hp_ablations_qwen_bsz256 / train_results.json
sedrickkeh's picture
End of training
3dfea0c verified
raw
history blame
220 Bytes
{
"epoch": 2.9982905982905983,
"total_flos": 2758364765356032.0,
"train_loss": 0.5923774672614808,
"train_runtime": 70850.8498,
"train_samples_per_second": 9.511,
"train_steps_per_second": 0.037
}