{
  "best_metric": 2.683746337890625,
  "best_model_checkpoint": "models/opt-babylm2-rewritten-clean-spacy-earlystop-bpe_seed-211_1e-3/checkpoint-45100",
  "epoch": 19.991464360935595,
  "eval_steps": 500,
  "global_step": 45100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.4434098215275468,
      "grad_norm": 0.535893440246582,
      "learning_rate": 3.125e-05,
      "loss": 5.5759,
      "step": 1000
    },
    {
      "epoch": 0.8868196430550936,
      "grad_norm": 0.5292273163795471,
      "learning_rate": 6.25e-05,
      "loss": 4.1,
      "step": 2000
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.36012162808432535,
      "eval_loss": 3.8208348751068115,
      "eval_runtime": 73.9931,
      "eval_samples_per_second": 820.076,
      "eval_steps_per_second": 12.826,
      "step": 2256
    },
    {
      "epoch": 1.3298969072164948,
      "grad_norm": 0.6377725601196289,
      "learning_rate": 9.375e-05,
      "loss": 3.7036,
      "step": 3000
    },
    {
      "epoch": 1.7733067287440418,
      "grad_norm": 0.5576291680335999,
      "learning_rate": 0.000125,
      "loss": 3.4569,
      "step": 4000
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.40816362273320667,
      "eval_loss": 3.308465003967285,
      "eval_runtime": 73.782,
      "eval_samples_per_second": 822.423,
      "eval_steps_per_second": 12.862,
      "step": 4512
    },
    {
      "epoch": 2.2163839929054427,
      "grad_norm": 0.5923783183097839,
      "learning_rate": 0.00015625,
      "loss": 3.2553,
      "step": 5000
    },
    {
      "epoch": 2.6597938144329896,
      "grad_norm": 0.4800017178058624,
      "learning_rate": 0.0001875,
      "loss": 3.1304,
      "step": 6000
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.4293314979254721,
      "eval_loss": 3.0989913940429688,
      "eval_runtime": 73.5868,
      "eval_samples_per_second": 824.605,
      "eval_steps_per_second": 12.896,
      "step": 6768
    },
    {
      "epoch": 3.102871078594391,
      "grad_norm": 0.40440040826797485,
      "learning_rate": 0.00021875,
      "loss": 3.0393,
      "step": 7000
    },
    {
      "epoch": 3.5462809001219378,
      "grad_norm": 0.3914831876754761,
      "learning_rate": 0.00025,
      "loss": 2.9686,
      "step": 8000
    },
    {
      "epoch": 3.9896907216494846,
      "grad_norm": 0.37147632241249084,
      "learning_rate": 0.00028125000000000003,
      "loss": 2.9241,
      "step": 9000
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.44051643465560253,
      "eval_loss": 2.9892077445983887,
      "eval_runtime": 73.1132,
      "eval_samples_per_second": 829.946,
      "eval_steps_per_second": 12.98,
      "step": 9024
    },
    {
      "epoch": 4.4327679858108855,
      "grad_norm": 0.3617306649684906,
      "learning_rate": 0.0003125,
      "loss": 2.8599,
      "step": 10000
    },
    {
      "epoch": 4.876177807338433,
      "grad_norm": 0.31954342126846313,
      "learning_rate": 0.00034375,
      "loss": 2.8456,
      "step": 11000
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.44670376258611555,
      "eval_loss": 2.9302828311920166,
      "eval_runtime": 72.5611,
      "eval_samples_per_second": 836.261,
      "eval_steps_per_second": 13.079,
      "step": 11280
    },
    {
      "epoch": 5.319255071499834,
      "grad_norm": 0.32251182198524475,
      "learning_rate": 0.000375,
      "loss": 2.7976,
      "step": 12000
    },
    {
      "epoch": 5.762664893027381,
      "grad_norm": 0.312112957239151,
      "learning_rate": 0.00040625000000000004,
      "loss": 2.7882,
      "step": 13000
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.4509860793361511,
      "eval_loss": 2.888702869415283,
      "eval_runtime": 72.4551,
      "eval_samples_per_second": 837.484,
      "eval_steps_per_second": 13.098,
      "step": 13536
    },
    {
      "epoch": 6.205742157188782,
      "grad_norm": 0.28876274824142456,
      "learning_rate": 0.0004375,
      "loss": 2.7559,
      "step": 14000
    },
    {
      "epoch": 6.649151978716328,
      "grad_norm": 0.29188501834869385,
      "learning_rate": 0.00046875,
      "loss": 2.7426,
      "step": 15000
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.4542816058526245,
      "eval_loss": 2.8618035316467285,
      "eval_runtime": 72.3378,
      "eval_samples_per_second": 838.842,
      "eval_steps_per_second": 13.119,
      "step": 15792
    },
    {
      "epoch": 7.09222924287773,
      "grad_norm": 0.28137603402137756,
      "learning_rate": 0.0005,
      "loss": 2.7305,
      "step": 16000
    },
    {
      "epoch": 7.535639064405276,
      "grad_norm": 0.25505542755126953,
      "learning_rate": 0.00053121875,
      "loss": 2.707,
      "step": 17000
    },
    {
      "epoch": 7.979048885932824,
      "grad_norm": 0.24025394022464752,
      "learning_rate": 0.0005624687499999999,
      "loss": 2.7124,
      "step": 18000
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.4565116264040224,
      "eval_loss": 2.840358257293701,
      "eval_runtime": 72.5997,
      "eval_samples_per_second": 835.816,
      "eval_steps_per_second": 13.072,
      "step": 18048
    },
    {
      "epoch": 8.422126150094225,
      "grad_norm": 0.24389903247356415,
      "learning_rate": 0.0005936875,
      "loss": 2.6715,
      "step": 19000
    },
    {
      "epoch": 8.865535971621771,
      "grad_norm": 0.23571985960006714,
      "learning_rate": 0.0006249375000000001,
      "loss": 2.689,
      "step": 20000
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.45795106440730543,
      "eval_loss": 2.8292860984802246,
      "eval_runtime": 72.5327,
      "eval_samples_per_second": 836.588,
      "eval_steps_per_second": 13.084,
      "step": 20304
    },
    {
      "epoch": 9.308613235783172,
      "grad_norm": 0.2347807139158249,
      "learning_rate": 0.0006561562500000001,
      "loss": 2.6603,
      "step": 21000
    },
    {
      "epoch": 9.75202305731072,
      "grad_norm": 0.21437448263168335,
      "learning_rate": 0.00068734375,
      "loss": 2.6674,
      "step": 22000
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.45918653948065713,
      "eval_loss": 2.817204236984253,
      "eval_runtime": 72.5446,
      "eval_samples_per_second": 836.45,
      "eval_steps_per_second": 13.082,
      "step": 22560
    },
    {
      "epoch": 10.19510032147212,
      "grad_norm": 0.21189159154891968,
      "learning_rate": 0.00071859375,
      "loss": 2.6505,
      "step": 23000
    },
    {
      "epoch": 10.638510142999667,
      "grad_norm": 0.20313581824302673,
      "learning_rate": 0.0007498437500000001,
      "loss": 2.651,
      "step": 24000
    },
    {
      "epoch": 11.0,
      "eval_accuracy": 0.4604101231791332,
      "eval_loss": 2.8066623210906982,
      "eval_runtime": 72.4079,
      "eval_samples_per_second": 838.03,
      "eval_steps_per_second": 13.106,
      "step": 24816
    },
    {
      "epoch": 11.081587407161068,
      "grad_norm": 0.20089128613471985,
      "learning_rate": 0.00078109375,
      "loss": 2.6463,
      "step": 25000
    },
    {
      "epoch": 11.524997228688616,
      "grad_norm": 0.20063355565071106,
      "learning_rate": 0.0008123125,
      "loss": 2.633,
      "step": 26000
    },
    {
      "epoch": 11.968407050216163,
      "grad_norm": 0.18298469483852386,
      "learning_rate": 0.0008435625,
      "loss": 2.6483,
      "step": 27000
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.4612719247224269,
      "eval_loss": 2.800128221511841,
      "eval_runtime": 72.4998,
      "eval_samples_per_second": 836.967,
      "eval_steps_per_second": 13.09,
      "step": 27072
    },
    {
      "epoch": 12.411484314377564,
      "grad_norm": 0.19970569014549255,
      "learning_rate": 0.00087478125,
      "loss": 2.6154,
      "step": 28000
    },
    {
      "epoch": 12.854894135905111,
      "grad_norm": 0.17127935588359833,
      "learning_rate": 0.0009060312499999999,
      "loss": 2.639,
      "step": 29000
    },
    {
      "epoch": 13.0,
      "eval_accuracy": 0.46190145669342225,
      "eval_loss": 2.7980594635009766,
      "eval_runtime": 72.3226,
      "eval_samples_per_second": 839.018,
      "eval_steps_per_second": 13.122,
      "step": 29328
    },
    {
      "epoch": 13.297971400066512,
      "grad_norm": 0.19391290843486786,
      "learning_rate": 0.00093725,
      "loss": 2.6128,
      "step": 30000
    },
    {
      "epoch": 13.741381221594057,
      "grad_norm": 0.1826080083847046,
      "learning_rate": 0.0009685000000000001,
      "loss": 2.6268,
      "step": 31000
    },
    {
      "epoch": 14.0,
      "eval_accuracy": 0.46204221438080834,
      "eval_loss": 2.7923851013183594,
      "eval_runtime": 72.136,
      "eval_samples_per_second": 841.189,
      "eval_steps_per_second": 13.156,
      "step": 31584
    },
    {
      "epoch": 14.18445848575546,
      "grad_norm": 0.1709497720003128,
      "learning_rate": 0.00099971875,
      "loss": 2.6164,
      "step": 32000
    },
    {
      "epoch": 14.627868307283006,
      "grad_norm": 0.1697809398174286,
      "learning_rate": 0.000924351145038168,
      "loss": 2.6116,
      "step": 33000
    },
    {
      "epoch": 15.0,
      "eval_accuracy": 0.46505868134992956,
      "eval_loss": 2.766456365585327,
      "eval_runtime": 72.5226,
      "eval_samples_per_second": 836.705,
      "eval_steps_per_second": 13.086,
      "step": 33840
    },
    {
      "epoch": 15.070945571444408,
      "grad_norm": 0.1920733004808426,
      "learning_rate": 0.0008480152671755725,
      "loss": 2.5972,
      "step": 34000
    },
    {
      "epoch": 15.514355392971954,
      "grad_norm": 0.17531998455524445,
      "learning_rate": 0.0007717557251908397,
      "loss": 2.5637,
      "step": 35000
    },
    {
      "epoch": 15.957765214499501,
      "grad_norm": 0.175877183675766,
      "learning_rate": 0.0006954198473282442,
      "loss": 2.5657,
      "step": 36000
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.4687057789496814,
      "eval_loss": 2.7436182498931885,
      "eval_runtime": 72.9943,
      "eval_samples_per_second": 831.298,
      "eval_steps_per_second": 13.001,
      "step": 36096
    },
    {
      "epoch": 16.400842478660902,
      "grad_norm": 0.18140995502471924,
      "learning_rate": 0.0006191603053435114,
      "loss": 2.509,
      "step": 37000
    },
    {
      "epoch": 16.84425230018845,
      "grad_norm": 0.17342546582221985,
      "learning_rate": 0.0005428244274809161,
      "loss": 2.5199,
      "step": 38000
    },
    {
      "epoch": 17.0,
      "eval_accuracy": 0.47174531777114276,
      "eval_loss": 2.7186553478240967,
      "eval_runtime": 72.9122,
      "eval_samples_per_second": 832.234,
      "eval_steps_per_second": 13.016,
      "step": 38352
    },
    {
      "epoch": 17.28732956434985,
      "grad_norm": 0.18188068270683289,
      "learning_rate": 0.0004665648854961832,
      "loss": 2.4714,
      "step": 39000
    },
    {
      "epoch": 17.730739385877396,
      "grad_norm": 0.18511821329593658,
      "learning_rate": 0.0003902290076335878,
      "loss": 2.4626,
      "step": 40000
    },
    {
      "epoch": 18.0,
      "eval_accuracy": 0.4746070676128065,
      "eval_loss": 2.7018399238586426,
      "eval_runtime": 72.5891,
      "eval_samples_per_second": 835.938,
      "eval_steps_per_second": 13.074,
      "step": 40608
    },
    {
      "epoch": 18.1738166500388,
      "grad_norm": 0.18930885195732117,
      "learning_rate": 0.0003138931297709924,
      "loss": 2.4292,
      "step": 41000
    },
    {
      "epoch": 18.617226471566344,
      "grad_norm": 0.19555842876434326,
      "learning_rate": 0.00023763358778625953,
      "loss": 2.3982,
      "step": 42000
    },
    {
      "epoch": 19.0,
      "eval_accuracy": 0.47732812439412153,
      "eval_loss": 2.6869590282440186,
      "eval_runtime": 72.705,
      "eval_samples_per_second": 834.605,
      "eval_steps_per_second": 13.053,
      "step": 42864
    },
    {
      "epoch": 19.060303735727746,
      "grad_norm": 0.2038242667913437,
      "learning_rate": 0.00016129770992366412,
      "loss": 2.3835,
      "step": 43000
    },
    {
      "epoch": 19.503713557255292,
      "grad_norm": 0.20078131556510925,
      "learning_rate": 8.50381679389313e-05,
      "loss": 2.3292,
      "step": 44000
    },
    {
      "epoch": 19.94712337878284,
      "grad_norm": 0.19901303946971893,
      "learning_rate": 8.702290076335879e-06,
      "loss": 2.3238,
      "step": 45000
    },
    {
      "epoch": 19.991464360935595,
      "eval_accuracy": 0.47886644176457666,
      "eval_loss": 2.683746337890625,
      "eval_runtime": 72.7112,
      "eval_samples_per_second": 834.534,
      "eval_steps_per_second": 13.052,
      "step": 45100
    },
    {
      "epoch": 19.991464360935595,
      "step": 45100,
      "total_flos": 1.507910045663232e+18,
      "train_loss": 2.80528530789056,
      "train_runtime": 30473.0676,
      "train_samples_per_second": 378.921,
      "train_steps_per_second": 1.48
    }
  ],
  "logging_steps": 1000,
  "max_steps": 45100,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 20,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 3,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.507910045663232e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}