Qwen3-0.6B-sft / train_results.json
wassname's picture
End of training
7150f1e verified
{
"epoch": 3.0,
"total_flos": 1.0094624425786737e+18,
"train_loss": 1.2982249011391638,
"train_runtime": 26293.112,
"train_samples": 117772,
"train_samples_per_second": 7.093,
"train_steps_per_second": 0.028
}