hp_ablations_qwen_bsz512_dcftv1.2 / train_results.json
sedrickkeh's picture
End of training
da2abab verified
raw
history blame contribute delete
219 Bytes
{
"epoch": 2.9956108266276518,
"total_flos": 2144987064041472.0,
"train_loss": 0.614284630744688,
"train_runtime": 55138.6251,
"train_samples_per_second": 9.519,
"train_steps_per_second": 0.019
}