Pith. sign in

REVIEW 23 cited by

RLEF: Grounding Code LLMs in Execution Feedback with Reinforcement Learning

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2410.02089 v2 pith:ZMMNP36S submitted 2024-10-02 cs.CL cs.AI

Jonas Gehring , Kunhao Zheng , Jade Copet , Vegard Mella , Quentin Carbonneaux , Taco Cohen , Gabriel Synnaeve This is my paper · ORCID
classification cs.CLcs.AI
keywords llmsfeedbackcodemodelsachieveexecutionlargelearning
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Large language models (LLMs) deployed as agents solve user-specified tasks over multiple steps while keeping the required manual engagement to a minimum. Crucially, such LLMs need to ground their generations in any feedback obtained to reliably achieve the desired outcomes. We propose an end-to-end reinforcement learning method for teaching models to leverage execution feedback in the realm of code synthesis, where state-of-the-art LLMs struggle to improve code iteratively compared to independent sampling. We benchmark on competitive programming tasks, where we achieve new state-of-the art results with both small (8B parameters) and large (70B) models while reducing the amount of samples required by an order of magnitude. Our analysis of inference-time behavior demonstrates that our method produces LLMs that effectively leverage automatic feedback over multiple steps.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 23 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Reinforcement Learning with Verifiable Physics: Post-training LLMs with Continuous Rewards

    cs.LG 2026-07 conditional novelty 7.0 of 10

    RLVP post-trains one LLM across eight PDE families with hybrid validity-plus-continuous physics rewards, improving solver accuracy and enabling selective compositional transfer to held-out PDEs.

  2. An Iterative Test-and-Repair Framework for Competitive Code Generation

    cs.SE 2026-04 unverdicted novelty 7.0 of 10

    FixAudit improves LLM code generation on competitive programming benchmarks by training a shared model for iterative code-aware test generation and repair, achieving 35%+ gains in Pass@1 over baselines on the same 7B model.

  3. Training Language Models to Generate Quality Code with Program Analysis Feedback

    cs.CL 2025-05 conditional novelty 7.0 of 10

    REAL trains LLMs with rewards from program analysis and unit tests, improving joint functionality and code quality over prior methods.

  4. Beyond the Need for Speed: Energy-Aware Code Generation via Simulation-Guided Reinforcement Learning

    cs.LG 2026-07 conditional novelty 6.5 of 10

    Simulation-in-the-loop GRPO on energy-labeled C++ pairs yields 12.63% CARET and beats human energy efficiency on 58.4% of valid held-out outputs, while IPC misranks energy on 67.8% of problems.

  5. LEAP: Lean Environment-Feedback via Adaptive Pruning for Code RL in GPU Kernel Generation

    cs.LG 2026-08 conditional novelty 6.0 of 10

    LEAP combines difficulty-conditioned pruning with pairwise rank-based rewards to improve multi-turn reinforcement learning for CUDA kernel generation.

  6. RLPF: Reinforcement Learning from Performance Feedback for Code Generation

    cs.LG 2026-07 conditional novelty 6.0 of 10

    RLPF's staged performance reward lifts Qwen3-32B on PerfCodeBench from 11.1% to 54.6% correct-and-runnable and from 8.1% to 38.6% relative efficiency.

  7. Skill Self-Play: Pushing the Frontier of LLM Capability with Co-Evolving Skills

    cs.CL 2026-07 conditional novelty 6.0 of 10

    Skill Self-Play uses an evolving skill library to guide LLM self-training, improving tool-call and reasoning accuracy beyond unguided self-play on five model backbones.

  8. Training Large Language Models for Self-Explanation Faithfulness

    cs.LG 2026-07 conditional novelty 6.0 of 10

    RL fine-tuning with a counterfactual mention/influence reward raises LLM self-explanation faithfulness (Phi-CCT) from near zero to ~0.66 in-distribution for two 8B models, with partial transfer to held-out tasks.

  9. CSPF: A Constrained Shared-Private Fusion Method for Non-Verifiable Preference Evaluation

    cs.CL 2026-07 conditional novelty 6.0 of 10

    Fusing hidden states of multiple frozen reward models under shared-private constraints improves pairwise preference accuracy over scalar-score fusion and single-expert adapters on LM-Arena and PPE.

  10. NEURA: A Unified and Retargetable Compilation Framework for Coarse-Grained Reconfigurable Architectures

    cs.PL 2026-04 unverdicted novelty 6.0 of 10

    NEURA flattens CGRA control flow into a pure predicated dataflow IR and reports 2.20× kernel and up to 2.71× application speedups over high-performance SOTA baselines.

  11. Beyond Binary: Turning Partial Success into Dense Verifiable Rewards for Reinforcement Learning in Code Generation

    cs.LG 2026-01 conditional novelty 6.0 of 10

    VeRPO turns partial test-suite success into dense, difficulty-weighted rewards for RL code generation, beating GRPO by up to 8.8 pass@1 points at negligible cost.

  12. KernelEvolve: Scaling Agentic Kernel Coding for Heterogeneous AI Accelerators at Meta

    cs.LG 2025-12 conditional novelty 6.0 of 10

    An agentic kernel-coding system combining tree search with hardware-knowledge retrieval generated optimized Triton kernels for NVIDIA, AMD, and Meta's MTIA accelerators: 100% correctness on 480 operator-platform confi...

  13. VERIRL: Boosting the LLM-based Verilog Code Generation via Reinforcement Learning

    cs.LG 2025-08 unverdicted novelty 6.0 of 10

    VERIRL applies reinforcement learning with a trace-back rescore mechanism and sample-balanced weighting to achieve state-of-the-art Verilog code generation.

  14. Adaptive Guidance Accelerates Reinforcement Learning of Reasoning Models

    cs.LG 2025-06 reject novelty 6.0 of 10

    Guide, a new RL method that injects hints on total failure with off-policy correction, improves math pass@1 by a few points, while the paper argues RLVR progress is mostly self-distillation.

  15. LlamaRL: A Distributed Asynchronous Reinforcement Learning Framework for Efficient Large-scale LLM Training

    cs.LG 2025-05 conditional novelty 6.0 of 10

    LlamaRL is a distributed asynchronous RL framework that reports up to 10.7x faster RL post-training steps for 405B-parameter LLMs than a co-located synchronous baseline, using off-policy corrections and GPU-direct wei...

  16. Afterburner: Reinforcement Learning Facilitates Self-Improving Code Efficiency Optimization

    cs.SE 2025-05 conditional novelty 6.0 of 10

    Reinforcement learning with execution feedback enables a code model to iteratively improve the efficiency of its own generated code, surpassing supervised and preference-based training methods.

  17. Reinforcing General Reasoning without Verifiers

    cs.LG 2025-05 conditional novelty 6.0 of 10

    VeriFree trains LLMs with RL by maximizing the likelihood of the reference answer after generated reasoning, matching verifier-based RL without any verifier.

  18. What I cannot execute, I do not understand: Training and Evaluating LLMs on Program Execution Traces

    cs.LG 2025-02 conditional novelty 6.0 of 10

    Training an 8B LLM on Python execution traces improves output prediction on CruxEval and MBPP to about 80%, with dynamic scratchpads extending accurate predictions to executions of over ten thousand steps.

  19. PCSD: Persistent Consistency for Self-Distillation in Agentic Reinforcement Learning

    cs.AI 2026-08 conditional novelty 5.0 of 10

    PCSD derives token-level distillation weights from the persistence of teacher-favoring signals and improves ALFWorld agent success by 15.6 points over GRPO.

  20. Improving LLM-Generated Code Quality with GRPO

    cs.AI 2025-06 conditional novelty 5.0 of 10

    Adding a hand-built code-quality score to the GRPO reward makes a 3B model produce code that blind raters prefer, but evidence rests on one model and a synthetic benchmark.

  21. Adopting Reinforcement Learning with Verifiable Rewards for Molecular Generation

    cs.LG 2026-07 reject novelty 4.0 of 10

    LLMol fine-tunes an LLM on simplified SELFIES and uses GRPO with RDKit-derived rewards for targeted molecular generation, but its own benchmark tables contradict the claimed state-of-the-art performance.

  22. CodeGrad: Integrating Multi-Step Verification with Gradient-Based LLM Refinement

    cs.SE 2025-08 reject novelty 4.0 of 10

    CodeGrad improves code generation by routing structured critic feedback into up to two refinement rounds, but its formal-verification claim rests on AI-written proofs judged by the AI.

  23. Reinforcement learning fine-tuning of language model for instruction following and math reasoning

    cs.CL 2025-06 reject novelty 2.0 of 10

    RLOO with a DeBERTa reward model reaches a 0.695 win rate on instruction following, while best-of-N with an exact-match verifier lifts Countdown accuracy from 0.37 to 0.811.

Pith tools