trainer_state.json

{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 7.64,
  "eval_steps": 500,
  "global_step": 30,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.64,
      "grad_norm": 3.945056676864624,
      "learning_rate": 7e-07,
      "logits/chosen": -3.070164442062378,
      "logits/rejected": -3.0701053142547607,
      "logps/chosen": -167.7537841796875,
      "logps/rejected": -176.98863220214844,
      "loss": 0.5708,
      "rewards/accuracies": 0.3333333432674408,
      "rewards/chosen": 0.0016566767590120435,
      "rewards/margins": 0.002745657227933407,
      "rewards/rejected": -0.001088749268092215,
      "step": 10
    },
    {
      "epoch": 5.0,
      "grad_norm": 1.3009190559387207,
      "learning_rate": 3.666666666666666e-07,
      "logits/chosen": -3.0801775455474854,
      "logits/rejected": -3.077198028564453,
      "logps/chosen": -166.36441040039062,
      "logps/rejected": -176.7521209716797,
      "loss": 0.5065,
      "rewards/accuracies": 0.6779661178588867,
      "rewards/chosen": 0.005665342789143324,
      "rewards/margins": 0.012972233816981316,
      "rewards/rejected": -0.007307084742933512,
      "step": 20
    },
    {
      "epoch": 7.64,
      "grad_norm": 4.014199256896973,
      "learning_rate": 3.3333333333333334e-08,
      "logits/chosen": -3.0732717514038086,
      "logits/rejected": -3.0763494968414307,
      "logps/chosen": -165.93182373046875,
      "logps/rejected": -177.59942626953125,
      "loss": 0.5649,
      "rewards/accuracies": 0.8409090638160706,
      "rewards/chosen": 0.007431319449096918,
      "rewards/margins": 0.016990546137094498,
      "rewards/rejected": -0.009561365470290184,
      "step": 30
    }
  ],
  "logging_steps": 10,
  "max_steps": 30,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}