qwen_l21_entropy / all_results.json
yakazimir's picture
End of training
7b7fa53 verified
raw
history blame contribute delete
779 Bytes
{
"epoch": 2.999297541394882,
"eval_logits/chosen": 0.9296241998672485,
"eval_logits/rejected": 1.3373150825500488,
"eval_logps/chosen": -4.961300849914551,
"eval_logps/rejected": -8.357980728149414,
"eval_loss": 0.6612383723258972,
"eval_rewards/accuracies": 0.6765578389167786,
"eval_rewards/chosen": -4.961300849914551,
"eval_rewards/margins": 3.396681308746338,
"eval_rewards/rejected": -8.357980728149414,
"eval_runtime": 40.3427,
"eval_samples": 1345,
"eval_samples_per_second": 33.339,
"eval_steps_per_second": 8.353,
"total_flos": 0.0,
"train_loss": 0.6706608949092182,
"train_runtime": 30654.4044,
"train_samples": 59790,
"train_samples_per_second": 5.851,
"train_steps_per_second": 0.183
}