6.7b-ri-reproduce-4-gpu / train_results.json
AlekseyKorshuk's picture
End of training
1f2302f
{
"epoch": 10.0,
"train_loss": 1.1160404185454051,
"train_runtime": 51742.5542,
"train_samples": 189,
"train_samples_per_second": 0.037,
"train_steps_per_second": 0.009
}