trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9984,
  "eval_steps": 100,
  "global_step": 78,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 6.25e-07,
      "logits/chosen": -2.7010061740875244,
      "logits/rejected": -2.6382064819335938,
      "logps/chosen": -391.27899169921875,
      "logps/rejected": -409.36163330078125,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.989935734988098e-06,
      "logits/chosen": -2.668689250946045,
      "logits/rejected": -2.6425745487213135,
      "logps/chosen": -330.8897705078125,
      "logps/rejected": -337.9491271972656,
      "loss": 0.6926,
      "rewards/accuracies": 0.4913194477558136,
      "rewards/chosen": -0.0030369642190635204,
      "rewards/margins": 0.0014728567330166698,
      "rewards/rejected": -0.004509821534156799,
      "step": 10
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.646121984004666e-06,
      "logits/chosen": -2.662149429321289,
      "logits/rejected": -2.6085782051086426,
      "logps/chosen": -327.1595458984375,
      "logps/rejected": -321.137939453125,
      "loss": 0.6875,
      "rewards/accuracies": 0.635937511920929,
      "rewards/chosen": -0.009846845641732216,
      "rewards/margins": 0.014805642887949944,
      "rewards/rejected": -0.02465249039232731,
      "step": 20
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.8772424536302565e-06,
      "logits/chosen": -2.605135440826416,
      "logits/rejected": -2.577131748199463,
      "logps/chosen": -320.46856689453125,
      "logps/rejected": -349.97857666015625,
      "loss": 0.6796,
      "rewards/accuracies": 0.6187499761581421,
      "rewards/chosen": -0.053877659142017365,
      "rewards/margins": 0.03185194730758667,
      "rewards/rejected": -0.08572960644960403,
      "step": 30
    },
    {
      "epoch": 0.51,
      "learning_rate": 2.835583164544139e-06,
      "logits/chosen": -2.6149070262908936,
      "logits/rejected": -2.5788040161132812,
      "logps/chosen": -337.58331298828125,
      "logps/rejected": -333.2330017089844,
      "loss": 0.6754,
      "rewards/accuracies": 0.5843750238418579,
      "rewards/chosen": -0.07810668647289276,
      "rewards/margins": 0.03418826684355736,
      "rewards/rejected": -0.11229495704174042,
      "step": 40
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.7274575140626318e-06,
      "logits/chosen": -2.6245689392089844,
      "logits/rejected": -2.574202537536621,
      "logps/chosen": -352.3199462890625,
      "logps/rejected": -360.7944030761719,
      "loss": 0.6687,
      "rewards/accuracies": 0.629687488079071,
      "rewards/chosen": -0.05965545028448105,
      "rewards/margins": 0.05930715054273605,
      "rewards/rejected": -0.1189626082777977,
      "step": 50
    },
    {
      "epoch": 0.77,
      "learning_rate": 7.723433775328385e-07,
      "logits/chosen": -2.583293914794922,
      "logits/rejected": -2.5575852394104004,
      "logps/chosen": -340.70123291015625,
      "logps/rejected": -365.50408935546875,
      "loss": 0.6609,
      "rewards/accuracies": 0.651562511920929,
      "rewards/chosen": -0.1051960214972496,
      "rewards/margins": 0.07506568729877472,
      "rewards/rejected": -0.18026170134544373,
      "step": 60
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.59412823400657e-07,
      "logits/chosen": -2.5673248767852783,
      "logits/rejected": -2.5238747596740723,
      "logps/chosen": -322.8841857910156,
      "logps/rejected": -349.5296936035156,
      "loss": 0.6624,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.09875378012657166,
      "rewards/margins": 0.07134519517421722,
      "rewards/rejected": -0.17009896039962769,
      "step": 70
    },
    {
      "epoch": 1.0,
      "step": 78,
      "total_flos": 0.0,
      "train_loss": 0.6737474944346991,
      "train_runtime": 6751.8687,
      "train_samples_per_second": 2.962,
      "train_steps_per_second": 0.012
    }
  ],
  "logging_steps": 10,
  "max_steps": 78,
  "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
}