Llama-3.1-8B-Instruct-SFT-500 / all_results.json
chchen's picture
End of training
97d9fef verified
raw
history blame contribute delete
360 Bytes
{
"epoch": 9.955555555555556,
"eval_loss": 0.07805962860584259,
"eval_runtime": 1.3061,
"eval_samples_per_second": 38.281,
"eval_steps_per_second": 19.141,
"total_flos": 2.516129863512883e+16,
"train_loss": 0.3581957229546138,
"train_runtime": 398.237,
"train_samples_per_second": 11.3,
"train_steps_per_second": 0.703
}