mean_token_accuracy": 0.7145083237439394,
      "num_tokens": 1337347.0,
      "step": 280
    },
    {
      "entropy": 1.3045797925442457,
      "epoch": 1.2,
      "grad_norm": 0.6553295850753784,
      "learning_rate": 1.4070053509147777e-05,
      "loss": 1.2605380058288573,
      "mean_token_accuracy": 0.7103811588138342,
      "num_tokens": 1424998.0,
      "step": 300
    },
    {
      "epoch": 1.2,
      "eval_entropy": 1.2426132311423619,
      "eval_loss": 1.2581056356430054,
      "eval_mean_token_accuracy": 0.7045401218864653,
      "eval_num_tokens": 1424998.0,
      "eval_runtime": 9.9679,
      "eval_samples_per_second": 28.592,
      "eval_steps_per_second": 3.612,
      "step": 300
    },
    {
      "entropy": 1.2380406064912677,
      "epoch": 1.28,
      "grad_norm": 0.5739977359771729,
      "learning_rate": 1.3249191065513857e-05,
      "loss": 1.208388614654541,
      "mean_token_accuracy": 0.7182681906968356,
      "num_tokens": 1523052.0,
      "step": 320
    },
    {
      "entropy": 1.2422979518771171,
      "epoch": 1.3599999999999999,
      "grad_norm": 0.6654541492462158,
      "learning_rate": 1.2403041864514005e-05,
      "loss": 1.20194673538208,
      "mean_token_accuracy": 0.7169692400842905,
      "num_tokens": 1619061.0,
      "step": 340
    },
    {
      "entropy": 1.2528483916074038,
      "epoch": 1.44,
      "grad_norm": 0.5971714854240417,
      "learning_rate": 1.153819104352846e-05,
      "loss": 1.2315282821655273,
      "mean_token_accuracy": 0.7162977565079928,
      "num_tokens": 1710487.0,
      "step": 360
    },
    {
      "entropy": 1.2364515990018845,
      "epoch": 1.52,
      "grad_norm": 0.7230240702629089,
      "learning_rate": 1.0661369284880005e-05,
      "loss": 1.1993119239807128,
      "mean_token_accuracy": 0.7190557043999434,
      "num_tokens": 1805041.0,
      "step": 380
    },
    {
      "entropy": 1.195871875807643,
      "epoch": 1.6,
      "grad_norm": 0.6772177219390869,
      "learning_rate": 9.779400434448137e-06,
      "loss": 1.1887290954589844,
      "mean_token_accuracy": 0.7250112488865852,
      "num_tokens": 1898912.0,
      "step": 400
    },
    {
      "epoch": 1.6,
      "eval_entropy": 1.218405197064082,
      "eval_loss": 1.2451484203338623,
      "eval_mean_token_accuracy": 0.706767362025049,
      "eval_num_tokens": 1898912.0,
      "eval_runtime": 9.9407,
      "eval_samples_per_second": 28.67,
      "eval_steps_per_second": 3.621,
      "step": 400
    },
    {
      "entropy": 1.2644711177796126,
      "epoch": 1.6800000000000002,
      "grad_norm": 0.6553245782852173,
      "learning_rate": 8.899148395239946e-06,
      "loss": 1.2322025299072266,
      "mean_token_accuracy": 0.7138959027826786,
      "num_tokens": 1997219.0,
      "step": 420
    },
    {
      "entropy": 1.245160198956728,
      "epoch": 1.76,
      "grad_norm": 0.6900923848152161,
      "learning_rate": 8.027463709217266e-06,
      "loss": 1.2088998794555663,
      "mean_token_accuracy": 0.7181152984499931,
      "num_tokens": 2089697.0,
      "step": 440
    },
    {
      "entropy": 1.2373197358101606,
      "epoch": 1.8399999999999999,
      "grad_norm": 0.6978564858436584,
      "learning_rate": 7.1711302431058795e-06,
      "loss": 1.2115466117858886,
      "mean_token_accuracy": 0.7191348981112242,
      "num_tokens": 2185513.0,
      "step": 460
    },
    {
      "entropy": 1.2476738758385182,
      "epoch": 1.92,
      "grad_norm": 0.7304058074951172,
      "learning_rate": 6.33681239310327e-06,
      "loss": 1.223146629333496,
      "mean_token_accuracy": 0.7194057062268258,
      "num_tokens": 2278767.0,
      "step": 480
    },
    {
      "entropy": 1.2755128189921379,
      "epoch": 2.0,
      "grad_norm": 0.6474395990371704,
      "learning_rate": 5.531003219363079e-06,
      "loss": 1.2292745590209961,
      "mean_token_accuracy": 0.7124675642699003,
      "num_tokens": 2371346.0,
      "step": 500
    },
    {
      "epoch": 2.0,
      "eval_entropy": 1.214999525083436,
      "eval_loss": 1.2368030548095703,
      "eval_mean_token_accuracy": 0.7081699487235811,
      "eval_num_tokens": 2371346.0,
      "eval_runtime": 9.9336,
      "eval_samples_per_second": 28.69,
      "eval_steps_per_second": 3.624,
      "step": 500
    },
    {
      "entropy": 1.167585114017129,
      "epoch": 2.08,
      "grad_norm": 0.7057409882545471,
      "learning_rate": 4.759973913898578e-06,
      "loss": 1.1290258407592773,
      "mean_token_accuracy": 0.733013367280364,
      "num_tokens": 2466083.0,
      "step": 520
    },
    {
      "entropy": 1.2017585724592208,
      "epoch": 2.16,
      "grad_norm": 0.7214968204498291,
      "learning_rate": 4.029724995169951e-06,
      "loss": 1.1627693176269531,
      "mean_token_accuracy": 0.7261429976671934,
      "num_tokens": 2561750.0,
      "step": 540
    },
    {
      "entropy": 1.185857030004263,
      "epoch": 2.24,
      "grad_norm": 0.7629018425941467,
      "learning_rate": 3.345939609182558e-06,
      "loss": 1.175245475769043,
      "mean_token_accuracy": 0.7282546646893024,
      "num_tokens": 2658568.0,
      "step": 560
    },
    {
      "entropy": 1.236609160900116,
      "epoch": 2.32,
      "grad_norm": 0.6379859447479248,
      "learning_rate": 2.713939300529255e-06,
      "loss": 1.1822381019592285,
      "mean_token_accuracy": 0.7198391534388066,
      "num_tokens": 2757472.0,
      "step": 580
    },
    {
      "entropy": 1.2224572278559207,
      "epoch": 2.4,
      "grad_norm": 0.607354462146759,
      "learning_rate": 2.138642597587686e-06,
      "loss": 1.1898154258728026,
      "mean_token_accuracy": 0.7245916619896888,
      "num_tokens": 2850420.0,
      "step": 600
    },
    {
      "epoch": 2.4,
      "eval_entropy": 1.1946484976344638,
      "eval_loss": 1.2342418432235718,
      "eval_mean_token_accuracy": 0.7090677950117323,
      "eval_num_tokens": 2850420.0,
      "eval_runtime": 11.5935,
      "eval_samples_per_second": 24.583,
      "eval_steps_per_second": 3.105,
      "step": 600
    }
  ],
  "logging_steps": 20,
  "max_steps": 750,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.371131850832691e+16,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}