Question Answering on UltraDomain Mix 1.0 (512 question-answer pairs)
54.49Exact Match (EM)FlexStructRAG
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| FlexStructRAGBackbone LLM=qwen-turbo-2025-07-15, Embedding Model=text-embedding-v3, Retrieval Mode=hybrid mode2026.02 | 54.49 | 65.21 | 73.09 | |
| HyperGraphRAGBackbone LLM=qwen-turbo-2025-07-15, Embedding Model=text-embedding-v32026.02 | 44.34 | 55.84 | 66.89 | |
| Hyper-RAGBackbone LLM=qwen-turbo-2025-07-15, Embedding Model=text-embedding-v32026.02 | 39.25 | 49.91 | 56.82 | |
| RAPTORBackbone LLM=qwen-turbo-2025-07-15, Embedding Model=text-embedding-v32026.02 | 37.5 | 49.17 | 62.75 | |
| GraphRAGBackbone LLM=qwen-turbo-2025-07-15, Embedding Model=text-embedding-v32026.02 | 36.52 | 47.52 | 60.99 | |
| StandardRAGBackbone LLM=qwen-turbo-2025-07-15, Embedding Model=text-embedding-v32026.02 | 35.16 | 47.4 | 61.85 | |
| LightRAGBackbone LLM=qwen-turbo-2025-07-15, Embedding Model=text-embedding-v32026.02 | 31.84 | 42.84 | 56.86 | |
| Cog-RAGBackbone LLM=qwen-turbo-2025-07-15, Embedding Model=text-embedding-v32026.02 | 29.88 | 40.94 | 48.92 | |
| HippoRAGBackbone LLM=qwen-turbo-2025-07-15, Embedding Model=text-embedding-v32026.02 | 28.71 | 38.02 | 52.83 | |
| NaiveGenerationBackbone LLM=qwen-turbo-2025-07-152026.02 | 10.16 | 17.48 | 36.22 |