Afterburner: Reinforcement Learning Facilitates Self-Improving Code Efficiency Optimization

TL;DR

This paper presents a reinforcement learning-based framework that enables large language models to iteratively improve code efficiency at test time using execution feedback, surpassing human benchmarks.

Contribution

It introduces a novel RL-based test-time optimization method for LLMs to self-improve code efficiency through iterative refinement with empirical feedback.

Findings

GRPO with RL significantly boosts code efficiency metrics.

SFT and DPO quickly saturate in efficiency gains.

The approach outperforms human submissions in code efficiency.

Abstract

Large Language Models (LLMs) generate functionally correct solutions but often fall short in code efficiency, a critical bottleneck for real-world deployment. In this paper, we introduce a novel test-time iterative optimization framework to address this, employing a closed-loop system where LLMs iteratively refine code based on empirical performance feedback from an execution sandbox. We explore three training strategies: Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and Group Relative Policy Optimization (GRPO). Experiments on our Venus dataset and the APPS benchmark show that SFT and DPO rapidly saturate in efficiency gains. In contrast, GRPO, using reinforcement learning (RL) with execution feedback, continuously optimizes code performance, significantly boosting both pass@1 (from 47% to 62%) and the likelihood of outperforming human submissions in efficiency (from 31% to 45%). Our work demonstrates effective test-time code efficiency improvement and critically reveals the power of RL in teaching LLMs to truly self-improve code efficiency.