trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9919137466307277,
  "eval_steps": 100,
  "global_step": 92,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "grad_norm": 381.81930096475645,
      "learning_rate": 5e-08,
      "logits/chosen": 110.69648742675781,
      "logits/rejected": 144.8971710205078,
      "logps/chosen": -802.6702880859375,
      "logps/rejected": -789.908203125,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.11,
      "grad_norm": 145.3169996556915,
      "learning_rate": 5e-07,
      "logits/chosen": 104.16372680664062,
      "logits/rejected": 147.7155303955078,
      "logps/chosen": -773.0805053710938,
      "logps/rejected": -786.1904907226562,
      "loss": 0.5786,
      "rewards/accuracies": 0.5416666865348816,
      "rewards/chosen": -0.04775738716125488,
      "rewards/margins": 0.49189090728759766,
      "rewards/rejected": -0.5396482944488525,
      "step": 10
    },
    {
      "epoch": 0.22,
      "grad_norm": 123.89226507692392,
      "learning_rate": 4.818756127755237e-07,
      "logits/chosen": 114.03608703613281,
      "logits/rejected": 162.7917022705078,
      "logps/chosen": -765.4005737304688,
      "logps/rejected": -855.3450927734375,
      "loss": 0.1177,
      "rewards/accuracies": 0.96875,
      "rewards/chosen": 0.9601849317550659,
      "rewards/margins": 7.298236846923828,
      "rewards/rejected": -6.3380513191223145,
      "step": 20
    },
    {
      "epoch": 0.32,
      "grad_norm": 34.79847456534606,
      "learning_rate": 4.301303984001967e-07,
      "logits/chosen": 103.0413818359375,
      "logits/rejected": 154.4637908935547,
      "logps/chosen": -750.0765380859375,
      "logps/rejected": -893.15478515625,
      "loss": 0.0543,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": 4.489648818969727,
      "rewards/margins": 12.956143379211426,
      "rewards/rejected": -8.4664945602417,
      "step": 30
    },
    {
      "epoch": 0.43,
      "grad_norm": 23.791340435812334,
      "learning_rate": 3.52267159292835e-07,
      "logits/chosen": 104.96720123291016,
      "logits/rejected": 149.83804321289062,
      "logps/chosen": -773.8189697265625,
      "logps/rejected": -879.3125,
      "loss": 0.0344,
      "rewards/accuracies": 0.981249988079071,
      "rewards/chosen": 4.7802629470825195,
      "rewards/margins": 12.39081859588623,
      "rewards/rejected": -7.610556125640869,
      "step": 40
    },
    {
      "epoch": 0.54,
      "grad_norm": 15.09725764927681,
      "learning_rate": 2.5957568342250883e-07,
      "logits/chosen": 92.78871154785156,
      "logits/rejected": 145.62005615234375,
      "logps/chosen": -734.4014892578125,
      "logps/rejected": -871.9119262695312,
      "loss": 0.0152,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 5.088149547576904,
      "rewards/margins": 11.76667594909668,
      "rewards/rejected": -6.678526401519775,
      "step": 50
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.1316228742026295,
      "learning_rate": 1.6549578039787434e-07,
      "logits/chosen": 103.61817932128906,
      "logits/rejected": 147.3481903076172,
      "logps/chosen": -762.0783081054688,
      "logps/rejected": -869.6064453125,
      "loss": 0.0127,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 4.692760944366455,
      "rewards/margins": 11.84466552734375,
      "rewards/rejected": -7.1519036293029785,
      "step": 60
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.649773974511314,
      "learning_rate": 8.366857495860869e-08,
      "logits/chosen": 100.9188232421875,
      "logits/rejected": 142.58493041992188,
      "logps/chosen": -742.1723022460938,
      "logps/rejected": -868.6980590820312,
      "loss": 0.0099,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 4.574300289154053,
      "rewards/margins": 11.782574653625488,
      "rewards/rejected": -7.208274841308594,
      "step": 70
    },
    {
      "epoch": 0.86,
      "grad_norm": 14.86224162055759,
      "learning_rate": 2.5958610759736126e-08,
      "logits/chosen": 105.79638671875,
      "logits/rejected": 157.96678161621094,
      "logps/chosen": -737.038818359375,
      "logps/rejected": -870.8370971679688,
      "loss": 0.0089,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 4.09164571762085,
      "rewards/margins": 12.486515045166016,
      "rewards/rejected": -8.394868850708008,
      "step": 80
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.750575547661943,
      "learning_rate": 7.335497040648897e-10,
      "logits/chosen": 102.36738586425781,
      "logits/rejected": 147.32630920410156,
      "logps/chosen": -761.7495727539062,
      "logps/rejected": -880.4935302734375,
      "loss": 0.009,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 4.6002397537231445,
      "rewards/margins": 12.243142127990723,
      "rewards/rejected": -7.642902374267578,
      "step": 90
    },
    {
      "epoch": 0.99,
      "step": 92,
      "total_flos": 0.0,
      "train_loss": 0.0926785598124337,
      "train_runtime": 1161.69,
      "train_samples_per_second": 10.208,
      "train_steps_per_second": 0.079
    }
  ],
  "logging_steps": 10,
  "max_steps": 92,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}