Book
Open access
Jul 2026
Can LLMs Really Reason about Code? Studying How Well LLMs Understand the Relation between Input, Code, and Output
In this work, the capacity of LLMs to reason about the semantics of code is examined—specifically, their ability to relate code, its inputs, and its outputs to each other, and four datasets are constructed to evaluate diverse program-understanding capabilities.
Norman Becker, T. Mammadov, Andreas Zeller
· AIware · 2 citations