Files
qwen3-4B-instruct-refiner-sft/all_results.json

12 lines
350 B
JSON
Raw Normal View History

{
"epoch": 5.0,
"eval_loss": 1.1232492923736572,
"eval_runtime": 111.5192,
"eval_samples_per_second": 4.484,
"eval_steps_per_second": 2.242,
"total_flos": 3.0080813400754176e+18,
"train_loss": 0.12334943315905438,
"train_runtime": 40705.595,
"train_samples_per_second": 2.097,
"train_steps_per_second": 0.066
}