GPT2_BABYLM_20000 / train_results.json
xiulinyang's picture
Add checkpoint
0490ec1
{
"epoch": 10.0,
"total_flos": 8.6487662592e+16,
"train_loss": 0.990625182810613,
"train_runtime": 16693.5239,
"train_samples": 33100,
"train_samples_per_second": 19.828,
"train_steps_per_second": 0.62
}