Retrieval on Flickr2.4M (test)
5.266Ave AesQCQC (Blip2)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| QCQC (Blip2)VLM=Blip2, Aes Cond=H, Rel Cond=L, Controllable=true2026.02 | 5.266 | 0.354 | |
| QCQC (Blip2)VLM=Blip2, Aes Cond=H, Rel Cond=H, Controllable=true2026.02 | 5.236 | 0.387 | |
| QCQC (CoCa)VLM=CoCa, Aes Cond=H, Rel Cond=L, Controllable=true2026.02 | 5.182 | 0.357 | |
| QCQC (Blip2)VLM=Blip2, Aes Cond=H, Rel Cond=M, Controllable=true2026.02 | 5.16 | 0.367 | |
| QCQC (CoCa)VLM=CoCa, Aes Cond=H, Rel Cond=H, Controllable=true2026.02 | 5.124 | 0.385 | |
| QCQC (CoCa)VLM=CoCa, Aes Cond=H, Rel Cond=M, Controllable=true2026.02 | 5.095 | 0.366 | |
| GPT-4oVLM=None, Aes Cond=H, Rel Cond=H, Controllable=false2026.02 | 5.056 | 0.361 | |
| QCQC (Blip2)VLM=Blip2, Aes Cond=M, Rel Cond=L, Controllable=true2026.02 | 4.948 | 0.354 | |
| QCQC (Blip2)VLM=Blip2, Aes Cond=M, Rel Cond=M, Controllable=true2026.02 | 4.946 | 0.374 | |
| QCQC (CoCa)VLM=CoCa, Aes Cond=M, Rel Cond=M, Controllable=true2026.02 | 4.921 | 0.369 | |
| FT (CoCa)VLM=CoCa, Aes Cond=H, Rel Cond=M, Controllable=false2026.02 | 4.896 | 0.366 | |
| QCQC (CoCa)VLM=CoCa, Aes Cond=M, Rel Cond=H, Controllable=true2026.02 | 4.894 | 0.386 | |
| QCQC (Blip2)VLM=Blip2, Aes Cond=M, Rel Cond=H, Controllable=true2026.02 | 4.885 | 0.391 | |
| QCQC (CoCa)VLM=CoCa, Aes Cond=M, Rel Cond=L, Controllable=true2026.02 | 4.878 | 0.355 | |
| FT (CoCa)VLM=CoCa, Aes Cond=M, Rel Cond=H, Controllable=false2026.02 | 4.876 | 0.366 | |
| FT (CoCa)VLM=CoCa, Aes Cond=L, Rel Cond=H, Controllable=false2026.02 | 4.864 | 0.367 | |
| FT (CoCa)VLM=CoCa, Aes Cond=H, Rel Cond=H, Controllable=false2026.02 | 4.853 | 0.368 | |
| FT (CoCa)VLM=CoCa, Aes Cond=L, Rel Cond=L, Controllable=false2026.02 | 4.848 | 0.366 | |
| FT (CoCa)VLM=CoCa, Aes Cond=M, Rel Cond=L, Controllable=false2026.02 | 4.847 | 0.365 | |
| LLaMA3VLM=None, Aes Cond=H, Rel Cond=H, Controllable=false2026.02 | 4.842 | 0.352 | |
| FT (Blip2)VLM=Blip2, Aes Cond=L, Rel Cond=L, Controllable=false2026.02 | 4.838 | 0.369 | |
| LLaMA3VLM=None, Aes Cond=M, Rel Cond=M, Controllable=false2026.02 | 4.837 | 0.351 | |
| LLaMA3VLM=None, Aes Cond=L, Rel Cond=H, Controllable=false2026.02 | 4.831 | 0.351 | |
| FT (CoCa)VLM=CoCa, Aes Cond=H, Rel Cond=L, Controllable=false2026.02 | 4.829 | 0.367 | |
| FT (CoCa)VLM=CoCa, Aes Cond=M, Rel Cond=M, Controllable=false2026.02 | 4.827 | 0.363 | |
| LLaMA3VLM=None, Aes Cond=M, Rel Cond=L, Controllable=false2026.02 | 4.823 | 0.353 | |
| LLaMA3VLM=None, Aes Cond=L, Rel Cond=M, Controllable=false2026.02 | 4.822 | 0.351 | |
| FT (CoCa)VLM=CoCa, Aes Cond=L, Rel Cond=M, Controllable=false2026.02 | 4.818 | 0.365 | |
| FT (Blip2)VLM=Blip2, Aes Cond=H, Rel Cond=H, Controllable=false2026.02 | 4.818 | 0.368 | |
| GPT-4oVLM=None, Aes Cond=H, Rel Cond=M, Controllable=false2026.02 | 4.816 | 0.357 | |
| LLaMA3VLM=None, Aes Cond=H, Rel Cond=L, Controllable=false2026.02 | 4.798 | 0.354 | |
| GPT-4oVLM=None, Aes Cond=M, Rel Cond=H, Controllable=false2026.02 | 4.791 | 0.356 | |
| GPT-4oVLM=None, Aes Cond=H, Rel Cond=L, Controllable=false2026.02 | 4.791 | 0.361 | |
| LLaMA3VLM=None, Aes Cond=M, Rel Cond=H, Controllable=false2026.02 | 4.784 | 0.35 | |
| FT (Blip2)VLM=Blip2, Aes Cond=M, Rel Cond=H, Controllable=false2026.02 | 4.772 | 0.365 | |
| FT (Blip2)VLM=Blip2, Aes Cond=H, Rel Cond=M, Controllable=false2026.02 | 4.749 | 0.359 | |
| FT (Blip2)VLM=Blip2, Aes Cond=L, Rel Cond=H, Controllable=false2026.02 | 4.744 | 0.369 | |
| PrefixVLM=None, Aes Cond=L, Rel Cond=L, Controllable=false2026.02 | 4.735 | 0.35 | |
| PrefixVLM=None, Aes Cond=L, Rel Cond=M, Controllable=false2026.02 | 4.735 | 0.35 | |
| PrefixVLM=None, Aes Cond=L, Rel Cond=H, Controllable=false2026.02 | 4.735 | 0.35 | |
| PrefixVLM=None, Aes Cond=M, Rel Cond=L, Controllable=false2026.02 | 4.735 | 0.35 | |
| PrefixVLM=None, Aes Cond=M, Rel Cond=M, Controllable=false2026.02 | 4.735 | 0.35 | |
| PrefixVLM=None, Aes Cond=M, Rel Cond=H, Controllable=false2026.02 | 4.735 | 0.35 | |
| PrefixVLM=None, Aes Cond=H, Rel Cond=L, Controllable=false2026.02 | 4.735 | 0.35 | |
| PrefixVLM=None, Aes Cond=H, Rel Cond=M, Controllable=false2026.02 | 4.735 | 0.35 | |
| PrefixVLM=None, Aes Cond=H, Rel Cond=H, Controllable=false2026.02 | 4.735 | 0.35 | |
| LLaMA3VLM=None, Aes Cond=L, Rel Cond=L, Controllable=false2026.02 | 4.73 | 0.351 | |
| GPT-4oVLM=None, Aes Cond=L, Rel Cond=H, Controllable=false2026.02 | 4.728 | 0.357 | |
| FT (Blip2)VLM=Blip2, Aes Cond=H, Rel Cond=L, Controllable=false2026.02 | 4.727 | 0.373 | |
| LLaMA3VLM=None, Aes Cond=H, Rel Cond=M, Controllable=false2026.02 | 4.722 | 0.354 | |
| GPT-4oVLM=None, Aes Cond=M, Rel Cond=L, Controllable=false2026.02 | 4.712 | 0.358 | |
| PTVLM=None, Aes Cond=H, Rel Cond=H, Controllable=false2026.02 | 4.692 | 0.352 | |
| PTVLM=None, Aes Cond=M, Rel Cond=L, Controllable=false2026.02 | 4.688 | 0.35 | |
| PTVLM=None, Aes Cond=L, Rel Cond=L, Controllable=false2026.02 | 4.681 | 0.351 | |
| QCQC (CoCa)VLM=CoCa, Aes Cond=L, Rel Cond=M, Controllable=true2026.02 | 4.674 | 0.372 | |
| FT (Blip2)VLM=Blip2, Aes Cond=L, Rel Cond=M, Controllable=false2026.02 | 4.674 | 0.36 | |
| PTVLM=None, Aes Cond=L, Rel Cond=H, Controllable=false2026.02 | 4.673 | 0.35 | |
| GPT-4oVLM=None, Aes Cond=M, Rel Cond=M, Controllable=false2026.02 | 4.668 | 0.36 | |
| PTVLM=None, Aes Cond=M, Rel Cond=H, Controllable=false2026.02 | 4.654 | 0.347 | |
| GPT-4oVLM=None, Aes Cond=L, Rel Cond=M, Controllable=false2026.02 | 4.651 | 0.361 | |
| QCQC (CoCa)VLM=CoCa, Aes Cond=L, Rel Cond=L, Controllable=true2026.02 | 4.646 | 0.354 | |
| PTVLM=None, Aes Cond=L, Rel Cond=M, Controllable=false2026.02 | 4.639 | 0.344 | |
| QCQC (CoCa)VLM=CoCa, Aes Cond=L, Rel Cond=H, Controllable=true2026.02 | 4.632 | 0.382 | |
| PTVLM=None, Aes Cond=H, Rel Cond=L, Controllable=false2026.02 | 4.61 | 0.349 | |
| FT (Blip2)VLM=Blip2, Aes Cond=M, Rel Cond=M, Controllable=false2026.02 | 4.599 | 0.362 | |
| FT (Blip2)VLM=Blip2, Aes Cond=M, Rel Cond=L, Controllable=false2026.02 | 4.592 | 0.365 | |
| QCQC (Blip2)VLM=Blip2, Aes Cond=L, Rel Cond=M, Controllable=true2026.02 | 4.56 | 0.374 | |
| PTVLM=None, Aes Cond=H, Rel Cond=M, Controllable=false2026.02 | 4.556 | 0.352 | |
| QCQC (Blip2)VLM=Blip2, Aes Cond=L, Rel Cond=L, Controllable=true2026.02 | 4.528 | 0.355 | |
| PTVLM=None, Aes Cond=M, Rel Cond=M, Controllable=false2026.02 | 4.504 | 0.346 | |
| QCQC (Blip2)VLM=Blip2, Aes Cond=L, Rel Cond=H, Controllable=true2026.02 | 4.47 | 0.393 | |
| GPT-4oVLM=None, Aes Cond=L, Rel Cond=L, Controllable=false2026.02 | 4.359 | 0.378 |