hp_ablations_qwen_epoch5 / train_results.json
sedrickkeh's picture
End of training
1d08d63 verified
raw
history blame
206 Bytes
{
"epoch": 5.0,
"total_flos": 4601905523195904.0,
"train_loss": 0.5737499028512177,
"train_runtime": 30529.5931,
"train_samples_per_second": 36.787,
"train_steps_per_second": 0.072
}