Files
Qwen2.5-3B-Instruct_Short_CoT/train_results.json

8 lines
203 B
JSON
Raw Permalink Normal View History

{
"epoch": 2.0,
"total_flos": 12178659213312.0,
"train_loss": 0.07465152751130169,
"train_runtime": 886.8941,
"train_samples_per_second": 12.017,
"train_steps_per_second": 0.753
}