metadata
license: apache-2.0
datasets:
- qingy2024/OCTAL-Math-CoT-47k
base_model:
- Qwen/Qwen2.5-Math-7B-Instruct
Experimental fine-tune of Qwen2.5-Math 7B using OCTAL-Math-CoT-47k, which took about 2-3 hours with MLX fine-tuning (on a single M2 Ultra).