Mathematical Reasoning on NuminaMath (Accuracy)
78.67AccuracyQwen2.5-Math-7B-Instruct
Evaluation results
| Method | Links | |
|---|---|---|
| Qwen2.5-Math-7B-Instructhint_setting=w/ LLM Hints2026.04 | 78.67 | |
| DeepSeek-R1-Distill-Qwen-7Bhint_setting=w/ LLM Hints2026.04 | 69.33 | |
| DeepSeek-R1-Distill-Llama-8Bhint_setting=w/ LLM Hints2026.04 | 68 | |
| FT DeepSeek-R1-Distill-Qwen-7Bhint_setting=w/ LLM Hints2026.04 | 65.33 | |
| DeepSeek-R1-Distill-Qwen-7Bhint_setting=w/ SLM Hints2026.04 | 60 | |
| Qwen2.5-Math-7B-Instructhint_setting=w/ SLM Hints2026.04 | 58.67 | |
| Qwen2.5-Math-7B-Instructhint_setting=w/o Hints2026.04 | 46.67 | |
| DeepSeek-R1-Distill-Llama-8Bhint_setting=w/ SLM Hints2026.04 | 45.33 | |
| FT DeepSeek-R1-Distill-Qwen-7Bhint_setting=w/ NFT SLM Hints2026.04 | 45.33 | |
| DeepSeek-R1-Distill-Qwen-7Bhint_setting=w/o Hints2026.04 | 44 | |
| DeepSeek-R1-Distill-Qwen-7Bhint_setting=w/ NFT SLM Hints2026.04 | 42.67 | |
| DeepSeek-R1-Distill-Llama-8Bhint_setting=w/o Hints2026.04 | 42.67 | |
| FT DeepSeek-R1-Distill-Qwen-7Bhint_setting=w/o Hints2026.04 | 38.67 | |
| FT DeepSeek-R1-Distill-Qwen-7Bhint_setting=w/ SLM Hints2026.04 | 37.33 | |
| Qwen2.5-Math-7B-Instructhint_setting=w/ NFT SLM Hints2026.04 | 36 | |
| Phi-4-mini-reasoninghint_setting=w/ LLM Hints2026.04 | 33.33 | |
| Phi-4-mini-reasoninghint_setting=w/ SLM Hints2026.04 | 29.33 | |
| DeepSeek-R1-Distill-Llama-8Bhint_setting=w/ NFT SLM Hints2026.04 | 26.67 | |
| Phi-4-mini-reasoninghint_setting=w/o Hints2026.04 | 21.33 | |
| Phi-4-mini-reasoninghint_setting=w/ NFT SLM Hints2026.04 | 21.33 |