Files
Qwen3-4B-Instruct-2507-SFT-tr5/train_results.json

9 lines
246 B
JSON
Raw Permalink Normal View History

{
"epoch": 4.6268656716417915,
"total_flos": 829937030004736.0,
"train_loss": 0.4202386662844689,
"train_runtime": 18585.0074,
"train_samples": 4281,
"train_samples_per_second": 1.068,
"train_steps_per_second": 0.033
}