CodeEvolve uses LLMs and runtime analysis to enhance code performance

Researchers have developed CodeEvolve, a new framework that uses Large Language Models (LLMs) to automatically enhance code quality and performance. This system integrates runtime profiling data to identify critical optimization targets, reducing the need for manual analysis. CodeEvolve then generates, evaluates, and refines code edits, ensuring functional correctness through various checks including LLM-based review. In tests, it achieved significant speedups on Java codebases and demonstrated reliable optimization for Salesforce Apex. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT Introduces an automated approach to code optimization, potentially improving developer productivity and software performance.

RANK_REASON This is a research paper detailing a novel framework for code enhancement using LLMs. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• CodeEvolve
• Large Language Models
• LLMs
• OpenEvolve
• Monte Carlo Tree Search
• MCTS
• Java
• Salesforce Apex
• Java Flight Recorder
• JFR

• paper
• other