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": -0.9686082005500793,
      "logits/rejected": -0.7667080163955688,
      "logps/chosen": -304.88916015625,
      "logps/pi_response": -174.6251220703125,
      "logps/ref_response": -174.6251220703125,
      "logps/rejected": -560.704833984375,
      "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": -1.5470714569091797,
      "logits/rejected": -1.1779179573059082,
      "logps/chosen": -347.94091796875,
      "logps/pi_response": -176.1438751220703,
      "logps/ref_response": -174.41168212890625,
      "logps/rejected": -613.462646484375,
      "loss": 0.6405,
      "rewards/accuracies": 0.6423611044883728,
      "rewards/chosen": -0.11746331304311752,
      "rewards/margins": 0.18620280921459198,
      "rewards/rejected": -0.3036660850048065,
      "step": 10
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.655786431300069e-07,
      "logits/chosen": -0.04107005149126053,
      "logits/rejected": 0.5205573439598083,
      "logps/chosen": -461.720947265625,
      "logps/pi_response": -247.1218719482422,
      "logps/ref_response": -176.59713745117188,
      "logps/rejected": -896.8013916015625,
      "loss": 0.6234,
      "rewards/accuracies": 0.746874988079071,
      "rewards/chosen": -1.4934273958206177,
      "rewards/margins": 1.6034536361694336,
      "rewards/rejected": -3.0968809127807617,
      "step": 20
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.9061232191019517e-07,
      "logits/chosen": 0.25087007880210876,
      "logits/rejected": 0.9124622344970703,
      "logps/chosen": -447.6351013183594,
      "logps/pi_response": -237.4333038330078,
      "logps/ref_response": -185.14527893066406,
      "logps/rejected": -821.3446044921875,
      "loss": 0.4964,
      "rewards/accuracies": 0.746874988079071,
      "rewards/chosen": -1.1763864755630493,
      "rewards/margins": 1.3816043138504028,
      "rewards/rejected": -2.557990789413452,
      "step": 30
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.8856223324132555e-07,
      "logits/chosen": 0.004806989338248968,
      "logits/rejected": 1.2118862867355347,
      "logps/chosen": -357.9244384765625,
      "logps/pi_response": -218.65719604492188,
      "logps/ref_response": -178.06735229492188,
      "logps/rejected": -763.8504028320312,
      "loss": 0.4663,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.6063495874404907,
      "rewards/margins": 1.2730048894882202,
      "rewards/rejected": -1.879354476928711,
      "step": 40
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.7908455541642582e-07,
      "logits/chosen": 0.6758335828781128,
      "logits/rejected": 1.2263845205307007,
      "logps/chosen": -416.4789123535156,
      "logps/pi_response": -241.4830322265625,
      "logps/ref_response": -178.06594848632812,
      "logps/rejected": -815.0631103515625,
      "loss": 0.4644,
      "rewards/accuracies": 0.7562500238418579,
      "rewards/chosen": -0.9604195356369019,
      "rewards/margins": 1.3460404872894287,
      "rewards/rejected": -2.30646014213562,
      "step": 50
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.32661172908373e-08,
      "logits/chosen": 1.4804556369781494,
      "logits/rejected": 2.2675604820251465,
      "logps/chosen": -387.3179931640625,
      "logps/pi_response": -213.3051300048828,
      "logps/ref_response": -167.53524780273438,
      "logps/rejected": -784.0960083007812,
      "loss": 0.4624,
      "rewards/accuracies": 0.784375011920929,
      "rewards/chosen": -0.7926492691040039,
      "rewards/margins": 1.3485177755355835,
      "rewards/rejected": -2.141166925430298,
      "step": 60
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.956279997278043e-08,
      "logits/chosen": 1.3448526859283447,
      "logits/rejected": 2.2897496223449707,
      "logps/chosen": -391.35345458984375,
      "logps/pi_response": -210.98974609375,
      "logps/ref_response": -175.8311004638672,
      "logps/rejected": -786.43701171875,
      "loss": 0.4534,
      "rewards/accuracies": 0.778124988079071,
      "rewards/chosen": -0.7257815003395081,
      "rewards/margins": 1.268803358078003,
      "rewards/rejected": -1.9945847988128662,
      "step": 70
    },
    {
      "epoch": 0.99,
      "step": 79,
      "total_flos": 0.0,
      "train_loss": 0.5050203770021849,
      "train_runtime": 4555.434,
      "train_samples_per_second": 4.473,
      "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
}