Question Answering on HotpotQA (EM and Latency)
48.1EM (%)LatentRAG△
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LatentRAG△Retrieval Model=jina-embeddings-v5-text-nano2026.05 | 48.1 | 467 | |
| LatentRAG♢Retrieval Model=jina-embeddings-v5-text-nano2026.05 | 47.66 | 540 | |
| LatentRAG♢Retrieval Model=harrier-oss-v1-270m2026.05 | 47.27 | 614 | |
| LatentRAG△Retrieval Model=Qwen3-Embedding-0.6B2026.05 | 47.16 | 528 | |
| LatentRAG△Retrieval Model=harrier-oss-v1-270m2026.05 | 46.74 | 507 | |
| LatentRAG♢Retrieval Model=Qwen3-Embedding-0.6B2026.05 | 46.73 | 626 | |
| LatentRAG♢Retrieval Model=e5-base-v22026.05 | 46.41 | 588 | |
| LatentRAG△Retrieval Model=e5-base-v22026.05 | 46.27 | 464 | |
| AutoRefine△Retrieval Model=harrier-oss-v1-270m2026.05 | 45.4 | 4,703 | |
| AutoRefine△Retrieval Model=e5-base-v22026.05 | 45.13 | 5,133 | |
| Search-R1♢Retrieval Model=harrier-oss-v1-270m2026.05 | 44.88 | 6,270 | |
| AutoRefine△Retrieval Model=jina-embeddings-v5-text-nano2026.05 | 44.79 | 4,962 | |
| LatentRAG△Retrieval Model=F2LLM-v2-330M2026.05 | 44.74 | 499 | |
| Search-R1♢Retrieval Model=e5-base-v22026.05 | 44.71 | 6,263 | |
| LatentRAG♢Retrieval Model=F2LLM-v2-330M2026.05 | 44.58 | 593 | |
| Search-R1♢Retrieval Model=jina-embeddings-v5-text-nano2026.05 | 44.31 | 6,161 | |
| AutoRefine△Retrieval Model=Qwen3-Embedding-0.6B2026.05 | 44.06 | 5,344 | |
| R1-SearcherRetrieval Model=Qwen3-Embedding-0.6B2026.05 | 43.34 | 9,306 | |
| Search-R1♢Retrieval Model=Qwen3-Embedding-0.6B2026.05 | 43.05 | 6,589 | |
| AutoRefine△Retrieval Model=F2LLM-v2-330M2026.05 | 42.63 | 4,992 | |
| Search-R1♢Retrieval Model=F2LLM-v2-330M2026.05 | 41.99 | 6,228 | |
| DeepRAGRetrieval Model=Qwen3-Embedding-0.6B2026.05 | 33.44 | 3,232 | |
| Iter-RetGenRetrieval Model=Qwen3-Embedding-0.6B2026.05 | 31.74 | 1,303 | |
| ZeroSearchRetrieval Model=Qwen3-Embedding-0.6B2026.05 | 30.61 | 4,136 | |
| Naive RAGRetrieval Model=Qwen3-Embedding-0.6B2026.05 | 27.41 | 300 | |
| Search-o1Retrieval Model=Qwen3-Embedding-0.6B2026.05 | 25.09 | 4,886 | |
| Direct InferRetrieval Model=N/A2026.05 | 16.68 | 179 |