Dimensions overall score 7.0

GitHub Code Pulse

No public code linked for this paper yet.

• Large Vision-Language Models (LVLMs) struggle with complex, multi-step reasoning when visual input serves as a clue rather than a direct answer.limitation 90%
• Successfully solving a rebus puzzle requires extracting visual/textual attributes, retrieving linguistic prior knowledge, and performing abstract mapping to synthesize elements into meaning outside pixel space.technical 85%
• RebusBench contains 1,164 puzzles designed to evaluate neurosymbolic capability integrating perception and knowledge.method 95%
• State-of-the-art models (including Qwen, InternVL, and LLaVA) achieve below 10% Exact Match and 20% semantic accuracy on RebusBench.result 90%
• No significant improvement is observed from model scaling or In-Context Learning (ICL) on rebus puzzle performance.result 85%
• Models possess necessary visual and linguistic components but lack the cognitive reasoning glue to connect them.limitation 80%
• Current LVLMs have achieved remarkable proficiency in explicit visual recognition and describing what is directly visible.technical 85%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card