qwen2.5-0.5b-expo-L2EXPO-ES-1 / all_results.json
hZzy's picture
End of training
7578e78 verified
{
"epoch": 1.8422295701464337,
"eval_dpo_loss": 2.573197364807129,
"eval_logits": -0.484580934047699,
"eval_logps": -80.18506622314453,
"eval_loss": 4.899093151092529,
"eval_objective": 4.920805931091309,
"eval_ranking_idealized": 0.5295031070709229,
"eval_ranking_idealized_expo": 0.5212215185165405,
"eval_ranking_simple": 0.523809552192688,
"eval_regularize": 4.920805931091309,
"eval_runtime": 307.8846,
"eval_samples": 5790,
"eval_samples_per_second": 18.806,
"eval_steps_per_second": 1.569,
"eval_wo_beta": 14.127137184143066,
"total_flos": 0.0,
"train_loss": 2.767508169504312,
"train_runtime": 17905.6458,
"train_samples": 50802,
"train_samples_per_second": 14.186,
"train_steps_per_second": 0.098
}