CAVE: A Structured Credit Assignment Approach for Fragmented Visual Evidence Reasoning
For researchers working on vision-language reasoning, this work addresses the understudied problem of integrating nonlocal visual evidence, though the method is incremental (applying GRPO with custom signals).
CAVE introduces a structured credit assignment method (based on GRPO) that improves vision-language models' ability to integrate fragmented visual evidence, achieving substantial gains on both a new benchmark (TRACER-Bench) and public benchmarks while maintaining competitive general performance.
Vision-Language Models (VLMs) have achieved strong performance on general multimodal reasoning, yet remain challenged in integrating nonlocal visual information to support semantically underdetermined visual reasoning. We describe this challenge as Fragmented Visual Reasoning. To this end, we propose Credit Assignment for Visual Evidence (CAVE), a structured process-reward method based on GRPO for interleaved visual reasoning. Specifically, CAVE evaluates the contribution of intermediate steps at the action level via three complementary reasoning process signals: belief update, evidence acquisition, and adaptive focus control, thereby guiding the model to optimize each reasoning action and learn more reliable visual reasoning strategies. Meanwhile, we construct TRACER-Bench, which covers four nonlocal and semantically confusable reasoning dimensions and provides key intermediate evidence to supervise reasoning paths. Experiments demonstrate that CAVE substantially improves performance on tasks requiring fragmented visual evidence integration, covering both public benchmarks and our newly introduced TRACER-Bench, while retaining competitive performance on general multimodal evaluations. Further analyses reveal that CAVE effectively improves the visual reasoning capacity and exhibits stronger robustness under longer-range and deeper cross-region dependencies.