HBERTv1_L12_H768_A12_ffn_3 / train_results.json
gokuls's picture
End of training
54f2056
raw
history blame contribute delete
202 Bytes
{
"epoch": 1.94,
"train_loss": 5.037873150938256,
"train_runtime": 197999.0051,
"train_samples": 5858758,
"train_samples_per_second": 2958.984,
"train_steps_per_second": 61.646
}