trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9905956112852664,
  "eval_steps": 500,
  "global_step": 79,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 6.25e-08,
      "logits/chosen": -2.903411865234375,
      "logits/rejected": -2.9185478687286377,
      "logps/chosen": -358.677001953125,
      "logps/rejected": -304.1064147949219,
      "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": 4.990217055187362e-07,
      "logits/chosen": -2.8053393363952637,
      "logits/rejected": -2.7965915203094482,
      "logps/chosen": -263.0384521484375,
      "logps/rejected": -252.10650634765625,
      "loss": 0.6921,
      "rewards/accuracies": 0.5243055820465088,
      "rewards/chosen": 0.0014568159822374582,
      "rewards/margins": 0.0028856326825916767,
      "rewards/rejected": -0.0014288168167695403,
      "step": 10
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.655786431300069e-07,
      "logits/chosen": -2.811448335647583,
      "logits/rejected": -2.793362617492676,
      "logps/chosen": -276.836181640625,
      "logps/rejected": -251.5447235107422,
      "loss": 0.6772,
      "rewards/accuracies": 0.659375011920929,
      "rewards/chosen": 0.029580000787973404,
      "rewards/margins": 0.03599635511636734,
      "rewards/rejected": -0.0064163520000875,
      "step": 20
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.9061232191019517e-07,
      "logits/chosen": -2.7584967613220215,
      "logits/rejected": -2.755288600921631,
      "logps/chosen": -261.2607421875,
      "logps/rejected": -263.68487548828125,
      "loss": 0.6542,
      "rewards/accuracies": 0.668749988079071,
      "rewards/chosen": -0.034566882997751236,
      "rewards/margins": 0.07634096592664719,
      "rewards/rejected": -0.11090785264968872,
      "step": 30
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.8856223324132555e-07,
      "logits/chosen": -2.76475191116333,
      "logits/rejected": -2.7666561603546143,
      "logps/chosen": -279.80926513671875,
      "logps/rejected": -270.8001708984375,
      "loss": 0.6295,
      "rewards/accuracies": 0.721875011920929,
      "rewards/chosen": -0.14598490297794342,
      "rewards/margins": 0.18713520467281342,
      "rewards/rejected": -0.3331201374530792,
      "step": 40
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.7908455541642582e-07,
      "logits/chosen": -2.7629075050354004,
      "logits/rejected": -2.7455806732177734,
      "logps/chosen": -316.5531311035156,
      "logps/rejected": -301.55743408203125,
      "loss": 0.6111,
      "rewards/accuracies": 0.7250000238418579,
      "rewards/chosen": -0.26032739877700806,
      "rewards/margins": 0.25766921043395996,
      "rewards/rejected": -0.517996609210968,
      "step": 50
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.32661172908373e-08,
      "logits/chosen": -2.762693405151367,
      "logits/rejected": -2.748161792755127,
      "logps/chosen": -270.93939208984375,
      "logps/rejected": -294.34466552734375,
      "loss": 0.5964,
      "rewards/accuracies": 0.721875011920929,
      "rewards/chosen": -0.27937284111976624,
      "rewards/margins": 0.3220180571079254,
      "rewards/rejected": -0.6013908386230469,
      "step": 60
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.956279997278043e-08,
      "logits/chosen": -2.765657424926758,
      "logits/rejected": -2.7370893955230713,
      "logps/chosen": -308.8651428222656,
      "logps/rejected": -304.91033935546875,
      "loss": 0.5938,
      "rewards/accuracies": 0.6812499761581421,
      "rewards/chosen": -0.30938825011253357,
      "rewards/margins": 0.29292166233062744,
      "rewards/rejected": -0.6023099422454834,
      "step": 70
    },
    {
      "epoch": 0.99,
      "step": 79,
      "total_flos": 0.0,
      "train_loss": 0.6337358317797697,
      "train_runtime": 2585.4134,
      "train_samples_per_second": 7.882,
      "train_steps_per_second": 0.031
    }
  ],
  "logging_steps": 10,
  "max_steps": 79,
  "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
}