opt-babylm2-subset-default-3e-4 / train_results.json
kanishka's picture
End of training
1c2c453 verified
raw
history blame contribute delete
237 Bytes
{
"epoch": 10.0,
"total_flos": 5.9695438005504e+17,
"train_loss": 2.237272139724321,
"train_runtime": 31036.7572,
"train_samples": 453383,
"train_samples_per_second": 146.079,
"train_steps_per_second": 4.565
}