nash_dpo_merge_iter_4 / all_results.json
YYYYYYibo's picture
Model save
aff606b verified
{
"epoch": 1.0,
"train_loss": 0.5863106256876236,
"train_runtime": 9239.1596,
"train_samples": 25000,
"train_samples_per_second": 2.706,
"train_steps_per_second": 0.021
}