Llama-3.1-8B-Instruct-SFT-Half / train_results.json
chchen's picture
End of training
7612cba verified
raw
history blame
217 Bytes
{
"epoch": 2.986666666666667,
"total_flos": 7565873752571904.0,
"train_loss": 0.9732896827516102,
"train_runtime": 178.1411,
"train_samples_per_second": 7.578,
"train_steps_per_second": 0.472
}