qwen_fUNL_entropy / all_results.json
yakazimir's picture
End of training
038dc1b verified
raw
history blame contribute delete
785 Bytes
{
"epoch": 2.999297541394882,
"eval_logits/chosen": 7.539320468902588,
"eval_logits/rejected": 7.256679534912109,
"eval_logps/chosen": -42.77936935424805,
"eval_logps/rejected": -43.914947509765625,
"eval_loss": 1.999994623247403e-07,
"eval_rewards/accuracies": 0.5667656064033508,
"eval_rewards/chosen": -42.77936935424805,
"eval_rewards/margins": 1.1355794668197632,
"eval_rewards/rejected": -43.914947509765625,
"eval_runtime": 40.6405,
"eval_samples": 1345,
"eval_samples_per_second": 33.095,
"eval_steps_per_second": 8.292,
"total_flos": 0.0,
"train_loss": 0.010826070491851839,
"train_runtime": 30412.7115,
"train_samples": 59790,
"train_samples_per_second": 5.898,
"train_steps_per_second": 0.184
}