Files
qwen3-8b-base-margin-dpo-hh…/train_results.json

9 lines
231 B
JSON
Raw Normal View History

{
"epoch": 0.999244142101285,
"total_flos": 0.0,
"train_loss": 1.122965409968516,
"train_runtime": 3224.9347,
"train_samples": 42336,
"train_samples_per_second": 13.128,
"train_steps_per_second": 0.205
}