hogru's picture
Update tokenizer, bump hf versions
ce10699
raw
history blame
195 Bytes
{
"epoch": 31.0,
"train_loss": 0.7912082670869731,
"train_runtime": 664.7096,
"train_samples": 7878,
"train_samples_per_second": 592.59,
"train_steps_per_second": 18.504
}