Text Question Answering on MuSiQue (Accuracy)
69.6AccuracyLensVLM
Evaluation Results
| Method | Links | |
|---|---|---|
| LensVLMICR=5×, ECR=4.3×2026.05 | 69.6 | |
| TextICR=1×, ECR=1.0×2026.05 | 68.7 | |
| LensVLMSFTICR=5×, ECR=4.6×2026.05 | 66.1 | |
| LensVLMICR=10×, ECR=7.4×2026.05 | 62.4 | |
| LensVLMSFTICR=10×, ECR=8.1×2026.05 | 59.9 | |
| BGE-M3ICR=5×, ECR=4.5×2026.05 | 59 | |
| Jina-v4-MMICR=5×, ECR=4.6×2026.05 | 58.5 | |
| Jina-v4ICR=5×, ECR=4.5×2026.05 | 57.5 | |
| Qwen3-EmbICR=5×, ECR=4.5×2026.05 | 56.5 | |
| BM25ICR=5×, ECR=4.5×2026.05 | 53.2 | |
| ColPaliICR=5×, ECR=4.7×2026.05 | 49.5 | |
| LensVLMICR=15×, ECR=10.1×2026.05 | 49 | |
| Qwen3-EmbICR=10×, ECR=8.3×2026.05 | 48.2 | |
| BGE-M3ICR=10×, ECR=8.3×2026.05 | 48 | |
| Jina-v4ICR=10×, ECR=8.3×2026.05 | 47.7 | |
| Jina-v4-MMICR=10×, ECR=8.3×2026.05 | 45 | |
| BM25ICR=10×, ECR=8.3×2026.05 | 43.8 | |
| Jina-v4ICR=15×, ECR=11.9×2026.05 | 41.7 | |
| Qwen3-EmbICR=15×, ECR=11.9×2026.05 | 41.3 | |
| LensVLMSFTICR=15×, ECR=10.7×2026.05 | 41.2 | |
| BGE-M3ICR=15×, ECR=11.9×2026.05 | 40.2 | |
| Jina-v4-MMICR=15×, ECR=11.9×2026.05 | 39 | |
| BM25ICR=15×, ECR=11.9×2026.05 | 36.8 | |
| Base + ExpandICR=5×, ECR=4.5×2026.05 | 34 | |
| LLMLingua-2ICR=5×, ECR=5.5×2026.05 | 33.8 | |
| GlyphICR=5×, ECR=4.5×2026.05 | 33 | |
| ColPaliICR=10×, ECR=8.3×2026.05 | 32.5 | |
| ColPaliICR=15×, ECR=11.9×2026.05 | 31.5 | |
| GlyphICR=10×, ECR=8.6×2026.05 | 26.5 | |
| LLMLingua-2ICR=10×, ECR=11.3×2026.05 | 26.5 | |
| LLMLingua-2ICR=15×, ECR=15.7×2026.05 | 21.8 | |
| GlyphICR=15×, ECR=10.8×2026.05 | 21 | |
| Comp. ImageICR=5×, ECR=5×2026.05 | 19.5 | |
| Base + ExpandICR=10×, ECR=8.0×2026.05 | 15 | |
| Comp. ImageICR=10×, ECR=10×2026.05 | 14 | |
| Comp. ImageICR=15×, ECR=15×2026.05 | 12.5 | |
| Base + ExpandICR=15×, ECR=11.0×2026.05 | 12.5 |