Composed Retrieval on Quilt-VQA Red
0.643R@1HOMIE (Qwen3-VL-2B)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| HOMIE (Qwen3-VL-2B)Model Category=MLLM-Based2025.02 | 0.643 | 0.897 | 0.929 | |
| HOMIE (Qwen3-VL-8B)Model Category=MLLM-Based2025.02 | 0.575 | 0.893 | 0.929 | |
| LamRA-7BModel Category=MLLM-Based2025.02 | 0.524 | 0.667 | 0.734 | |
| MUSKFusion Strategy=Add, Model Category=Dual-Encoder2025.02 | 0.5 | 0.698 | 0.778 | |
| GME-7BModel Category=MLLM-Based2025.02 | 0.488 | 0.611 | 0.671 | |
| MUSKFusion Strategy=RRF, Model Category=Dual-Encoder2025.02 | 0.373 | 0.623 | 0.714 | |
| Pathgen-CLIP-LFusion Strategy=Add, Model Category=Dual-Encoder2025.02 | 0.369 | 0.492 | 0.532 | |
| Pathgen-CLIPFusion Strategy=Add, Model Category=Dual-Encoder2025.02 | 0.353 | 0.5 | 0.563 | |
| Patho-CLIP-LFusion Strategy=Add, Model Category=Dual-Encoder2025.02 | 0.353 | 0.587 | 0.663 | |
| PathCLIPFusion Strategy=Add, Model Category=Dual-Encoder2025.02 | 0.337 | 0.464 | 0.508 | |
| PubmedCLIPFusion Strategy=Add, Model Category=Dual-Encoder2025.02 | 0.302 | 0.452 | 0.508 | |
| QuiltNetFusion Strategy=Add, Model Category=Dual-Encoder2025.02 | 0.282 | 0.369 | 0.413 | |
| PLIPFusion Strategy=Add, Model Category=Dual-Encoder2025.02 | 0.266 | 0.425 | 0.46 | |
| Pathgen-CLIP-LFusion Strategy=RRF, Model Category=Dual-Encoder2025.02 | 0.266 | 0.456 | 0.536 | |
| Patho-CLIP-LFusion Strategy=RRF, Model Category=Dual-Encoder2025.02 | 0.25 | 0.488 | 0.591 | |
| Qwen2.5-VL-7BModel Category=MLLM-Based2025.02 | 0.242 | 0.421 | 0.5 | |
| Lingshu-7BModel Category=MLLM-Based2025.02 | 0.238 | 0.397 | 0.484 | |
| Pathgen-CLIPFusion Strategy=RRF, Model Category=Dual-Encoder2025.02 | 0.23 | 0.44 | 0.54 | |
| Patho-CLIPFusion Strategy=Add, Model Category=Dual-Encoder2025.02 | 0.218 | 0.401 | 0.472 | |
| PathoR1-7BModel Category=MLLM-Based2025.02 | 0.202 | 0.417 | 0.524 | |
| BiomedCLIPFusion Strategy=Add, Model Category=Dual-Encoder2025.02 | 0.19 | 0.337 | 0.46 | |
| HuatuoGPT-V-7BModel Category=MLLM-Based2025.02 | 0.19 | 0.341 | 0.409 | |
| CONCHFusion Strategy=Add, Model Category=Dual-Encoder2025.02 | 0.183 | 0.425 | 0.583 | |
| Qwen3-VL-8BModel Category=MLLM-Based2025.02 | 0.167 | 0.294 | 0.341 | |
| QuiltNetFusion Strategy=RRF, Model Category=Dual-Encoder2025.02 | 0.147 | 0.278 | 0.369 | |
| Patho-CLIPFusion Strategy=RRF, Model Category=Dual-Encoder2025.02 | 0.147 | 0.337 | 0.448 | |
| CONCHFusion Strategy=RRF, Model Category=Dual-Encoder2025.02 | 0.135 | 0.333 | 0.492 | |
| PathCLIPFusion Strategy=RRF, Model Category=Dual-Encoder2025.02 | 0.123 | 0.286 | 0.393 | |
| E5-V-7BModel Category=MLLM-Based2025.02 | 0.091 | 0.155 | 0.206 | |
| PLIPFusion Strategy=RRF, Model Category=Dual-Encoder2025.02 | 0.083 | 0.242 | 0.353 | |
| BiomedCLIPFusion Strategy=RRF, Model Category=Dual-Encoder2025.02 | 0.079 | 0.218 | 0.361 | |
| PubmedCLIPFusion Strategy=RRF, Model Category=Dual-Encoder2025.02 | 0.036 | 0.155 | 0.278 | |
| Qwen3-VL-2BModel Category=MLLM-Based2025.02 | 0.02 | 0.096 | 0.139 |