trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.989247311827957,
  "eval_steps": 100,
  "global_step": 46,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 1e-07,
      "logits/chosen": -2.7254130840301514,
      "logits/rejected": -2.9678733348846436,
      "logps/chosen": -138.15951538085938,
      "logps/rejected": -195.77276611328125,
      "loss": 6953.125,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "rewards/safe_rewards": 0.0,
      "rewards/unsafe_rewards": 0.0,
      "step": 1
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.818756127755237e-07,
      "logits/chosen": -2.8674046993255615,
      "logits/rejected": -2.8880205154418945,
      "logps/chosen": -153.06439208984375,
      "logps/rejected": -155.67388916015625,
      "loss": 6371.6432,
      "rewards/accuracies": 0.5,
      "rewards/chosen": -0.003938814625144005,
      "rewards/margins": -0.0019423332996666431,
      "rewards/rejected": -0.0019964815583080053,
      "rewards/safe_rewards": -0.0036866520531475544,
      "rewards/unsafe_rewards": -0.004190977197140455,
      "step": 10
    },
    {
      "epoch": 0.43,
      "learning_rate": 3.52267159292835e-07,
      "logits/chosen": -2.7803285121917725,
      "logits/rejected": -2.840688943862915,
      "logps/chosen": -166.37338256835938,
      "logps/rejected": -179.26712036132812,
      "loss": 5594.4398,
      "rewards/accuracies": 0.612500011920929,
      "rewards/chosen": -0.06914173811674118,
      "rewards/margins": 0.02756076492369175,
      "rewards/rejected": -0.09670250117778778,
      "rewards/safe_rewards": -0.07211428880691528,
      "rewards/unsafe_rewards": -0.06616917997598648,
      "step": 20
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.6549578039787434e-07,
      "logits/chosen": -2.67883038520813,
      "logits/rejected": -2.7650046348571777,
      "logps/chosen": -171.47341918945312,
      "logps/rejected": -193.48011779785156,
      "loss": 4781.5113,
      "rewards/accuracies": 0.6499999761581421,
      "rewards/chosen": -0.24309594929218292,
      "rewards/margins": 0.06528899818658829,
      "rewards/rejected": -0.3083849549293518,
      "rewards/safe_rewards": -0.23420853912830353,
      "rewards/unsafe_rewards": -0.2519833743572235,
      "step": 30
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.5958610759736126e-08,
      "logits/chosen": -2.738619089126587,
      "logits/rejected": -2.759796142578125,
      "logps/chosen": -200.98873901367188,
      "logps/rejected": -213.40988159179688,
      "loss": 4346.852,
      "rewards/accuracies": 0.512499988079071,
      "rewards/chosen": -0.4134730398654938,
      "rewards/margins": 0.0412062332034111,
      "rewards/rejected": -0.4546792507171631,
      "rewards/safe_rewards": -0.40580517053604126,
      "rewards/unsafe_rewards": -0.4211408495903015,
      "step": 40
    },
    {
      "epoch": 0.99,
      "step": 46,
      "total_flos": 0.0,
      "train_loss": 5218.799974524457,
      "train_runtime": 254.7215,
      "train_samples_per_second": 5.838,
      "train_steps_per_second": 0.181
    }
  ],
  "logging_steps": 10,
  "max_steps": 46,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}