hBERTv2_new_pretrain_w_init__sst2 / train_results.json
gokuls's picture
End of training
90fa98c
{
"epoch": 6.0,
"train_loss": 0.1679188283767073,
"train_runtime": 2130.856,
"train_samples": 67349,
"train_samples_per_second": 1580.327,
"train_steps_per_second": 12.366
}