deepspeed_no_offload_liger / train_results.json
sedrickkeh's picture
End of training
71dd683 verified
raw
history blame
221 Bytes
{
"epoch": 2.980891719745223,
"total_flos": 6.084353864735949e+16,
"train_loss": 0.7682466140160193,
"train_runtime": 1871.3915,
"train_samples_per_second": 8.017,
"train_steps_per_second": 0.083
}