Files
Qwen3-4B-Base-SFT-tr5/train_results.json

9 lines
247 B
JSON
Raw Permalink Normal View History

{
"epoch": 4.6268656716417915,
"total_flos": 829202911068160.0,
"train_loss": 0.33651591361530364,
"train_runtime": 10914.7384,
"train_samples": 4281,
"train_samples_per_second": 1.818,
"train_steps_per_second": 0.057
}