AI Insight
This paper addresses the readability of code generated by large language models (LLMs) by proposing a multitask representation engineering (RepE) framework. While existing research primarily focuses on code correctness, the authors argue that readability is equally important and develop a low-cost method to simultaneously control multiple aspects of code quality. The theoretical and experimental analysis examines the tradeoff between improving code readability and maintaining correctness when applying multitask steering techniques.
Why it matters
As LLMs become increasingly used for code generation in software development, ensuring that generated code is not only functionally correct but also human-readable is critical for maintainability and collaboration. This work provides a computationally efficient approach to enhance code quality across multiple dimensions, which could improve developer productivity and code review processes.
arXiv:2606.06214v1 Announce Type: cross
Abstract: Correctness and readability are key measures of code quality, respectively ensuring functional fidelity and ease of comprehension. While most existing research focuses on improving the correctness of large language models~(LLMs) generated codes, readability remains under-addressed. Enhancing readability through targeted control is challenging due to its subjective nature. In this article, we employ representation engineering~(RepE) as the targeted control method given its characteristics of low data dependency and low computational cost. Prior work on RepE has primarily focused on the targeted control for a single task, but improving the code readability requires the control across multiple tasks. Accordingly we proposes the multitask RepE framework and theoretically discuss the impact of the multitask steering method on the tradeoff between the code readability and correctness. We further provide comprehensive experiments in support. All the relevant implementations are open-source and available upon request.
Source: Towards the Readability of LLM-Generated Codes through Multitask Representation Engineering