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": -2.199268341064453,
      "logits/rejected": -2.1993045806884766,
      "logps/chosen": -444.4443359375,
      "logps/pi_response": -249.17221069335938,
      "logps/ref_response": -249.17221069335938,
      "logps/rejected": -656.5706787109375,
      "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": -2.2055578231811523,
      "logits/rejected": -2.1615922451019287,
      "logps/chosen": -339.12615966796875,
      "logps/pi_response": -225.56497192382812,
      "logps/ref_response": -218.07363891601562,
      "logps/rejected": -589.6128540039062,
      "loss": 0.6499,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.1427740752696991,
      "rewards/margins": 0.1424369066953659,
      "rewards/rejected": -0.2852109670639038,
      "step": 10
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.655786431300069e-07,
      "logits/chosen": -1.9026947021484375,
      "logits/rejected": -1.8411662578582764,
      "logps/chosen": -455.3936462402344,
      "logps/pi_response": -284.5272521972656,
      "logps/ref_response": -217.56393432617188,
      "logps/rejected": -904.8353271484375,
      "loss": 0.521,
      "rewards/accuracies": 0.784375011920929,
      "rewards/chosen": -1.3254053592681885,
      "rewards/margins": 1.8717145919799805,
      "rewards/rejected": -3.197119951248169,
      "step": 20
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.9061232191019517e-07,
      "logits/chosen": -2.0006089210510254,
      "logits/rejected": -1.9059054851531982,
      "logps/chosen": -461.3187561035156,
      "logps/pi_response": -265.7572326660156,
      "logps/ref_response": -216.7289581298828,
      "logps/rejected": -917.9791259765625,
      "loss": 0.5136,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": -1.244099497795105,
      "rewards/margins": 1.8518199920654297,
      "rewards/rejected": -3.095919132232666,
      "step": 30
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.8856223324132555e-07,
      "logits/chosen": -2.099766731262207,
      "logits/rejected": -2.014448642730713,
      "logps/chosen": -427.6517028808594,
      "logps/pi_response": -249.57998657226562,
      "logps/ref_response": -217.0635986328125,
      "logps/rejected": -758.6690673828125,
      "loss": 0.4599,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.7826972603797913,
      "rewards/margins": 0.9983444213867188,
      "rewards/rejected": -1.7810415029525757,
      "step": 40
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.7908455541642582e-07,
      "logits/chosen": -2.0843515396118164,
      "logits/rejected": -1.9910612106323242,
      "logps/chosen": -448.34722900390625,
      "logps/pi_response": -273.9801940917969,
      "logps/ref_response": -225.388671875,
      "logps/rejected": -805.3088989257812,
      "loss": 0.4435,
      "rewards/accuracies": 0.793749988079071,
      "rewards/chosen": -1.0244767665863037,
      "rewards/margins": 1.1706058979034424,
      "rewards/rejected": -2.195082426071167,
      "step": 50
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.32661172908373e-08,
      "logits/chosen": -2.002279043197632,
      "logits/rejected": -1.9122340679168701,
      "logps/chosen": -430.016845703125,
      "logps/pi_response": -267.3531799316406,
      "logps/ref_response": -211.29519653320312,
      "logps/rejected": -843.6468505859375,
      "loss": 0.42,
      "rewards/accuracies": 0.784375011920929,
      "rewards/chosen": -1.0640926361083984,
      "rewards/margins": 1.4147446155548096,
      "rewards/rejected": -2.478837490081787,
      "step": 60
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.956279997278043e-08,
      "logits/chosen": -1.9821714162826538,
      "logits/rejected": -1.8856136798858643,
      "logps/chosen": -447.898681640625,
      "logps/pi_response": -278.2708740234375,
      "logps/ref_response": -217.0521697998047,
      "logps/rejected": -860.4439697265625,
      "loss": 0.4232,
      "rewards/accuracies": 0.7906249761581421,
      "rewards/chosen": -1.141607403755188,
      "rewards/margins": 1.4329321384429932,
      "rewards/rejected": -2.5745396614074707,
      "step": 70
    },
    {
      "epoch": 0.99,
      "step": 79,
      "total_flos": 0.0,
      "train_loss": 0.4766373211824441,
      "train_runtime": 4474.9999,
      "train_samples_per_second": 4.554,
      "train_steps_per_second": 0.018
    }
  ],
  "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
}