Question Answering on MuSiQue 1,000 questions sample (val)
32.3Containment AccuracyLinearRAG
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LinearRAGLLM=GPT-4o-mini2026.06 | 32.3 | 35.1 | |
| FlowRAGLLM=GPT-4o-mini2026.06 | 32.2 | 37.1 | |
| HippoRAG2LLM=GPT-4o-mini2026.06 | 31 | 35 | |
| GFM-RAGLLM=GPT-4o-mini2026.06 | 29.9 | 34.6 | |
| HippoRAGLLM=GPT-4o-mini2026.06 | 29.3 | 24.1 | |
| LightRAGLLM=GPT-4o-mini2026.06 | 27.4 | 28.6 | |
| Vanilla RAGLLM=GPT-4o-mini2026.06 | 26.1 | 29.6 | |
| KGPLLM=GPT-4o-mini2026.06 | 25.6 | 30.1 | |
| E2GraphRAGLLM=GPT-4o-mini2026.06 | 23.8 | 26.2 | |
| RAPTORLLM=GPT-4o-mini2026.06 | 23.3 | 27.4 | |
| G-retrieverLLM=GPT-4o-mini2026.06 | 14.4 | 15.5 |