{
  "best_metric": 2.989222764968872,
  "best_model_checkpoint": "models/opt-babylm2-rewritten-clean-spacy-32k-earlystop-40epochs_seed-42_3e-4/checkpoint-40566",
  "epoch": 24.0,
  "eval_steps": 500,
  "global_step": 46362,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.5176653293645658,
      "grad_norm": 0.485008180141449,
      "learning_rate": 9.375e-06,
      "loss": 6.8783,
      "step": 1000
    },
    {
      "epoch": 0.9996117510029766,
      "eval_accuracy": 0.289060183567583,
      "eval_loss": 4.46536922454834,
      "eval_runtime": 111.7462,
      "eval_samples_per_second": 467.04,
      "eval_steps_per_second": 7.302,
      "step": 1931
    },
    {
      "epoch": 1.0353306587291315,
      "grad_norm": 0.6161150932312012,
      "learning_rate": 1.875e-05,
      "loss": 4.6542,
      "step": 2000
    },
    {
      "epoch": 1.5529959880936974,
      "grad_norm": 0.6660575270652771,
      "learning_rate": 2.8125e-05,
      "loss": 4.2263,
      "step": 3000
    },
    {
      "epoch": 1.9997411673353178,
      "eval_accuracy": 0.3337067802786951,
      "eval_loss": 3.917879343032837,
      "eval_runtime": 112.2732,
      "eval_samples_per_second": 464.848,
      "eval_steps_per_second": 7.268,
      "step": 3863
    },
    {
      "epoch": 2.070661317458263,
      "grad_norm": 0.6615888476371765,
      "learning_rate": 3.75e-05,
      "loss": 3.9582,
      "step": 4000
    },
    {
      "epoch": 2.588326646822829,
      "grad_norm": 0.7149969339370728,
      "learning_rate": 4.6874999999999994e-05,
      "loss": 3.7724,
      "step": 5000
    },
    {
      "epoch": 2.9998705836676587,
      "eval_accuracy": 0.3562454681048507,
      "eval_loss": 3.6396915912628174,
      "eval_runtime": 112.1753,
      "eval_samples_per_second": 465.254,
      "eval_steps_per_second": 7.274,
      "step": 5795
    },
    {
      "epoch": 3.105991976187395,
      "grad_norm": 0.6370624899864197,
      "learning_rate": 5.625e-05,
      "loss": 3.6294,
      "step": 6000
    },
    {
      "epoch": 3.6236573055519608,
      "grad_norm": 0.6175299286842346,
      "learning_rate": 6.5625e-05,
      "loss": 3.5091,
      "step": 7000
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.3723592153857136,
      "eval_loss": 3.456868886947632,
      "eval_runtime": 111.9653,
      "eval_samples_per_second": 466.126,
      "eval_steps_per_second": 7.288,
      "step": 7727
    },
    {
      "epoch": 4.141322634916526,
      "grad_norm": 0.5864672064781189,
      "learning_rate": 7.5e-05,
      "loss": 3.4133,
      "step": 8000
    },
    {
      "epoch": 4.658987964281092,
      "grad_norm": 0.5497373342514038,
      "learning_rate": 8.437499999999999e-05,
      "loss": 3.3306,
      "step": 9000
    },
    {
      "epoch": 4.999611751002977,
      "eval_accuracy": 0.3838290710037608,
      "eval_loss": 3.3310000896453857,
      "eval_runtime": 112.0797,
      "eval_samples_per_second": 465.651,
      "eval_steps_per_second": 7.281,
      "step": 9658
    },
    {
      "epoch": 5.176653293645658,
      "grad_norm": 0.5599066019058228,
      "learning_rate": 9.374999999999999e-05,
      "loss": 3.2608,
      "step": 10000
    },
    {
      "epoch": 5.694318623010224,
      "grad_norm": 0.5453025698661804,
      "learning_rate": 0.00010312499999999999,
      "loss": 3.2012,
      "step": 11000
    },
    {
      "epoch": 5.999741167335317,
      "eval_accuracy": 0.3918066341309469,
      "eval_loss": 3.2469069957733154,
      "eval_runtime": 112.2873,
      "eval_samples_per_second": 464.79,
      "eval_steps_per_second": 7.267,
      "step": 11590
    },
    {
      "epoch": 6.21198395237479,
      "grad_norm": 0.5039867162704468,
      "learning_rate": 0.0001125,
      "loss": 3.152,
      "step": 12000
    },
    {
      "epoch": 6.729649281739356,
      "grad_norm": 0.4990510642528534,
      "learning_rate": 0.000121875,
      "loss": 3.1088,
      "step": 13000
    },
    {
      "epoch": 6.999870583667659,
      "eval_accuracy": 0.39817732342985096,
      "eval_loss": 3.1827573776245117,
      "eval_runtime": 112.5977,
      "eval_samples_per_second": 463.509,
      "eval_steps_per_second": 7.247,
      "step": 13522
    },
    {
      "epoch": 7.2473146111039215,
      "grad_norm": 0.46827396750450134,
      "learning_rate": 0.00013125,
      "loss": 3.0691,
      "step": 14000
    },
    {
      "epoch": 7.764979940468487,
      "grad_norm": 0.4815771281719208,
      "learning_rate": 0.000140625,
      "loss": 3.0364,
      "step": 15000
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.40226096953439355,
      "eval_loss": 3.1404154300689697,
      "eval_runtime": 112.1878,
      "eval_samples_per_second": 465.202,
      "eval_steps_per_second": 7.274,
      "step": 15454
    },
    {
      "epoch": 8.282645269833052,
      "grad_norm": 0.4806288778781891,
      "learning_rate": 0.000149990625,
      "loss": 3.0022,
      "step": 16000
    },
    {
      "epoch": 8.80031059919762,
      "grad_norm": 0.4368716776371002,
      "learning_rate": 0.00015936562499999999,
      "loss": 2.9837,
      "step": 17000
    },
    {
      "epoch": 8.999611751002977,
      "eval_accuracy": 0.4056956294684956,
      "eval_loss": 3.1080353260040283,
      "eval_runtime": 112.3344,
      "eval_samples_per_second": 464.595,
      "eval_steps_per_second": 7.264,
      "step": 17385
    },
    {
      "epoch": 9.317975928562184,
      "grad_norm": 0.44254282116889954,
      "learning_rate": 0.00016873124999999998,
      "loss": 2.9495,
      "step": 18000
    },
    {
      "epoch": 9.835641257926751,
      "grad_norm": 0.4331772029399872,
      "learning_rate": 0.00017809687499999998,
      "loss": 2.9377,
      "step": 19000
    },
    {
      "epoch": 9.999741167335317,
      "eval_accuracy": 0.40769541156182726,
      "eval_loss": 3.083951711654663,
      "eval_runtime": 112.4544,
      "eval_samples_per_second": 464.099,
      "eval_steps_per_second": 7.256,
      "step": 19317
    },
    {
      "epoch": 10.353306587291316,
      "grad_norm": 0.425889790058136,
      "learning_rate": 0.00018747187499999998,
      "loss": 2.9051,
      "step": 20000
    },
    {
      "epoch": 10.870971916655883,
      "grad_norm": 0.40670710802078247,
      "learning_rate": 0.00019684687499999998,
      "loss": 2.9019,
      "step": 21000
    },
    {
      "epoch": 10.99987058366766,
      "eval_accuracy": 0.41011139539164965,
      "eval_loss": 3.063291311264038,
      "eval_runtime": 112.0395,
      "eval_samples_per_second": 465.818,
      "eval_steps_per_second": 7.283,
      "step": 21249
    },
    {
      "epoch": 11.388637246020448,
      "grad_norm": 0.4176114499568939,
      "learning_rate": 0.00020622187499999998,
      "loss": 2.8676,
      "step": 22000
    },
    {
      "epoch": 11.906302575385013,
      "grad_norm": 0.400343120098114,
      "learning_rate": 0.00021559687499999997,
      "loss": 2.8713,
      "step": 23000
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.4117278120291995,
      "eval_loss": 3.0504517555236816,
      "eval_runtime": 111.9439,
      "eval_samples_per_second": 466.216,
      "eval_steps_per_second": 7.289,
      "step": 23181
    },
    {
      "epoch": 12.42396790474958,
      "grad_norm": 0.3873593509197235,
      "learning_rate": 0.00022496249999999997,
      "loss": 2.8353,
      "step": 24000
    },
    {
      "epoch": 12.941633234114144,
      "grad_norm": 0.3830123543739319,
      "learning_rate": 0.00023433749999999997,
      "loss": 2.8449,
      "step": 25000
    },
    {
      "epoch": 12.999611751002977,
      "eval_accuracy": 0.41297589125705847,
      "eval_loss": 3.037565231323242,
      "eval_runtime": 112.2612,
      "eval_samples_per_second": 464.898,
      "eval_steps_per_second": 7.269,
      "step": 25112
    },
    {
      "epoch": 13.459298563478711,
      "grad_norm": 0.390222430229187,
      "learning_rate": 0.00024369374999999997,
      "loss": 2.8068,
      "step": 26000
    },
    {
      "epoch": 13.976963892843276,
      "grad_norm": 0.3650892376899719,
      "learning_rate": 0.00025306875,
      "loss": 2.8231,
      "step": 27000
    },
    {
      "epoch": 13.999741167335317,
      "eval_accuracy": 0.41429497800269754,
      "eval_loss": 3.026965856552124,
      "eval_runtime": 112.249,
      "eval_samples_per_second": 464.948,
      "eval_steps_per_second": 7.27,
      "step": 27044
    },
    {
      "epoch": 14.494629222207843,
      "grad_norm": 0.3750116527080536,
      "learning_rate": 0.00026244375,
      "loss": 2.7828,
      "step": 28000
    },
    {
      "epoch": 14.99987058366766,
      "eval_accuracy": 0.4150191044035932,
      "eval_loss": 3.0222034454345703,
      "eval_runtime": 112.1123,
      "eval_samples_per_second": 465.516,
      "eval_steps_per_second": 7.278,
      "step": 28976
    },
    {
      "epoch": 15.012294551572408,
      "grad_norm": 0.39160436391830444,
      "learning_rate": 0.00027181875,
      "loss": 2.804,
      "step": 29000
    },
    {
      "epoch": 15.529959880936975,
      "grad_norm": 0.36566755175590515,
      "learning_rate": 0.00028118437499999993,
      "loss": 2.7644,
      "step": 30000
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.4156012157689288,
      "eval_loss": 3.0159823894500732,
      "eval_runtime": 112.075,
      "eval_samples_per_second": 465.671,
      "eval_steps_per_second": 7.281,
      "step": 30908
    },
    {
      "epoch": 16.04762521030154,
      "grad_norm": 0.37446266412734985,
      "learning_rate": 0.000290559375,
      "loss": 2.7817,
      "step": 31000
    },
    {
      "epoch": 16.565290539666105,
      "grad_norm": 0.3600151836872101,
      "learning_rate": 0.000299925,
      "loss": 2.7508,
      "step": 32000
    },
    {
      "epoch": 16.999611751002977,
      "eval_accuracy": 0.4166737674184447,
      "eval_loss": 3.010031223297119,
      "eval_runtime": 112.0828,
      "eval_samples_per_second": 465.638,
      "eval_steps_per_second": 7.28,
      "step": 32839
    },
    {
      "epoch": 17.082955869030673,
      "grad_norm": 0.35380175709724426,
      "learning_rate": 0.00029342175066312994,
      "loss": 2.7621,
      "step": 33000
    },
    {
      "epoch": 17.60062119839524,
      "grad_norm": 0.3466060161590576,
      "learning_rate": 0.0002867970822281167,
      "loss": 2.7296,
      "step": 34000
    },
    {
      "epoch": 17.99974116733532,
      "eval_accuracy": 0.41775473477377156,
      "eval_loss": 3.002298355102539,
      "eval_runtime": 112.139,
      "eval_samples_per_second": 465.405,
      "eval_steps_per_second": 7.277,
      "step": 34771
    },
    {
      "epoch": 18.118286527759803,
      "grad_norm": 0.35987117886543274,
      "learning_rate": 0.0002801657824933687,
      "loss": 2.7304,
      "step": 35000
    },
    {
      "epoch": 18.63595185712437,
      "grad_norm": 0.345238596200943,
      "learning_rate": 0.0002735411140583554,
      "loss": 2.7053,
      "step": 36000
    },
    {
      "epoch": 18.999870583667658,
      "eval_accuracy": 0.41883562698886795,
      "eval_loss": 2.9966797828674316,
      "eval_runtime": 112.4594,
      "eval_samples_per_second": 464.079,
      "eval_steps_per_second": 7.256,
      "step": 36703
    },
    {
      "epoch": 19.153617186488933,
      "grad_norm": 0.35703110694885254,
      "learning_rate": 0.0002669098143236074,
      "loss": 2.701,
      "step": 37000
    },
    {
      "epoch": 19.671282515853502,
      "grad_norm": 0.3403065800666809,
      "learning_rate": 0.00026028514588859414,
      "loss": 2.6821,
      "step": 38000
    },
    {
      "epoch": 20.0,
      "eval_accuracy": 0.419519853927392,
      "eval_loss": 2.9925894737243652,
      "eval_runtime": 112.0678,
      "eval_samples_per_second": 465.7,
      "eval_steps_per_second": 7.281,
      "step": 38635
    },
    {
      "epoch": 20.188947845218067,
      "grad_norm": 0.36988621950149536,
      "learning_rate": 0.0002536538461538461,
      "loss": 2.6747,
      "step": 39000
    },
    {
      "epoch": 20.706613174582632,
      "grad_norm": 0.3578760623931885,
      "learning_rate": 0.0002470291777188329,
      "loss": 2.6601,
      "step": 40000
    },
    {
      "epoch": 20.999611751002977,
      "eval_accuracy": 0.4202126468521879,
      "eval_loss": 2.989222764968872,
      "eval_runtime": 112.0379,
      "eval_samples_per_second": 465.824,
      "eval_steps_per_second": 7.283,
      "step": 40566
    },
    {
      "epoch": 21.224278503947197,
      "grad_norm": 0.36684709787368774,
      "learning_rate": 0.00024039787798408487,
      "loss": 2.6476,
      "step": 41000
    },
    {
      "epoch": 21.741943833311765,
      "grad_norm": 0.3607647120952606,
      "learning_rate": 0.0002337732095490716,
      "loss": 2.6406,
      "step": 42000
    },
    {
      "epoch": 21.99974116733532,
      "eval_accuracy": 0.4203659329223163,
      "eval_loss": 2.9898085594177246,
      "eval_runtime": 112.2513,
      "eval_samples_per_second": 464.939,
      "eval_steps_per_second": 7.269,
      "step": 42498
    },
    {
      "epoch": 22.25960916267633,
      "grad_norm": 0.3788565695285797,
      "learning_rate": 0.0002271419098143236,
      "loss": 2.6251,
      "step": 43000
    },
    {
      "epoch": 22.777274492040895,
      "grad_norm": 0.3606908619403839,
      "learning_rate": 0.00022051061007957558,
      "loss": 2.621,
      "step": 44000
    },
    {
      "epoch": 22.999870583667658,
      "eval_accuracy": 0.42054829826163076,
      "eval_loss": 2.992137908935547,
      "eval_runtime": 112.0224,
      "eval_samples_per_second": 465.889,
      "eval_steps_per_second": 7.284,
      "step": 44430
    },
    {
      "epoch": 23.29493982140546,
      "grad_norm": 0.3780686855316162,
      "learning_rate": 0.00021389257294429705,
      "loss": 2.5996,
      "step": 45000
    },
    {
      "epoch": 23.812605150770025,
      "grad_norm": 0.3720705807209015,
      "learning_rate": 0.00020726127320954907,
      "loss": 2.6048,
      "step": 46000
    },
    {
      "epoch": 24.0,
      "eval_accuracy": 0.4207038385386728,
      "eval_loss": 2.992816925048828,
      "eval_runtime": 111.6981,
      "eval_samples_per_second": 467.242,
      "eval_steps_per_second": 7.305,
      "step": 46362
    },
    {
      "epoch": 24.0,
      "step": 46362,
      "total_flos": 1.550560221462528e+18,
      "train_loss": 3.0875192876465034,
      "train_runtime": 49059.1729,
      "train_samples_per_second": 403.2,
      "train_steps_per_second": 1.574
    }
  ],
  "logging_steps": 1000,
  "max_steps": 77240,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 40,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 3,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 3
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.550560221462528e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}