trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.985781990521327,
  "eval_steps": 100,
  "global_step": 52,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "grad_norm": 1597.1610924726594,
      "learning_rate": 8.333333333333333e-08,
      "logits/chosen": 123.11854553222656,
      "logits/rejected": 97.00198364257812,
      "logps/chosen": -425.18585205078125,
      "logps/rejected": -424.1869201660156,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.19,
      "grad_norm": 1960.9556077938475,
      "learning_rate": 4.907293218369498e-07,
      "logits/chosen": 117.075439453125,
      "logits/rejected": 135.90403747558594,
      "logps/chosen": -441.22174072265625,
      "logps/rejected": -524.5287475585938,
      "loss": 2.1431,
      "rewards/accuracies": 0.5486111044883728,
      "rewards/chosen": 1.300018310546875,
      "rewards/margins": 1.5483126640319824,
      "rewards/rejected": -0.24829429388046265,
      "step": 10
    },
    {
      "epoch": 0.38,
      "grad_norm": 1386.5508807535327,
      "learning_rate": 3.941700805287168e-07,
      "logits/chosen": 126.82967376708984,
      "logits/rejected": 134.35858154296875,
      "logps/chosen": -432.8621520996094,
      "logps/rejected": -501.55560302734375,
      "loss": 2.1158,
      "rewards/accuracies": 0.637499988079071,
      "rewards/chosen": -3.958549976348877,
      "rewards/margins": 4.468470096588135,
      "rewards/rejected": -8.427020072937012,
      "step": 20
    },
    {
      "epoch": 0.57,
      "grad_norm": 1222.8384880370088,
      "learning_rate": 2.3293939665883228e-07,
      "logits/chosen": 127.63995361328125,
      "logits/rejected": 130.95755004882812,
      "logps/chosen": -463.97509765625,
      "logps/rejected": -524.4500122070312,
      "loss": 2.103,
      "rewards/accuracies": 0.668749988079071,
      "rewards/chosen": -5.313003063201904,
      "rewards/margins": 5.028109073638916,
      "rewards/rejected": -10.34111213684082,
      "step": 30
    },
    {
      "epoch": 0.76,
      "grad_norm": 1242.2684060861586,
      "learning_rate": 7.936171419533652e-08,
      "logits/chosen": 127.74715423583984,
      "logits/rejected": 122.52357482910156,
      "logps/chosen": -481.22698974609375,
      "logps/rejected": -501.05059814453125,
      "loss": 2.0594,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -4.827347755432129,
      "rewards/margins": 4.944188117980957,
      "rewards/rejected": -9.771535873413086,
      "step": 40
    },
    {
      "epoch": 0.95,
      "grad_norm": 1482.0031541291255,
      "learning_rate": 2.328513490917311e-09,
      "logits/chosen": 136.43093872070312,
      "logits/rejected": 136.77505493164062,
      "logps/chosen": -492.9546813964844,
      "logps/rejected": -509.92645263671875,
      "loss": 2.0029,
      "rewards/accuracies": 0.6312500238418579,
      "rewards/chosen": -6.079165458679199,
      "rewards/margins": 5.127718448638916,
      "rewards/rejected": -11.206883430480957,
      "step": 50
    },
    {
      "epoch": 0.99,
      "step": 52,
      "total_flos": 0.0,
      "train_loss": 2.0260768074255724,
      "train_runtime": 571.9301,
      "train_samples_per_second": 11.802,
      "train_steps_per_second": 0.091
    }
  ],
  "logging_steps": 10,
  "max_steps": 52,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}