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": -2.9087584018707275,
      "logits/rejected": -2.8338208198547363,
      "logps/chosen": -352.96319580078125,
      "logps/pi_response": -77.43819427490234,
      "logps/ref_response": -77.43819427490234,
      "logps/rejected": -170.26690673828125,
      "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": -2.7865679264068604,
      "logits/rejected": -2.7636632919311523,
      "logps/chosen": -238.10020446777344,
      "logps/pi_response": -73.99278259277344,
      "logps/ref_response": -73.52912139892578,
      "logps/rejected": -170.78558349609375,
      "loss": 0.6904,
      "rewards/accuracies": 0.5243055820465088,
      "rewards/chosen": 0.003248802153393626,
      "rewards/margins": 0.00418997323140502,
      "rewards/rejected": -0.0009411710198037326,
      "step": 10
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.655786431300069e-07,
      "logits/chosen": -2.739081859588623,
      "logits/rejected": -2.6856980323791504,
      "logps/chosen": -250.3759765625,
      "logps/pi_response": -82.33766174316406,
      "logps/ref_response": -71.4201431274414,
      "logps/rejected": -176.65232849121094,
      "loss": 0.6582,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": 0.026225481182336807,
      "rewards/margins": 0.08432629704475403,
      "rewards/rejected": -0.05810081958770752,
      "step": 20
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.9061232191019517e-07,
      "logits/chosen": -2.6405444145202637,
      "logits/rejected": -2.6057441234588623,
      "logps/chosen": -238.1879119873047,
      "logps/pi_response": -110.0510025024414,
      "logps/ref_response": -68.3199691772461,
      "logps/rejected": -194.5875244140625,
      "loss": 0.6206,
      "rewards/accuracies": 0.731249988079071,
      "rewards/chosen": -0.12873004376888275,
      "rewards/margins": 0.19655928015708923,
      "rewards/rejected": -0.3252893388271332,
      "step": 30
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.8856223324132555e-07,
      "logits/chosen": -2.6333813667297363,
      "logits/rejected": -2.6113696098327637,
      "logps/chosen": -266.65216064453125,
      "logps/pi_response": -156.62741088867188,
      "logps/ref_response": -75.2870864868164,
      "logps/rejected": -239.99850463867188,
      "loss": 0.5682,
      "rewards/accuracies": 0.7406250238418579,
      "rewards/chosen": -0.29538047313690186,
      "rewards/margins": 0.37797990441322327,
      "rewards/rejected": -0.6733604669570923,
      "step": 40
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.7908455541642582e-07,
      "logits/chosen": -2.627793073654175,
      "logits/rejected": -2.594454288482666,
      "logps/chosen": -294.68048095703125,
      "logps/pi_response": -167.09994506835938,
      "logps/ref_response": -74.21197509765625,
      "logps/rejected": -259.50811767578125,
      "loss": 0.536,
      "rewards/accuracies": 0.793749988079071,
      "rewards/chosen": -0.39494794607162476,
      "rewards/margins": 0.4701464772224426,
      "rewards/rejected": -0.8650944828987122,
      "step": 50
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.32661172908373e-08,
      "logits/chosen": -2.622882604598999,
      "logits/rejected": -2.587216854095459,
      "logps/chosen": -264.9446716308594,
      "logps/pi_response": -174.61488342285156,
      "logps/ref_response": -66.34608459472656,
      "logps/rejected": -277.74664306640625,
      "loss": 0.5127,
      "rewards/accuracies": 0.7906249761581421,
      "rewards/chosen": -0.49885016679763794,
      "rewards/margins": 0.6039000153541565,
      "rewards/rejected": -1.1027500629425049,
      "step": 60
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.956279997278043e-08,
      "logits/chosen": -2.61812686920166,
      "logits/rejected": -2.5810680389404297,
      "logps/chosen": -314.621337890625,
      "logps/pi_response": -201.65823364257812,
      "logps/ref_response": -77.6395034790039,
      "logps/rejected": -290.4659423828125,
      "loss": 0.5012,
      "rewards/accuracies": 0.7749999761581421,
      "rewards/chosen": -0.5674734711647034,
      "rewards/margins": 0.6503495573997498,
      "rewards/rejected": -1.2178230285644531,
      "step": 70
    },
    {
      "epoch": 0.99,
      "step": 79,
      "total_flos": 0.0,
      "train_loss": 0.5758069738557067,
      "train_runtime": 4675.5433,
      "train_samples_per_second": 4.358,
      "train_steps_per_second": 0.017
    }
  ],
  "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
}