Background
Vision--language models (VLMs) are tested for fine-grained understanding with image--text matching probes, a model must prefer the correct image or caption over handcrafted negative. In retrieval, the correct item must beat thousands of alternatives.
Methodology
We measure how these two evaluation methods relate by using two probing benchmarks, \svo{} and SugarCrepe, for retrieval tasks. For 100 image and caption queries we judge the top-10 retrievals from full dataset of six VLMs, with human annotators and an LLM assessor. We than compare how well models rank pool of items on the same queries, while changing only where the negatives comes from:
- the benchmark,
- a random draw, or
- the judged-incorrect retrievals of other models.
Results
On \svo{} the accuracy of retrieving a positive over handcrafted negatives is significantly higher than for retrieved ones (the correct item is preferred in 85\% against 77\% of the time). Using pool of ten retrieved negatives leads to same accuracy as when one uses pool of 200 to 430 random negatives.
On SugarCrepe the order is reversed: its handcrafted negatives are harder than the retrieved ones. Incomplete relevance labels explain most of the low retrieval scores on \svo{}, where the top-ranked item is correct for 54--90\% of the queries under human labels against 6--21\% under the benchmark's labels.
However, we detect no association between passing the probe for a query and retrieving a correct item first.
Agreement
The LLM assessor o3 agrees with individual annotators ($κ$ 0.57--0.61) slightly less than they agree with each other (0.64--0.68).