Compositional Neuro-Symbolic Reasoning

[-0.5em] Anugyan Das, Omkar Ghugarkar, Vishvesh Bhat, Asad Aali · 2026

Read Compositional Neuro-Symbolic Reasoning by [-0.5em] Anugyan Das, Omkar Ghugarkar, Vishvesh Bhat, Asad Aali (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-02434.txt.

Description

Das and team build a hybrid reasoner for the Abstraction and Reasoning Corpus that splits seeing, neural guesses and symbolic checks. Object structure is read from grids, neural priors propose new changes from a fixed domain language. On ARC-AGI-2 it lifts base model scores from 16 percent to about 24 percent alone and to about 31 percent paired with ARC Lang Solver via a meta judge. Gains come without task tuning or reward learning, and with less brute force search.

Catalog record

CREATOR[-0.5em] Anugyan Das, Omkar Ghugarkar, Vishvesh Bhat, Asad Aali
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE55873
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.02434

Browse the full library · About the Tower of Babel