vanilla_dpo_iter_4 / train_results.json
YYYYYYibo's picture
Model save
b98266a verified
{
"epoch": 1.0,
"train_loss": 0.6200787784197391,
"train_runtime": 7189.6884,
"train_samples": 20000,
"train_samples_per_second": 2.782,
"train_steps_per_second": 0.022
}