Document Retrieval on DUDE
25.17Recall@1only Segmentation
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| only SegmentationEmbedding Model=gte-base-en-v1.52026.04 | 25.17 | 30.54 | 33.22 | — | — | — | — | |
| only SegmentationEmbedding Model=e5-large-v22026.04 | 24.83 | 30.2 | 33.89 | — | — | — | — | |
| REVISE_metaEmbedding Model=gte-base-en-v1.52026.04 | 23.15 | 31.21 | 35.91 | — | — | — | — | |
| REVISE_metaEmbedding Model=bge-large-en-v1.52026.04 | 22.82 | 31.21 | 35.23 | — | — | — | — | |
| only ColumnEmbedding Model=gte-base-en-v1.52026.04 | 22.82 | 30.2 | 34.23 | — | — | — | — | |
| REVISE_metaEmbedding Model=e5-large-v22026.04 | 22.48 | 29.87 | 32.55 | — | — | — | — | |
| only ColumnEmbedding Model=bge-large-en-v1.52026.04 | 22.15 | 33.22 | 36.91 | — | — | — | — | |
| only SegmentationEmbedding Model=bge-large-en-v1.52026.04 | 22.15 | 30.87 | 36.58 | — | — | — | — | |
| only DeletionEmbedding Model=e5-large-v22026.04 | 22.15 | 29.19 | 32.89 | — | — | — | — | |
| only SubstitutionEmbedding Model=gte-base-en-v1.52026.04 | 22.15 | 30.2 | 33.56 | — | — | — | — | |
| only TranspositionEmbedding Model=gte-base-en-v1.52026.04 | 21.81 | 28.86 | 33.56 | — | — | — | — | |
| only ColumnEmbedding Model=e5-large-v22026.04 | 21.48 | 32.21 | 37.92 | — | — | — | — | |
| only SubstitutionEmbedding Model=e5-large-v22026.04 | 20.47 | 28.19 | 32.21 | — | — | — | — | |
| BaselineEmbedding Model=gte-base-en-v1.52026.04 | 20.47 | 28.86 | 31.88 | — | — | — | — | |
| only DeletionEmbedding Model=gte-base-en-v1.52026.04 | 20.47 | 29.87 | 34.23 | — | — | — | — | |
| BaselineEmbedding Model=bge-large-en-v1.52026.04 | 20.13 | 30.87 | 34.9 | — | — | — | — | |
| only SubstitutionEmbedding Model=bge-large-en-v1.52026.04 | 20.13 | 29.87 | 34.56 | — | — | — | — | |
| BaselineEmbedding Model=e5-large-v22026.04 | 20.13 | 27.18 | 31.88 | — | — | — | — | |
| only InsertionEmbedding Model=e5-large-v22026.04 | 19.8 | 28.19 | 30.54 | — | — | — | — | |
| only TranspositionEmbedding Model=e5-large-v22026.04 | 19.8 | 27.18 | 31.88 | — | — | — | — | |
| REVISE_metaEmbedding Model=jina-embeddings-v2-base2026.04 | 19.8 | 28.19 | 32.21 | — | — | — | — | |
| only DeletionEmbedding Model=bge-large-en-v1.52026.04 | 19.46 | 29.53 | 32.89 | — | — | — | — | |
| only InsertionEmbedding Model=bge-large-en-v1.52026.04 | 19.13 | 29.53 | 34.56 | — | — | — | — | |
| only SubstitutionEmbedding Model=jina-embeddings-v2-base2026.04 | 19.13 | 28.52 | 30.87 | — | — | — | — | |
| only TranspositionEmbedding Model=bge-large-en-v1.52026.04 | 18.46 | 29.87 | 34.23 | — | — | — | — | |
| only InsertionEmbedding Model=gte-base-en-v1.52026.04 | 18.46 | 29.87 | 34.23 | — | — | — | — | |
| only ColumnEmbedding Model=jina-embeddings-v2-base2026.04 | 18.12 | 27.85 | 31.54 | — | — | — | — | |
| only DeletionEmbedding Model=jina-embeddings-v2-base2026.04 | 17.79 | 25.5 | 28.86 | — | — | — | — | |
| only SegmentationEmbedding Model=jina-embeddings-v2-base2026.04 | 17.79 | 23.49 | 28.86 | — | — | — | — | |
| only TranspositionEmbedding Model=jina-embeddings-v2-base2026.04 | 17.79 | 23.83 | 28.86 | — | — | — | — | |
| BaselineEmbedding Model=jina-embeddings-v2-base2026.04 | 13.42 | 18.46 | 25.84 | — | — | — | — | |
| only InsertionEmbedding Model=jina-embeddings-v2-base2026.04 | 13.09 | 17.11 | 26.17 | — | — | — | — | |
| BaselineEmbedding Model=Aggregate (All Models)2026.04 | — | — | — | 25.34 | — | — | — | |
| Length chunkingtop-k=1 ~ 42026.04 | — | — | — | — | 26.28 | 16.86 | 21.66 | |
| Length chunkingRetrievers=Averaged over BGE, E5, BM25, MM-Embed, Top-k=macro-averaged for {1, 2, 3, 4}, RAG Protocol=Shared corpus-level2026.04 | — | — | — | — | 26.28 | 16.86 | 21.66 | |
| LumberChunkertop-k=1 ~ 42026.04 | — | — | — | — | 23.95 | 15.33 | 19.86 | |
| LumberChunkerRetrievers=Averaged over BGE, E5, BM25, MM-Embed, Top-k=macro-averaged for {1, 2, 3, 4}, RAG Protocol=Shared corpus-level2026.04 | — | — | — | — | 23.95 | 15.33 | 19.86 | |
| M3DOCDEPRetrievers=Averaged over BGE, E5, BM25, MM-Embed, Top-k=macro-averaged for {1, 2, 3, 4}, RAG Protocol=Shared corpus-level2026.04 | — | — | — | — | 35.12 | 24.91 | 27.81 | |
| MultiDocFusiontop-k=1 ~ 42026.04 | — | — | — | — | 29.27 | 20.01 | 25.05 | |
| MultiDocFusionRetrievers=Averaged over BGE, E5, BM25, MM-Embed, Top-k=macro-averaged for {1, 2, 3, 4}, RAG Protocol=Shared corpus-level2026.04 | — | — | — | — | 29.27 | 20.01 | 25.05 | |
| only ColumnEmbedding Model=Aggregate (All Models)2026.04 | — | — | — | 30.76 | — | — | — | |
| only DeletionEmbedding Model=Aggregate (All Models)2026.04 | — | — | — | 27.29 | — | — | — | |
| only InsertionEmbedding Model=Aggregate (All Models)2026.04 | — | — | — | 27.74 | — | — | — | |
| only SegmentationEmbedding Model=Aggregate (All Models)2026.04 | — | — | — | 29.87 | — | — | — | |
| only SubstitutionEmbedding Model=Aggregate (All Models)2026.04 | — | — | — | 28.19 | — | — | — | |
| only TranspositionEmbedding Model=Aggregate (All Models)2026.04 | — | — | — | 27.52 | — | — | — | |
| Perplexity chunkingtop-k=1 ~ 42026.04 | — | — | — | — | 24.28 | 15.59 | 20.2 | |
| Perplexity chunkingRetrievers=Averaged over BGE, E5, BM25, MM-Embed, Top-k=macro-averaged for {1, 2, 3, 4}, RAG Protocol=Shared corpus-level2026.04 | — | — | — | — | 24.28 | 15.59 | 20.2 | |
| REVISE_metaEmbedding Model=Aggregate (All Models)2026.04 | — | — | — | 29.75 | — | — | — | |
| Semantic chunkingtop-k=1 ~ 42026.04 | — | — | — | — | 9.56 | 5.49 | 7.75 | |
| Semantic chunkingRetrievers=Averaged over BGE, E5, BM25, MM-Embed, Top-k=macro-averaged for {1, 2, 3, 4}, RAG Protocol=Shared corpus-level2026.04 | — | — | — | — | 9.56 | 5.49 | 7.75 | |
| Structure-based chunkingtop-k=1 ~ 42026.04 | — | — | — | — | 22.19 | 14.5 | 18.62 | |
| Structure-based chunkingRetrievers=Averaged over BGE, E5, BM25, MM-Embed, Top-k=macro-averaged for {1, 2, 3, 4}, RAG Protocol=Shared corpus-level2026.04 | — | — | — | — | 22.19 | 14.5 | 18.62 |