chinese_extract_bert_scratch / train_results.json
frett's picture
End of training
6d24e55 verified
raw
history blame contribute delete
235 Bytes
{
"epoch": 5.0,
"total_flos": 3.611513123226624e+16,
"train_loss": 4.290382243968822,
"train_runtime": 7091.0117,
"train_samples": 27643,
"train_samples_per_second": 19.492,
"train_steps_per_second": 0.609
}