Visual document retrieval on ViDoRe V2
66.9Avg nDCG@5Qwen3-VL-Emb-2B
Evaluation Results
| Method | Links | ||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Qwen3-VL-Emb-2BSize=2B, Late Interaction=true2026.05 | 66.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| NemColEmbed V2 8BSize=8B, Late Interaction=true, Neural Query Encoder=true2026.05 | 65.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Evo-Retriever-7BSize (B)=7.02026.03 | 65.2 | — | — | — | — | — | 65.2 | 59.1 | 77 | — | — | 59.7 | — | — | — | — | — | — | |
| llama-nemoretriever-3b-v1Size (B)=4.42026.03 | 63.5 | — | — | — | — | — | 64.3 | 55.9 | 75.4 | — | — | 58.6 | — | — | — | — | — | — | |
| Evo-Retriever-3BSize (B)=3.02026.03 | 63.3 | — | — | — | — | — | 63.3 | 59.6 | 73 | — | — | 57.3 | — | — | — | — | — | — | |
| GME-Qwen2-7BSize=7B, Late Interaction=true2026.05 | 63.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| llama-nemoretriever-1b-v1Size (B)=2.42026.03 | 62.1 | — | — | — | — | — | 62.9 | 53.2 | 76.4 | — | — | 55.9 | — | — | — | — | — | — | |
| colnomic-embed-multimodal-7bSize (B)=7.02026.03 | 60.8 | — | — | — | — | — | 63.4 | 56.2 | 68.7 | — | — | 54.8 | — | — | — | — | — | — | |
| BaseTok=12972026.02 | 60 | — | — | — | — | — | 61.4 | 53.9 | 57 | 67.6 | — | — | — | — | — | — | — | — | |
| colqwen2.5-v0.2Size (B)=3.02026.03 | 59.3 | — | — | — | — | — | 59.2 | 53.3 | 66.4 | — | — | 58.3 | — | — | — | — | — | — | |
| MEmbedTok=642026.02 | 58.8 | — | — | — | — | — | 58.7 | 55.5 | 57.4 | 63.7 | — | — | — | — | — | — | — | — | |
| jina-embeddings-v4Size (B)=3.82026.03 | 57.6 | — | — | — | — | — | 60.9 | 51.9 | 65.1 | — | — | 52.5 | — | — | — | — | — | — | |
| AGCTok=642026.02 | 56.7 | — | — | — | — | — | 59 | 54.5 | 55.8 | 57.3 | — | — | — | — | — | — | — | — | |
| CQOTok=-2026.02 | 56.5 | — | — | — | — | — | 56.5 | 53.2 | 54.2 | 62.2 | — | — | — | — | — | — | — | — | |
| H-PoolTok=642026.02 | 56.4 | — | — | — | — | — | 59.6 | 52.1 | 53.4 | 60.6 | — | — | — | — | — | — | — | — | |
| ColModernVBERTSize=0.25B, Late Interaction=true, Neural Query Encoder=true2026.05 | 56 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| colnomic-embed-multimodal-3bSize (B)=3.02026.03 | 55.5 | — | — | — | — | — | 62.5 | 53.3 | 57 | — | — | 49.2 | — | — | — | — | — | — | |
| colqwen2-v1.0Size (B)=2.22026.03 | 55 | — | — | — | — | — | 56.3 | 50.6 | 60.4 | — | — | 52.5 | — | — | — | — | — | — | |
| MemTokTok=642026.02 | 54.3 | — | — | — | — | — | 56.8 | 53 | 46.4 | 61.4 | — | — | — | — | — | — | — | — | |
| ColPaliSize=3B, Late Interaction=true, Neural Query Encoder=true2026.05 | 54.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BGE-M3 (caption)Size=0.57B, Late Interaction=true2026.05 | 53.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ColPaliTok=-2026.02 | 53.3 | — | — | — | — | — | 56.5 | 49.9 | 55.7 | 51.1 | — | — | — | — | — | — | — | — | |
| V-SPLADE + BiModernVBERT + BM25 (unstr.)Size=0.25/0.25B, Neural Query Encoder=true2026.05 | 53.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| V-SPLADE + BiModernVBERTSize=0.25/0.25B, Neural Query Encoder=true2026.05 | 51.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SeqResizeTok=642026.02 | 51.7 | — | — | — | — | — | 54.7 | 53.5 | 45.2 | 53.5 | — | — | — | — | — | — | — | — | |
| V-SPLADE + BM25 (unstr.)Size=0.25B2026.05 | 51.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BGE-M3 (unstr.)Size=0.57B, Late Interaction=true2026.05 | 50.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| V-SPLADE qualitySize=0.25B, FLOPs=1.92026.05 | 49.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| E5-VSize=8B, Late Interaction=true2026.05 | 49.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| V-SPLADE efficientSize=0.25B, FLOPs=1.12026.05 | 46.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BM25 (caption)FLOPs=1.32026.05 | 44.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ColFlorSize=0.17B, Late Interaction=true, Neural Query Encoder=true2026.05 | 43.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BM25 (unstr.)FLOPs=1.42026.05 | 41.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VLM2VecSize=4B, Late Interaction=true2026.05 | 41.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BiModernVBERTSize=0.25B, Neural Query Encoder=true2026.05 | 35.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Jina CLIP v2Size=0.9B, Neural Query Encoder=true2026.05 | 28.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SigLIP2-LSize=0.9B, Neural Query Encoder=true2026.05 | 27 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| JSDO-SparseSize=0.25B, Neural Query Encoder=true, FLOPs=18142026.05 | 12.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| D2SSize=0.25B, Neural Query Encoder=true, FLOPs=14.72026.05 | 4.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| 1D-PoolingBase Architecture=ColQwen2.5, Token=322026.01 | — | 25.1 | 47.2 | 41.7 | 26.1 | 35 | — | — | — | — | — | — | — | — | — | — | — | — | |
| 1D-PoolingBase Architecture=ColPali, Token=322026.01 | — | 16.3 | 28.6 | 35.3 | 21.6 | 25.5 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Average RankingTuning Status=Untuned2025.10 | — | — | — | — | — | — | — | — | — | — | — | — | -3 | 2.5 | -6.6 | 0.5 | -6.4 | 0.4 | |
| Average RankingTuning Status=Tuned2025.10 | — | — | — | — | — | — | — | — | — | — | — | — | -0.3 | 1.8 | -2.9 | 1.3 | -2.8 | 3.2 | |
| BaseBase Architecture=ColQwen2.5, Token=49622026.01 | — | 54.9 | 59.1 | 54.4 | 62 | 57.6 | — | — | — | — | — | — | — | — | — | — | — | — | |
| BaseBase Architecture=ColPali, Token=10312026.01 | — | 35 | 43.3 | 39.7 | 40.6 | 39.7 | — | — | — | — | — | — | — | — | — | — | — | — | |
| BiPaliBase Architecture=ColPali, Token=12026.01 | — | 15.5 | 25.9 | 36.9 | 21.6 | 26.1 | — | — | — | — | — | — | — | — | — | — | — | — | |
| BiQwen2Base Architecture=ColQwen2.5, Token=12026.01 | — | 41.3 | 48.7 | 49.3 | 50.4 | 46.4 | — | — | — | — | — | — | — | — | — | — | — | — | |
| c-semBackbone=VLM2Vec-V1-2B2026.03 | — | — | 20.32 | 11.01 | 11.86 | 13.41 | — | — | — | — | 10.44 | — | — | — | — | — | — | — | |
| CausalPaliBase Architecture=ColPali, Token=322026.01 | — | 35.7 | 49.3 | 50.3 | 46.4 | 45.4 | — | — | — | — | — | — | — | — | — | — | — | — | |
| CausalQwenBase Architecture=ColQwen2.5, Token=322026.01 | — | 49.8 | 50.5 | 52.2 | 53.7 | 51.6 | — | — | — | — | — | — | — | — | — | — | — | — | |
| cl-s-cBackbone=VLM2Vec-V1-2B2026.03 | — | — | 16.38 | 10.03 | 4.93 | 9 | — | — | — | — | 4.64 | — | — | — | — | — | — | — | |
| cl-s-mBackbone=VLM2Vec-V1-2B2026.03 | — | — | 13.46 | 6.5 | 6.89 | 7.65 | — | — | — | — | 3.74 | — | — | — | — | — | — | — | |
| cl-t-cBackbone=VLM2Vec-V1-2B2026.03 | — | — | 14.86 | 14.49 | 4.14 | 9.49 | — | — | — | — | 4.48 | — | — | — | — | — | — | — | |
| cl-t-mBackbone=VLM2Vec-V1-2B2026.03 | — | — | 13.34 | 4.6 | 8.78 | 7.97 | — | — | — | — | 5.15 | — | — | — | — | — | — | — | |
| ColParseBackbone=VLM2Vec-V1-2B2026.03 | — | — | 30.33 | 29.55 | 33.21 | 32.86 | — | — | — | — | 38.33 | — | — | — | — | — | — | — | |
| ColParseBackbone=VLM2Vec-V1-7B2026.03 | — | — | 42.63 | 42.89 | 50.55 | 44.73 | — | — | — | — | 42.86 | — | — | — | — | — | — | — | |
| ColParseBackbone=VLM2Vec-V2-2B2026.03 | — | — | 50.06 | 53.76 | 57.41 | 51.91 | — | — | — | — | 46.4 | — | — | — | — | — | — | — | |
| ColParseBackbone=LamRA-Ret2026.03 | — | — | 15.81 | 17.65 | 23.75 | 20.67 | — | — | — | — | 25.48 | — | — | — | — | — | — | — | |
| ColParseBackbone=GME-2B2026.03 | — | — | 55.14 | 52.08 | 56.24 | 53.71 | — | — | — | — | 51.36 | — | — | — | — | — | — | — | |
| ColParseBackbone=GME-7B2026.03 | — | — | 62.4 | 59.73 | 68.43 | 62.12 | — | — | — | — | 57.9 | — | — | — | — | — | — | — | |
| GME-2BOptimization=Base2026.03 | — | — | 54.25 | 50.65 | 59.44 | 53.37 | — | — | — | — | 49.15 | — | — | — | — | — | — | — | |
| GME-7BOptimization=Base2026.03 | — | — | 53.66 | 54.34 | 65.38 | 56.93 | — | — | — | — | 54.32 | — | — | — | — | — | — | — | |
| Guided Query Refinement (GQR)Tuning Status=Tuned2025.10 | — | — | — | — | — | — | — | — | — | — | — | — | 3.9 | 1.9 | 1.5 | 2 | 3.3 | 8.7 | |
| K-MeansBase Architecture=ColQwen2.5, Token=322026.01 | — | 41.3 | 50.4 | 53 | 35.9 | 45.2 | — | — | — | — | — | — | — | — | — | — | — | — | |
| K-MeansBase Architecture=ColPali, Token=322026.01 | — | 29.9 | 35.4 | 33.4 | 24.3 | 30.8 | — | — | — | — | — | — | — | — | — | — | — | — | |
| LamRA-RetOptimization=Base2026.03 | — | — | 10.75 | 9.65 | 6.32 | 9.48 | — | — | — | — | 11.18 | — | — | — | — | — | — | — | |
| multi-imgBackbone=VLM2Vec-V1-2B2026.03 | — | — | 15.59 | 14.9 | 5.7 | 10.54 | — | — | — | — | 5.95 | — | — | — | — | — | — | — | |
| RandomBase Architecture=ColQwen2.5, Token=322026.01 | — | 29.3 | 42.5 | 35.4 | 30.4 | 34.4 | — | — | — | — | — | — | — | — | — | — | — | — | |
| RandomBase Architecture=ColPali, Token=322026.01 | — | 23 | 27.2 | 33.7 | 19.1 | 25.8 | — | — | — | — | — | — | — | — | — | — | — | — | |
| RRFTuning Status=Untuned2025.10 | — | — | — | — | — | — | — | — | — | — | — | — | -2.8 | 2.6 | -6.5 | 0.3 | -6.2 | 1.3 | |
| RRFTuning Status=Tuned2025.10 | — | — | — | — | — | — | — | — | — | — | — | — | -0.1 | 1.9 | -2.9 | 1.2 | -4 | 5.5 | |
| s2m-addBackbone=VLM2Vec-V1-2B2026.03 | — | — | 10.4 | 8.59 | 4.68 | 6.78 | — | — | — | — | 3.46 | — | — | — | — | — | — | — | |
| s2m-addBackbone=VLM2Vec-V1-7B2026.03 | — | — | 29.49 | 38.26 | 31.73 | 30.57 | — | — | — | — | 22.8 | — | — | — | — | — | — | — | |
| s2m-mulBackbone=VLM2Vec-V1-2B2026.03 | — | — | 10 | 8.4 | 5.09 | 6.75 | — | — | — | — | 3.51 | — | — | — | — | — | — | — | |
| Score Aggregation (Min-Max)Tuning Status=Untuned2025.10 | — | — | — | — | — | — | — | — | — | — | — | — | 0.4 | 2.6 | -3 | 1.7 | -1.8 | 4.7 | |
| Score Aggregation (Min-Max)Tuning Status=Tuned2025.10 | — | — | — | — | — | — | — | — | — | — | — | — | 3.4 | 2.1 | 1.2 | 2.9 | 2.9 | 6.7 | |
| Score Aggregation (Softmax)Tuning Status=Untuned2025.10 | — | — | — | — | — | — | — | — | — | — | — | — | 1.5 | 1.9 | -0.9 | 0.8 | 0.6 | 5.4 | |
| Score Aggregation (Softmax)Tuning Status=Tuned2025.10 | — | — | — | — | — | — | — | — | — | — | — | — | 2.6 | 2.3 | 0.3 | 0.9 | 2.2 | 7.1 | |
| SemClusterBase Architecture=ColQwen2.5, Token=322026.01 | — | 42.6 | 51.5 | 48.5 | 39.5 | 45.5 | — | — | — | — | — | — | — | — | — | — | — | — | |
| SemClusterBase Architecture=ColPali, Token=322026.01 | — | 29.1 | 36.6 | 37.1 | 27.1 | 32.5 | — | — | — | — | — | — | — | — | — | — | — | — | |
| UniME-V2-2BOptimization=Base2026.03 | — | — | 9.5 | 15.78 | 15.51 | 14.96 | — | — | — | — | 19.03 | — | — | — | — | — | — | — | |
| VLM2Vec-V1-2BOptimization=Base2026.03 | — | — | 6.88 | 14.15 | 12.25 | 13.46 | — | — | — | — | 20.54 | — | — | — | — | — | — | — | |
| VLM2Vec-V1-7BOptimization=Base2026.03 | — | — | 4.93 | 13.74 | 6.82 | 9.19 | — | — | — | — | 11.27 | — | — | — | — | — | — | — | |
| VLM2Vec-V2-2BOptimization=Base2026.03 | — | — | 44.45 | 45.77 | 48.77 | 46.49 | — | — | — | — | 46.98 | — | — | — | — | — | — | — |