Files
Qwen3-4B_Paper_Impact_code_…/train_results.json

8 lines
209 B
JSON
Raw Permalink Normal View History

{
"epoch": 1.0,
"total_flos": 2.1110553137958093e+17,
"train_loss": 0.10852315148402905,
"train_runtime": 1273.7115,
"train_samples_per_second": 2.893,
"train_steps_per_second": 0.046
}