/Neural-Comprehension

[ICLR 2024] We have enabled language models to more fundamental comprehension of the concepts, to achieve completely absolute accuracy in symbolic reasoning without additional tools.

Primary LanguagePythonMIT LicenseMIT

Stargazers