Retrieval on MisstepMath
67.49Cosine SimilarityMisEdu-RAG
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MisEdu-RAGBackbone=Qwen-Plus2026.04 | 67.49 | 39.65 | |
| MisEdu-RAGBackbone=GPT-4o-mini2026.04 | 64.83 | 34.37 | |
| HypergraphRAGBackbone=Qwen-Plus2026.04 | 62.46 | 28.7 | |
| HypergraphRAGBackbone=GPT-4o-mini2026.04 | 62.02 | 28.01 | |
| StandardRAGBackbone=GPT-4o-mini2026.04 | 61.36 | 23.98 | |
| MisEdu-RAGBackbone=DeepSeek-R12026.04 | 59.47 | 25.48 | |
| StandardRAGBackbone=Qwen-Plus2026.04 | 58.05 | 19.22 | |
| LLM generationBackbone=GPT-4o-mini2026.04 | 57.7 | 16.55 | |
| MisEdu-RAGBackbone=LLaMa3.3-70B2026.04 | 56.33 | 23.17 | |
| LLM generationBackbone=Qwen-Plus2026.04 | 56.12 | 18.37 | |
| HypergraphRAGBackbone=DeepSeek-R12026.04 | 53.77 | 17.91 | |
| HypergraphRAGBackbone=LLaMa3.3-70B2026.04 | 50.26 | 16.2 | |
| StandardRAGBackbone=DeepSeek-R12026.04 | 48.36 | 13.82 | |
| StandardRAGBackbone=LLaMa3.3-70B2026.04 | 47.8 | 17.33 | |
| LLM generationBackbone=DeepSeek-R12026.04 | 47.66 | 13.23 | |
| LLM generationBackbone=LLaMa3.3-70B2026.04 | 43.1 | 12.74 |