RLPF: Reinforcement Learning from Performance Feedback for Code Generation

• URL: http://arxiv.org/abs/2607.27271v1
• Date: Wed, 29 Jul 2026 11:39:55 GMT
• Title: RLPF: Reinforcement Learning from Performance Feedback for Code Generation
• Abstract summary: We study how to train code agents to prefer faster correct implementations, rather than treating efficiency only as an evaluation metric.<n>We propose textbfRLPF, reinforcement learning from performance feedback, which turns execution outcomes into a staged reward.<n>Fine-tuning Qwen3-32B with RLPF on PerfCodeBench raises correct-and-runnable solutions from $11.1%$ to $54.6%$.
• Score: 44.39557506834191
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Code models are increasingly trained with execution feedback, but most training signals still stop at correctness. This leaves an important gap for systems code: two programs can pass the same tests while differing greatly in runtime. We study how to train code agents to prefer faster correct implementations, rather than treating efficiency only as an evaluation metric. The key difficulty is that runtime is a fragile reward. It is meaningful only after a program is correct, varies across tasks, and gives little guidance when most sampled programs fail to compile or run. We propose \textbf{RLPF}, reinforcement learning from performance feedback, which turns execution outcomes into a staged reward. Failed programs are ordered by execution progress, while correct programs are ranked by their relative improvement from the baseline toward the expert reference. This gives useful feedback before correctness and performance-sensitive feedback after correctness. Fine-tuning Qwen3-32B with RLPF on PerfCodeBench raises correct-and-runnable solutions from $11.1\%$ to $54.6\%$ and improves relative efficiency from $8.1\%$ to $38.6\%$. The trained model becomes competitive with stronger open-weight systems, and its optimization behavior transfers modestly to EffiBench-X. Additional studies show that model-generated references provide useful but weaker supervision, and that the full composite reward is more reliable than correctness-only or runtime-only baselines. These results suggest that code agents can be trained not only to pass tests, but also to optimize the programs they write.

Related papers

• Write, Execute, Refine: From Skill Followers to Skill Optimizers via Reinforcement Learning from Execution Feedback [52.78692582507693]
  Expert-written natural language skills can improve tool-using agents, yet agent-authored skills perform 8-11 points worse than using no skill.<n>We study how to organize execution experience from intermediate skills into training states for an executor.
  arXiv  Detail & Related papers  (2026-08-18T09:52:48Z)
• When Do Intrinsic Rewards Work for Code Reasoning? A Comprehensive Study [77.64957118012423]
  Reinforcement learning with verifiable rewards (RLVR) has driven substantial progress in large language model reasoning.<n>Recent work addresses this by deriving rewards from a model's own signals, such as majority voting or confidence-based scores.<n>We present a systematic empirical study of intrinsic reward methods for code generation.
  arXiv  Detail & Related papers  (2026-06-18T19:15:50Z)
• REVES: REvision and VErification--Augmented Training for Test-Time Scaling [53.197756110943395]
  We propose a two-stage iterative framework that alternates between online data/prompt augmentation and policy optimization.<n>We observe gains of +6.5 points over the RL baseline and +4.0 points over standard multi-turn training.
  arXiv  Detail & Related papers  (2026-06-17T10:37:23Z)
• Correct Is Not Enough: Training Reasoning Planners with Executor-Grounded Rewards [15.458797431655961]
  Reinforcement learning with verifiable rewards has become a common way to improve explicit reasoning in large language models.<n>But final-answer correctness alone does not reveal whether the reasoning trace is faithful, reliable, or useful to the model that consumes it.<n>We propose TraceLift, a planner-executor training framework that treats reasoning as a consumable intermediate artifact.
  arXiv  Detail & Related papers  (2026-05-05T15:24:40Z)
• ACE: Self-Evolving LLM Coding Framework via Adversarial Unit Test Generation and Preference Optimization [28.464748466548965]
  ACE is a self-evolving code generation framework based on a solver--adversary architecture.<n>It prioritizes active failure discovery through execution-centric supervision.<n>ACE consistently outperforms strong solver--verifier baselines.
  arXiv  Detail & Related papers  (2026-04-17T07:20:01Z)
• ExecVerify: White-Box RL with Verifiable Stepwise Rewards for Code Execution Reasoning [9.137158235106941]
  We introduce ExecVerify, which goes beyond text imitation by incorporating verifiable white-box rewards derived from execution traces.<n>Experiments demonstrate that a 7B model trained with ExecVerify achieves performance comparable to 32B models on code reasoning benchmarks.
  arXiv  Detail & Related papers  (2026-03-11T18:49:45Z)
• CVeDRL: An Efficient Code Verifier via Difficulty-aware Reinforcement Learning [57.24524263804788]
  Code verifiers play a critical role in post-verification for LLM-based code generation.<n>Existing supervised fine-tuning methods suffer from data scarcity, high failure rates, and poor inference efficiency.<n>We show that naive RL with only functionality rewards fails to generate effective unit tests for difficult branches and samples.
  arXiv  Detail & Related papers  (2026-01-30T10:33:29Z)
• NExT: Teaching Large Language Models to Reason about Code Execution [50.93581376646064]
  Large language models (LLMs) of code are typically trained on the surface textual form of programs. We propose NExT, a method to teach LLMs to inspect the execution traces of programs and reason about their run-time behavior.
  arXiv  Detail & Related papers  (2024-04-23T01:46:32Z)
• Learning from Self-Sampled Correct and Partially-Correct Programs [96.66452896657991]
  We propose to let the model perform sampling during training and learn from both self-sampled fully-correct programs and partially-correct programs. We show that our use of self-sampled correct and partially-correct programs can benefit learning and help guide the sampling process. Our proposed method improves the pass@k performance by 3.1% to 12.3% compared to learning from a single reference program with MLE.
  arXiv  Detail & Related papers  (2022-05-28T03:31:07Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.