qwen_unl_entropy_0_0 / all_results.json
yakazimir's picture
End of training
47e127f verified
raw
history blame contribute delete
785 Bytes
{
"epoch": 2.999297541394882,
"eval_logits/chosen": 0.07478896528482437,
"eval_logits/rejected": 0.14641596376895905,
"eval_logps/chosen": -1.303154706954956,
"eval_logps/rejected": -1.4992746114730835,
"eval_loss": 1.6478711366653442,
"eval_rewards/accuracies": 0.5712166428565979,
"eval_rewards/chosen": -1.303154706954956,
"eval_rewards/margins": 0.19611996412277222,
"eval_rewards/rejected": -1.4992746114730835,
"eval_runtime": 40.2789,
"eval_samples": 1345,
"eval_samples_per_second": 33.392,
"eval_steps_per_second": 8.367,
"total_flos": 0.0,
"train_loss": 1.6264377747324006,
"train_runtime": 30342.8146,
"train_samples": 59790,
"train_samples_per_second": 5.911,
"train_steps_per_second": 0.185
}