{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.5150988345888867,
  "eval_steps": 500,
  "global_step": 10000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.9996726480206315e-05,
      "loss": 0.7422,
      "step": 100
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9986906778099784e-05,
      "loss": 0.6673,
      "step": 200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9970543465279565e-05,
      "loss": 0.6749,
      "step": 300
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994764082699591e-05,
      "loss": 0.6481,
      "step": 400
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991820486102801e-05,
      "loss": 0.6505,
      "step": 500
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9882243276113245e-05,
      "loss": 0.6708,
      "step": 600
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.983976548992841e-05,
      "loss": 0.6596,
      "step": 700
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9790782626623436e-05,
      "loss": 0.6657,
      "step": 800
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.97353075139081e-05,
      "loss": 0.659,
      "step": 900
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9673354679692785e-05,
      "loss": 0.6519,
      "step": 1000
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.960494034828381e-05,
      "loss": 0.638,
      "step": 1100
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9530082436134614e-05,
      "loss": 0.629,
      "step": 1200
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.944880054715378e-05,
      "loss": 0.6464,
      "step": 1300
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9361115967571094e-05,
      "loss": 0.642,
      "step": 1400
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.926705166036311e-05,
      "loss": 0.6431,
      "step": 1500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.916663225923953e-05,
      "loss": 0.6465,
      "step": 1600
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.905988406219217e-05,
      "loss": 0.6439,
      "step": 1700
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8946835024607885e-05,
      "loss": 0.652,
      "step": 1800
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8827514751947656e-05,
      "loss": 0.6376,
      "step": 1900
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8701954491993426e-05,
      "loss": 0.6357,
      "step": 2000
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.85701871266649e-05,
      "loss": 0.6338,
      "step": 2100
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8432247163408365e-05,
      "loss": 0.6591,
      "step": 2200
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8288170726159815e-05,
      "loss": 0.621,
      "step": 2300
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8137995545884794e-05,
      "loss": 0.6337,
      "step": 2400
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.798176095069731e-05,
      "loss": 0.637,
      "step": 2500
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.781950785556054e-05,
      "loss": 0.6464,
      "step": 2600
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7651278751571984e-05,
      "loss": 0.6539,
      "step": 2700
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.747711769483576e-05,
      "loss": 0.6233,
      "step": 2800
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.729707029492521e-05,
      "loss": 0.6476,
      "step": 2900
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.711118370293852e-05,
      "loss": 0.6302,
      "step": 3000
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.691950659915074e-05,
      "loss": 0.632,
      "step": 3100
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.672208918026535e-05,
      "loss": 0.6281,
      "step": 3200
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.6518983146268604e-05,
      "loss": 0.6514,
      "step": 3300
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.63102416868903e-05,
      "loss": 0.6254,
      "step": 3400
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.609591946767437e-05,
      "loss": 0.6412,
      "step": 3500
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.587607261566294e-05,
      "loss": 0.6647,
      "step": 3600
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.565075870469777e-05,
      "loss": 0.644,
      "step": 3700
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.542003674034263e-05,
      "loss": 0.6328,
      "step": 3800
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.5183967144430904e-05,
      "loss": 0.6454,
      "step": 3900
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.4942611739242166e-05,
      "loss": 0.6619,
      "step": 4000
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.469603373131214e-05,
      "loss": 0.6326,
      "step": 4100
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.444429769487997e-05,
      "loss": 0.6373,
      "step": 4200
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.4187469554977503e-05,
      "loss": 0.6375,
      "step": 4300
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.392561657016472e-05,
      "loss": 0.6181,
      "step": 4400
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.3658807314915915e-05,
      "loss": 0.6415,
      "step": 4500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.338711166166135e-05,
      "loss": 0.6235,
      "step": 4600
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.3110600762488915e-05,
      "loss": 0.643,
      "step": 4700
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.282934703051076e-05,
      "loss": 0.6488,
      "step": 4800
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.2543424120899556e-05,
      "loss": 0.6536,
      "step": 4900
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.2252906911599646e-05,
      "loss": 0.6396,
      "step": 5000
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.195787148371782e-05,
      "loss": 0.6456,
      "step": 5100
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.165839510159914e-05,
      "loss": 0.6213,
      "step": 5200
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.1354556192592766e-05,
      "loss": 0.6408,
      "step": 5300
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.1046434326513305e-05,
      "loss": 0.6394,
      "step": 5400
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.073411019480297e-05,
      "loss": 0.6338,
      "step": 5500
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.041766558939991e-05,
      "loss": 0.6271,
      "step": 5600
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.009718338131851e-05,
      "loss": 0.6269,
      "step": 5700
    },
    {
      "epoch": 0.3,
      "learning_rate": 3.97727474989469e-05,
      "loss": 0.6465,
      "step": 5800
    },
    {
      "epoch": 0.3,
      "learning_rate": 3.944444290606768e-05,
      "loss": 0.6372,
      "step": 5900
    },
    {
      "epoch": 0.31,
      "learning_rate": 3.911235557960752e-05,
      "loss": 0.6399,
      "step": 6000
    },
    {
      "epoch": 0.31,
      "learning_rate": 3.877657248712131e-05,
      "loss": 0.62,
      "step": 6100
    },
    {
      "epoch": 0.32,
      "learning_rate": 3.844059304094324e-05,
      "loss": 0.6468,
      "step": 6200
    },
    {
      "epoch": 0.32,
      "learning_rate": 3.8097717913954655e-05,
      "loss": 0.6362,
      "step": 6300
    },
    {
      "epoch": 0.33,
      "learning_rate": 3.775141273585859e-05,
      "loss": 0.6302,
      "step": 6400
    },
    {
      "epoch": 0.33,
      "learning_rate": 3.740176819760346e-05,
      "loss": 0.6334,
      "step": 6500
    },
    {
      "epoch": 0.34,
      "learning_rate": 3.704887586465459e-05,
      "loss": 0.6507,
      "step": 6600
    },
    {
      "epoch": 0.35,
      "learning_rate": 3.669282815301495e-05,
      "loss": 0.6403,
      "step": 6700
    },
    {
      "epoch": 0.35,
      "learning_rate": 3.6333718305023066e-05,
      "loss": 0.6407,
      "step": 6800
    },
    {
      "epoch": 0.36,
      "learning_rate": 3.597164036493457e-05,
      "loss": 0.6477,
      "step": 6900
    },
    {
      "epoch": 0.36,
      "learning_rate": 3.560668915429376e-05,
      "loss": 0.656,
      "step": 7000
    },
    {
      "epoch": 0.37,
      "learning_rate": 3.5238960247101584e-05,
      "loss": 0.6252,
      "step": 7100
    },
    {
      "epoch": 0.37,
      "learning_rate": 3.486854994478655e-05,
      "loss": 0.6345,
      "step": 7200
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.4499297670475796e-05,
      "loss": 0.6355,
      "step": 7300
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.412384064706181e-05,
      "loss": 0.6468,
      "step": 7400
    },
    {
      "epoch": 0.39,
      "learning_rate": 3.374599425781161e-05,
      "loss": 0.6356,
      "step": 7500
    },
    {
      "epoch": 0.39,
      "learning_rate": 3.336585745373593e-05,
      "loss": 0.6335,
      "step": 7600
    },
    {
      "epoch": 0.4,
      "learning_rate": 3.2983529785662964e-05,
      "loss": 0.6092,
      "step": 7700
    },
    {
      "epoch": 0.4,
      "learning_rate": 3.2599111378167855e-05,
      "loss": 0.6337,
      "step": 7800
    },
    {
      "epoch": 0.41,
      "learning_rate": 3.221657650708416e-05,
      "loss": 0.6455,
      "step": 7900
    },
    {
      "epoch": 0.41,
      "learning_rate": 3.182829754439824e-05,
      "loss": 0.627,
      "step": 8000
    },
    {
      "epoch": 0.42,
      "learning_rate": 3.143823037633881e-05,
      "loss": 0.6272,
      "step": 8100
    },
    {
      "epoch": 0.42,
      "learning_rate": 3.104647715431351e-05,
      "loss": 0.6337,
      "step": 8200
    },
    {
      "epoch": 0.43,
      "learning_rate": 3.0653140471276476e-05,
      "loss": 0.6562,
      "step": 8300
    },
    {
      "epoch": 0.43,
      "learning_rate": 3.0258323334861104e-05,
      "loss": 0.6184,
      "step": 8400
    },
    {
      "epoch": 0.44,
      "learning_rate": 2.9862129140404272e-05,
      "loss": 0.6175,
      "step": 8500
    },
    {
      "epoch": 0.44,
      "learning_rate": 2.9464661643868984e-05,
      "loss": 0.6489,
      "step": 8600
    },
    {
      "epoch": 0.45,
      "learning_rate": 2.9066024934672632e-05,
      "loss": 0.6423,
      "step": 8700
    },
    {
      "epoch": 0.45,
      "learning_rate": 2.866632340842786e-05,
      "loss": 0.6127,
      "step": 8800
    },
    {
      "epoch": 0.46,
      "learning_rate": 2.8265661739603277e-05,
      "loss": 0.6247,
      "step": 8900
    },
    {
      "epoch": 0.46,
      "learning_rate": 2.7864144854111168e-05,
      "loss": 0.6307,
      "step": 9000
    },
    {
      "epoch": 0.47,
      "learning_rate": 2.7461877901829303e-05,
      "loss": 0.6231,
      "step": 9100
    },
    {
      "epoch": 0.47,
      "learning_rate": 2.7058966229064143e-05,
      "loss": 0.6304,
      "step": 9200
    },
    {
      "epoch": 0.48,
      "learning_rate": 2.6655515350962545e-05,
      "loss": 0.6408,
      "step": 9300
    },
    {
      "epoch": 0.48,
      "learning_rate": 2.625163092387934e-05,
      "loss": 0.6198,
      "step": 9400
    },
    {
      "epoch": 0.49,
      "learning_rate": 2.584741871770784e-05,
      "loss": 0.6378,
      "step": 9500
    },
    {
      "epoch": 0.49,
      "learning_rate": 2.5442984588180657e-05,
      "loss": 0.6303,
      "step": 9600
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.5038434449148057e-05,
      "loss": 0.6297,
      "step": 9700
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.463387424484106e-05,
      "loss": 0.6258,
      "step": 9800
    },
    {
      "epoch": 0.51,
      "learning_rate": 2.4229409922126572e-05,
      "loss": 0.6349,
      "step": 9900
    },
    {
      "epoch": 0.52,
      "learning_rate": 2.382514740276191e-05,
      "loss": 0.6259,
      "step": 10000
    }
  ],
  "logging_steps": 100,
  "max_steps": 19413,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 2000,
  "total_flos": 2.268305935999461e+19,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}