deepspeed_no_offload / all_results.json
sedrickkeh's picture
End of training
e57a620 verified
{
"epoch": 2.980891719745223,
"total_flos": 55447011295232.0,
"train_loss": 0.7628553406550334,
"train_runtime": 1911.8295,
"train_samples_per_second": 7.847,
"train_steps_per_second": 0.082
}