tinyllama_moe_sft_routeraux_ep3 / train_results.json
hushell's picture
Model save
eb5a167 verified
{
"epoch": 3.0,
"train_loss": 1.4016100346188847,
"train_runtime": 49299.9861,
"train_samples": 207865,
"train_samples_per_second": 8.888,
"train_steps_per_second": 0.069
}