nash_dpo_iter_2 / all_results.json
YYYYYYibo's picture
Model save
6389088
{
"epoch": 1.0,
"train_loss": 0.6604029192374303,
"train_runtime": 7258.1064,
"train_samples": 20000,
"train_samples_per_second": 2.756,
"train_steps_per_second": 0.021
}