trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.992,
  "eval_steps": 200,
  "global_step": 62,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.016,
      "grad_norm": 96.80832392576436,
      "learning_rate": 7.142857142857142e-08,
      "logits/generated": -1.177710771560669,
      "logits/real": -0.5424066185951233,
      "logps/generated": -206.68331909179688,
      "logps/real": -268.4350280761719,
      "loss": 0.956,
      "rewards/accuracies": 0.0,
      "rewards/generated": 0.0,
      "rewards/margins": 0.0,
      "rewards/real": 0.0,
      "step": 1
    },
    {
      "epoch": 0.16,
      "grad_norm": 186.85535273079182,
      "learning_rate": 4.727272727272727e-07,
      "logits/generated": -0.9581692218780518,
      "logits/real": -0.7258952260017395,
      "logps/generated": -263.182373046875,
      "logps/real": -268.8551025390625,
      "loss": 0.9057,
      "rewards/accuracies": 0.5972222089767456,
      "rewards/generated": 0.11249147355556488,
      "rewards/margins": 0.11431904882192612,
      "rewards/real": 0.22681055963039398,
      "step": 10
    },
    {
      "epoch": 0.32,
      "grad_norm": 118.11338598390404,
      "learning_rate": 3.818181818181818e-07,
      "logits/generated": -0.891255259513855,
      "logits/real": -0.6647359132766724,
      "logps/generated": -264.0565185546875,
      "logps/real": -256.42901611328125,
      "loss": 0.785,
      "rewards/accuracies": 0.7250000238418579,
      "rewards/generated": 0.7130780220031738,
      "rewards/margins": 0.5250438451766968,
      "rewards/real": 1.2381219863891602,
      "step": 20
    },
    {
      "epoch": 0.48,
      "grad_norm": 89.31545609503772,
      "learning_rate": 2.909090909090909e-07,
      "logits/generated": -0.8781732320785522,
      "logits/real": -0.5925976634025574,
      "logps/generated": -256.00799560546875,
      "logps/real": -247.38980102539062,
      "loss": 0.7838,
      "rewards/accuracies": 0.75,
      "rewards/generated": 1.1103136539459229,
      "rewards/margins": 0.712864875793457,
      "rewards/real": 1.8231786489486694,
      "step": 30
    },
    {
      "epoch": 0.64,
      "grad_norm": 88.34237989806086,
      "learning_rate": 2e-07,
      "logits/generated": -0.8793425559997559,
      "logits/real": -0.6354281306266785,
      "logps/generated": -263.5761413574219,
      "logps/real": -255.0084686279297,
      "loss": 0.7397,
      "rewards/accuracies": 0.7749999761581421,
      "rewards/generated": 1.0507025718688965,
      "rewards/margins": 0.881763756275177,
      "rewards/real": 1.9324661493301392,
      "step": 40
    },
    {
      "epoch": 0.8,
      "grad_norm": 91.22613773952635,
      "learning_rate": 1.0909090909090908e-07,
      "logits/generated": -0.8102799654006958,
      "logits/real": -0.5707312822341919,
      "logps/generated": -248.6565704345703,
      "logps/real": -260.7277526855469,
      "loss": 0.7153,
      "rewards/accuracies": 0.800000011920929,
      "rewards/generated": 1.1541557312011719,
      "rewards/margins": 0.9233818054199219,
      "rewards/real": 2.0775375366210938,
      "step": 50
    },
    {
      "epoch": 0.96,
      "grad_norm": 86.27514871900985,
      "learning_rate": 1.818181818181818e-08,
      "logits/generated": -0.9142637252807617,
      "logits/real": -0.6623004078865051,
      "logps/generated": -254.97494506835938,
      "logps/real": -244.65640258789062,
      "loss": 0.7192,
      "rewards/accuracies": 0.8125,
      "rewards/generated": 1.215023159980774,
      "rewards/margins": 0.8442209959030151,
      "rewards/real": 2.059244155883789,
      "step": 60
    },
    {
      "epoch": 0.992,
      "step": 62,
      "total_flos": 0.0,
      "train_loss": 0.770801761457997,
      "train_runtime": 772.1202,
      "train_samples_per_second": 2.585,
      "train_steps_per_second": 0.08
    }
  ],
  "logging_steps": 10,
  "max_steps": 62,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 200,
  "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": 4,
  "trial_name": null,
  "trial_params": null
}