llama-3-8b-instruct-sppo-iter3 / all_results.json
jcmei's picture
End of training
ec3ffcf verified
{
"epoch": 0.9984,
"total_flos": 0.0,
"train_loss": 125356.69771634616,
"train_runtime": 2759.785,
"train_samples": 19996,
"train_samples_per_second": 7.245,
"train_steps_per_second": 0.113
}