Files
DReP-SFT-Qwen3-8B/train_results.json

8 lines
206 B
JSON
Raw Permalink Normal View History

{
"epoch": 4.0,
"total_flos": 1671471016509440.0,
"train_loss": 0.5990438958973561,
"train_runtime": 156717.3164,
"train_samples_per_second": 0.671,
"train_steps_per_second": 0.005
}