trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 100,
  "global_step": 78,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "grad_norm": 5.038770139725663,
      "learning_rate": 6.25e-07,
      "logits/chosen": -1.3596761226654053,
      "logits/rejected": -1.0023326873779297,
      "logps/chosen": -450.79583740234375,
      "logps/rejected": -781.127197265625,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.222015988708302,
      "learning_rate": 4.989935734988098e-06,
      "logits/chosen": -1.0887341499328613,
      "logits/rejected": -1.0588740110397339,
      "logps/chosen": -564.32080078125,
      "logps/rejected": -855.5671997070312,
      "loss": 0.6684,
      "rewards/accuracies": 0.6944444179534912,
      "rewards/chosen": -0.007345028221607208,
      "rewards/margins": 0.07519946992397308,
      "rewards/rejected": -0.08254450559616089,
      "step": 10
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.800980531623392,
      "learning_rate": 4.646121984004666e-06,
      "logits/chosen": -1.4986072778701782,
      "logits/rejected": -2.113752603530884,
      "logps/chosen": -606.9659423828125,
      "logps/rejected": -1018.4754638671875,
      "loss": 0.4124,
      "rewards/accuracies": 0.8374999761581421,
      "rewards/chosen": -0.3804299235343933,
      "rewards/margins": 1.465038537979126,
      "rewards/rejected": -1.845468521118164,
      "step": 20
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.950457492388352,
      "learning_rate": 3.8772424536302565e-06,
      "logits/chosen": -1.7479159832000732,
      "logits/rejected": -2.567039966583252,
      "logps/chosen": -614.0457153320312,
      "logps/rejected": -1108.7679443359375,
      "loss": 0.3199,
      "rewards/accuracies": 0.893750011920929,
      "rewards/chosen": -0.10914883762598038,
      "rewards/margins": 2.6420645713806152,
      "rewards/rejected": -2.751213550567627,
      "step": 30
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.011791311794506,
      "learning_rate": 2.835583164544139e-06,
      "logits/chosen": -1.7483808994293213,
      "logits/rejected": -2.6819939613342285,
      "logps/chosen": -644.5540161132812,
      "logps/rejected": -1273.226318359375,
      "loss": 0.2279,
      "rewards/accuracies": 0.9125000238418579,
      "rewards/chosen": 0.04285002499818802,
      "rewards/margins": 4.04502010345459,
      "rewards/rejected": -4.002170085906982,
      "step": 40
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.2682180093043582,
      "learning_rate": 1.7274575140626318e-06,
      "logits/chosen": -1.5594079494476318,
      "logits/rejected": -2.4641501903533936,
      "logps/chosen": -627.8310546875,
      "logps/rejected": -1211.9755859375,
      "loss": 0.2181,
      "rewards/accuracies": 0.9375,
      "rewards/chosen": 0.09700597822666168,
      "rewards/margins": 3.8286356925964355,
      "rewards/rejected": -3.7316298484802246,
      "step": 50
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.445366662132522,
      "learning_rate": 7.723433775328385e-07,
      "logits/chosen": -1.6432807445526123,
      "logits/rejected": -2.286984920501709,
      "logps/chosen": -544.6373291015625,
      "logps/rejected": -1150.681396484375,
      "loss": 0.1556,
      "rewards/accuracies": 0.956250011920929,
      "rewards/chosen": 0.27296125888824463,
      "rewards/margins": 3.291428327560425,
      "rewards/rejected": -3.0184669494628906,
      "step": 60
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.67000331566183,
      "learning_rate": 1.59412823400657e-07,
      "logits/chosen": -1.5773608684539795,
      "logits/rejected": -2.4722862243652344,
      "logps/chosen": -636.8548583984375,
      "logps/rejected": -1234.2066650390625,
      "loss": 0.1503,
      "rewards/accuracies": 0.956250011920929,
      "rewards/chosen": 0.26559901237487793,
      "rewards/margins": 4.093817234039307,
      "rewards/rejected": -3.828218460083008,
      "step": 70
    },
    {
      "epoch": 1.0,
      "step": 78,
      "total_flos": 0.0,
      "train_loss": 0.2915988182410216,
      "train_runtime": 1026.489,
      "train_samples_per_second": 4.863,
      "train_steps_per_second": 0.076
    }
  ],
  "logging_steps": 10,
  "max_steps": 78,
  "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
}