4 articles tagged with "Reasoning"
The recent study on compositional neuro-symbolic reasoning emphasizes structured abstraction for improved AI generalization, particularly within the Abstraction and Reasoning Corpus (ARC).
A recent study on Large Reasoning Models (LRMs) highlights their reasoning capabilities while also addressing the pitfalls of overthinking and computational inefficiency.
The TRACED framework offers a novel approach to assess LLM reasoning quality, moving beyond traditional scalar probability evaluations by focusing on structural dynamics.
The HEAL framework introduces a novel approach to distilling reasoning capabilities from larger models into smaller, more efficient ones, addressing traditional sampling limitations.