trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9936305732484076,
  "eval_steps": 500,
  "global_step": 39,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.025477707006369428,
      "grad_norm": 9.541497731022089,
      "learning_rate": 1.25e-07,
      "logits/chosen": -2.9487781524658203,
      "logits/rejected": -2.858518123626709,
      "logps/chosen": -304.58709716796875,
      "logps/pi_response": -66.34120178222656,
      "logps/ref_response": -66.34120178222656,
      "logps/rejected": -133.11962890625,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.25477707006369427,
      "grad_norm": 7.426043579963901,
      "learning_rate": 4.646121984004665e-07,
      "logits/chosen": -2.768293619155884,
      "logits/rejected": -2.7385334968566895,
      "logps/chosen": -256.4764404296875,
      "logps/pi_response": -72.72017669677734,
      "logps/ref_response": -72.56087493896484,
      "logps/rejected": -173.5285186767578,
      "loss": 0.6872,
      "rewards/accuracies": 0.5972222089767456,
      "rewards/chosen": 0.02276873216032982,
      "rewards/margins": 0.0161480400711298,
      "rewards/rejected": 0.006620692554861307,
      "step": 10
    },
    {
      "epoch": 0.5095541401273885,
      "grad_norm": 6.093230482473603,
      "learning_rate": 2.8355831645441387e-07,
      "logits/chosen": -2.7015206813812256,
      "logits/rejected": -2.6768670082092285,
      "logps/chosen": -233.2022247314453,
      "logps/pi_response": -80.76576232910156,
      "logps/ref_response": -70.75907897949219,
      "logps/rejected": -180.19044494628906,
      "loss": 0.6568,
      "rewards/accuracies": 0.737500011920929,
      "rewards/chosen": 0.03768356516957283,
      "rewards/margins": 0.07391206920146942,
      "rewards/rejected": -0.03622850403189659,
      "step": 20
    },
    {
      "epoch": 0.7643312101910829,
      "grad_norm": 7.083904309264922,
      "learning_rate": 7.723433775328384e-08,
      "logits/chosen": -2.6932857036590576,
      "logits/rejected": -2.6497244834899902,
      "logps/chosen": -257.22515869140625,
      "logps/pi_response": -100.95832824707031,
      "logps/ref_response": -70.76898193359375,
      "logps/rejected": -196.4232635498047,
      "loss": 0.6258,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.04754001647233963,
      "rewards/margins": 0.1733211874961853,
      "rewards/rejected": -0.22086119651794434,
      "step": 30
    },
    {
      "epoch": 0.9936305732484076,
      "step": 39,
      "total_flos": 0.0,
      "train_loss": 0.6476471485235752,
      "train_runtime": 1779.1919,
      "train_samples_per_second": 5.621,
      "train_steps_per_second": 0.022
    }
  ],
  "logging_steps": 10,
  "max_steps": 39,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}