hp_ablations_qwen_adambeta2_0.995 / train_results.json
sedrickkeh's picture
End of training
b986ae6 verified
raw
history blame
219 Bytes
{
"epoch": 2.996011396011396,
"total_flos": 2755219238682624.0,
"train_loss": 0.6040605018127999,
"train_runtime": 70543.6764,
"train_samples_per_second": 9.552,
"train_steps_per_second": 0.019
}