trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 40,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.025,
      "grad_norm": 6.698416653714476,
      "learning_rate": 1.25e-07,
      "logits/chosen": -2.8582587242126465,
      "logits/rejected": -2.842068910598755,
      "logps/chosen": -261.8958435058594,
      "logps/rejected": -226.5897216796875,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.883361102730669,
      "learning_rate": 4.6650635094610966e-07,
      "logits/chosen": -2.7813661098480225,
      "logits/rejected": -2.772981643676758,
      "logps/chosen": -254.5003204345703,
      "logps/rejected": -251.1380157470703,
      "loss": 0.6908,
      "rewards/accuracies": 0.5381944179534912,
      "rewards/chosen": 0.005406244192272425,
      "rewards/margins": 0.0035838009789586067,
      "rewards/rejected": 0.001822443911805749,
      "step": 10
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.43798743744834,
      "learning_rate": 2.934120444167326e-07,
      "logits/chosen": -2.7719171047210693,
      "logits/rejected": -2.7447400093078613,
      "logps/chosen": -265.3804931640625,
      "logps/rejected": -252.05618286132812,
      "loss": 0.6742,
      "rewards/accuracies": 0.6812499761581421,
      "rewards/chosen": 0.03401705250144005,
      "rewards/margins": 0.04408254474401474,
      "rewards/rejected": -0.01006548572331667,
      "step": 20
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.502439531203193,
      "learning_rate": 8.930309757836516e-08,
      "logits/chosen": -2.761791467666626,
      "logits/rejected": -2.740384578704834,
      "logps/chosen": -258.3692626953125,
      "logps/rejected": -249.21566772460938,
      "loss": 0.6515,
      "rewards/accuracies": 0.6781250238418579,
      "rewards/chosen": -0.02331426367163658,
      "rewards/margins": 0.07361713796854019,
      "rewards/rejected": -0.09693139791488647,
      "step": 30
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.122574066847829,
      "learning_rate": 0.0,
      "logits/chosen": -2.7583651542663574,
      "logits/rejected": -2.733116626739502,
      "logps/chosen": -266.1463317871094,
      "logps/rejected": -273.2423095703125,
      "loss": 0.6468,
      "rewards/accuracies": 0.6000000238418579,
      "rewards/chosen": -0.05725393444299698,
      "rewards/margins": 0.06221733242273331,
      "rewards/rejected": -0.11947127431631088,
      "step": 40
    },
    {
      "epoch": 1.0,
      "step": 40,
      "total_flos": 0.0,
      "train_loss": 0.6658878326416016,
      "train_runtime": 1111.8376,
      "train_samples_per_second": 9.164,
      "train_steps_per_second": 0.036
    }
  ],
  "logging_steps": 10,
  "max_steps": 40,
  "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
}