nash_dpo_iter_1 / train_results.json
YYYYYYibo's picture
Model save
5a1b0f2 verified
{
"epoch": 1.0,
"train_loss": 0.6446946973984058,
"train_runtime": 7192.2445,
"train_samples": 20000,
"train_samples_per_second": 2.781,
"train_steps_per_second": 0.022
}