zephyr-7b-dpo-qlora / all_results.json
ketchup123's picture
End of training
4a1941e verified
raw
history blame contribute delete
772 Bytes
{
"epoch": 0.9984301412872841,
"eval_logits/chosen": -1.2525343894958496,
"eval_logits/rejected": -1.1810396909713745,
"eval_logps/chosen": -443.7073059082031,
"eval_logps/rejected": -543.7128295898438,
"eval_loss": 0.5031002759933472,
"eval_rewards/accuracies": 0.76953125,
"eval_rewards/chosen": -1.9727791547775269,
"eval_rewards/margins": 0.9890241622924805,
"eval_rewards/rejected": -2.961803436279297,
"eval_runtime": 110.64,
"eval_samples": 2000,
"eval_samples_per_second": 18.077,
"eval_steps_per_second": 0.289,
"total_flos": 0.0,
"train_loss": 0.5421361258444796,
"train_runtime": 7660.497,
"train_samples": 61134,
"train_samples_per_second": 7.98,
"train_steps_per_second": 0.062
}