Hugging Face Trending Papers

Knowledge Knows, Verbalization Tells: Disentangling Latent Directions for Mathematical Solvability in LLMs

Read the original on Hugging Face Trending Papers →

Although LLMs have made significant progress in mathematical reasoning, determining whether a mathematical problem is solvable remains a fundamental yet challenging capability. While recent studies have probed internal representations of model solvability beliefs, verbalization has primarily been studied behaviorally rather than as an internal representation, limiting its analysis and manipulation.

Summary generated by The Flow from the publisher's feed. The full article lives at Hugging Face Trending Papers.

arXiv AI
Jul 22

Fluid Reasoning Representations

arXiv:2602. 04843v2 Announce Type: replace Abstract: Frontier large language models increasingly solve complex tasks involving abstract concepts through extended test-time thinking.

By Dmitrii Kharlapenko, Terry Jingchen Zhang, Arth Singh, Alessandro Stolfo, Arthur Conmy, Mrinmaya Sachan, Zhijing Jin