cooking_sft_fail_new_mem / train_results.json
izzcw's picture
End of training
e268c95 verified
{
"epoch": 0.999983338054218,
"total_flos": 1934604910592000.0,
"train_loss": 0.2527618171993302,
"train_runtime": 72440.3327,
"train_samples_per_second": 6.628,
"train_steps_per_second": 0.052
}