Question Answering on eManual RAGBench (test)
48.11Mean Faithfulness ScoreUnWeaver
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| UnWeaverLLM=GPT-oss-120B, Embedder=Qwen3-Embedding-4B, Query Prompt=Fixed2026.02 | 48.11 | 2.47 | 98 | 940 | 198 | 96 | 105 | 5.7 | |
| VectorRAGLLM=GPT-oss-120B, Embedder=Qwen3-Embedding-4B, Query Prompt=Fixed2026.02 | 47.93 | 2.16 | — | 343 | — | 83 | 36 | 5.7 | |
| RAPTORLLM=GPT-oss-120B, Embedder=Qwen3-Embedding-4B, Query Prompt=Fixed2026.02 | 44.66 | 2.17 | 57 | 1 | 30 | 115 | 60 | 5.7 | |
| GraphRAGLLM=GPT-oss-120B, Embedder=Qwen3-Embedding-4B, Query Prompt=Fixed2026.02 | 34.24 | 5.87 | 1 | 3 | 687 | 112 | 1 | 5.7 |