pretrain / eval_results.json
furmaniak's picture
End of training
417153d verified
raw
history blame contribute delete
213 Bytes
{
"epoch": 0.9454545454545454,
"eval_loss": 1.1903643608093262,
"eval_runtime": 40.1525,
"eval_samples_per_second": 2.441,
"eval_steps_per_second": 0.324,
"perplexity": 3.2882791091731414
}