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": 8.161267569589059,
      "learning_rate": 1.25e-07,
      "logits/chosen": -2.73935866355896,
      "logits/rejected": -2.784054756164551,
      "logps/chosen": -260.22711181640625,
      "logps/pi_response": -172.46438598632812,
      "logps/ref_response": -172.46438598632812,
      "logps/rejected": -284.8775634765625,
      "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": 8.362628417432258,
      "learning_rate": 4.646121984004665e-07,
      "logits/chosen": -2.6796317100524902,
      "logits/rejected": -2.6501288414001465,
      "logps/chosen": -210.9478302001953,
      "logps/pi_response": -135.48536682128906,
      "logps/ref_response": -137.83311462402344,
      "logps/rejected": -229.79412841796875,
      "loss": 0.6853,
      "rewards/accuracies": 0.5659722089767456,
      "rewards/chosen": 0.0010574186453595757,
      "rewards/margins": 0.016374297440052032,
      "rewards/rejected": -0.01531688030809164,
      "step": 10
    },
    {
      "epoch": 0.5095541401273885,
      "grad_norm": 11.834934635253203,
      "learning_rate": 2.8355831645441387e-07,
      "logits/chosen": -2.712939739227295,
      "logits/rejected": -2.6800286769866943,
      "logps/chosen": -234.5419921875,
      "logps/pi_response": -130.1776123046875,
      "logps/ref_response": -137.71713256835938,
      "logps/rejected": -254.71969604492188,
      "loss": 0.6276,
      "rewards/accuracies": 0.6781250238418579,
      "rewards/chosen": -0.08936784416437149,
      "rewards/margins": 0.1819755733013153,
      "rewards/rejected": -0.2713434100151062,
      "step": 20
    },
    {
      "epoch": 0.7643312101910829,
      "grad_norm": 10.70634334365682,
      "learning_rate": 7.723433775328384e-08,
      "logits/chosen": -2.7188973426818848,
      "logits/rejected": -2.702681541442871,
      "logps/chosen": -253.486328125,
      "logps/pi_response": -143.45779418945312,
      "logps/ref_response": -144.81507873535156,
      "logps/rejected": -301.4783020019531,
      "loss": 0.6049,
      "rewards/accuracies": 0.675000011920929,
      "rewards/chosen": -0.28306064009666443,
      "rewards/margins": 0.31163957715034485,
      "rewards/rejected": -0.5947002172470093,
      "step": 30
    },
    {
      "epoch": 0.9936305732484076,
      "step": 39,
      "total_flos": 0.0,
      "train_loss": 0.6222284512642102,
      "train_runtime": 1801.62,
      "train_samples_per_second": 5.551,
      "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
}