Question Answering on STAGE Main Results
70.12Overall ScoreNKW
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| NKWBackbone=Qwen3-235B2026.06 | 70.12 | 81.48 | |
| NKWBackbone=GPT-5.52026.06 | 69.84 | 81.14 | |
| NKWBackbone=Llama-3.1-405B2026.06 | 65.35 | 74.29 | |
| Hybrid RAGBackbone=GPT-5.52026.06 | 62.16 | 65.25 | |
| Hybrid RAGBackbone=Llama-3.1-405B2026.06 | 61.12 | 63.85 | |
| Hybrid RAGBackbone=Qwen3-235B2026.06 | 60.56 | 64.65 | |
| NKWBackbone=Qwen3-30B2026.06 | 58.92 | 66.43 | |
| NKWBackbone=Llama-3.1-70B2026.06 | 56.29 | 64.31 | |
| Hybrid RAGBackbone=Qwen3-30B2026.06 | 51.44 | 55.63 | |
| Hybrid RAGBackbone=Llama-3.1-70B2026.06 | 51.24 | 55.47 | |
| HippoRAGBackbone=GPT-5.52026.06 | 51.18 | 54.29 | |
| HippoRAGBackbone=Llama-3.1-405B2026.06 | 48.74 | 51.12 | |
| LightRAGBackbone=GPT-5.52026.06 | 47.84 | 61.22 | |
| LightRAGBackbone=Llama-3.1-405B2026.06 | 45.27 | 58.94 | |
| LightRAGBackbone=Qwen3-235B2026.06 | 44.07 | 59.88 | |
| Hybrid RAGBackbone=Qwen3-8B2026.06 | 42.16 | 46.37 | |
| A-RAGBackbone=GPT-5.52026.06 | 42.16 | 53.25 | |
| HippoRAGBackbone=Llama-3.1-70B2026.06 | 41.02 | 44.15 | |
| GraphRAGBackbone=GPT-5.52026.06 | 40.24 | 53.57 | |
| HippoRAGBackbone=Qwen3-235B2026.06 | 40.14 | 43.01 | |
| A-RAGBackbone=Llama-3.1-405B2026.06 | 39.52 | 50.2 | |
| LightRAGBackbone=Llama-3.1-70B2026.06 | 39.12 | 52.16 | |
| LightRAGBackbone=Qwen3-30B2026.06 | 38.74 | 53.93 | |
| Hybrid RAGBackbone=Llama-3.1-8B2026.06 | 38.54 | 41.24 | |
| NKWBackbone=Qwen3-8B2026.06 | 38.52 | 50.14 | |
| NKWBackbone=Llama-3.1-8B2026.06 | 37.21 | 48.54 | |
| LightRAGBackbone=Qwen3-8B2026.06 | 36.85 | 43.29 | |
| A-RAGBackbone=Qwen3-235B2026.06 | 36.25 | 48.62 | |
| GraphRAGBackbone=Llama-3.1-405B2026.06 | 35.25 | 49.82 | |
| LightRAGBackbone=Llama-3.1-8B2026.06 | 34.19 | 41.06 | |
| A-RAGBackbone=Llama-3.1-70B2026.06 | 33.49 | 42.14 | |
| GraphRAGBackbone=Qwen3-235B2026.06 | 32.06 | 47.9 | |
| HippoRAGBackbone=Qwen3-30B2026.06 | 31.86 | 34.41 | |
| GraphRAGBackbone=Llama-3.1-70B2026.06 | 31.22 | 46.91 | |
| GraphRAGBackbone=Qwen3-30B2026.06 | 30.28 | 41.26 | |
| A-RAGBackbone=Qwen3-30B2026.06 | 29.8 | 37.56 | |
| GraphRAGBackbone=Llama-3.1-8B2026.06 | 29.16 | 32.46 | |
| GraphRAGBackbone=Qwen3-8B2026.06 | 28.54 | 31.24 | |
| HippoRAGBackbone=Llama-3.1-8B2026.06 | 27.19 | 29.32 | |
| HippoRAGBackbone=Qwen3-8B2026.06 | 26.43 | 28.9 | |
| A-RAGBackbone=Qwen3-8B2026.06 | 23.15 | 27.41 | |
| A-RAGBackbone=Llama-3.1-8B2026.06 | 21.46 | 25.29 |