hp_ablations_qwen_epoch2_dcftv1.2 / train_results.json
sedrickkeh's picture
End of training
d688721 verified
{
"epoch": 1.9967081199707388,
"total_flos": 1429904000286720.0,
"train_loss": 0.6341017776570362,
"train_runtime": 36533.8026,
"train_samples_per_second": 9.578,
"train_steps_per_second": 0.019
}