NP-hard Graph Reasoning on Small-scale NP-hard graph problems Average
86.7AccuracyNPG-Muse-7B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| NPG-Muse-7BCategory=NPG-Muse2025.08 | 86.7 | 98.1 | |
| QwQ-32BCategory=Reasoning2025.08 | 85.3 | 98.7 | |
| NPG-Muse-8BCategory=NPG-Muse2025.08 | 79.6 | 98.7 | |
| LLaMA-8B-GTCategory=Graph-oriented2025.08 | 60.1 | — | |
| Claude-3.5-sonnetCategory=Closed-source2025.08 | 48.3 | 94.1 | |
| GPT-4oCategory=Closed-source2025.08 | 46.4 | 92.7 | |
| G1-7BCategory=Graph-oriented2025.08 | 40 | — | |
| S1.1-7BCategory=Reasoning2025.08 | 34.3 | 72.6 | |
| LLaMA3-70B-InsCategory=Non-reasoning2025.08 | 32.5 | 80.3 | |
| Qwen2.5-7B-Ins-1MCategory=Non-reasoning2025.08 | 25.7 | 65.2 | |
| Qwen3-8B-BaseCategory=Non-reasoning2025.08 | 23.5 | 61.1 | |
| GraphWiz-7B-DPOCategory=Graph-oriented2025.08 | 1.5 | 19.2 |