nash_dpo_rank4_iter_real_plus_3 / train_results.json
YYYYYYibo's picture
Model save
d8c02b1 verified
{
"epoch": 1.0,
"train_loss": 0.5628757874170939,
"train_runtime": 1697.0362,
"train_samples": 5000,
"train_samples_per_second": 2.946,
"train_steps_per_second": 0.023
}