ConMe: Rethinking Evaluation of Compositional Reasoning for Modern VLMs
This addresses the need for more rigorous evaluation of compositional reasoning in VLMs, which is incremental as it improves benchmarking rather than solving the reasoning problem itself.
The paper tackles the problem that existing benchmarks may not adequately test compositional reasoning in modern Vision-Language Models (VLMs) by introducing ConMe, a new benchmark and data generation pipeline that uses VLMs to create hard questions, resulting in up to a 33% decrease in performance compared to prior benchmarks.
Compositional Reasoning (CR) entails grasping the significance of attributes, relations, and word order. Recent Vision-Language Models (VLMs), comprising a visual encoder and a Large Language Model (LLM) decoder, have demonstrated remarkable proficiency in such reasoning tasks. This prompts a crucial question: have VLMs effectively tackled the CR challenge? We conjecture that existing CR benchmarks may not adequately push the boundaries of modern VLMs due to the reliance on an LLM-only negative text generation pipeline. Consequently, the negatives produced either appear as outliers from the natural language distribution learned by VLMs' LLM decoders or as improbable within the corresponding image context. To address these limitations, we introduce ConMe -- a compositional reasoning benchmark and a novel data generation pipeline leveraging VLMs to produce `hard CR Q&A'. Through a new concept of VLMs conversing with each other to collaboratively expose their weaknesses, our pipeline autonomously generates, evaluates, and selects challenging compositional reasoning questions, establishing a robust CR benchmark, also subsequently validated manually. Our benchmark provokes a noteworthy, up to 33%, decrease in CR performance compared to preceding benchmarks, reinstating the CR challenge even for state-of-the-art VLMs.