REVIEW 20 cited by
StepCoder: Improve Code Generation with Reinforcement Learning from Compiler Feedback
Not yet reviewed by Pith; the record is open.
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
Signed reviews
read the original abstract
The advancement of large language models (LLMs) has significantly propelled the field of code generation. Previous work integrated reinforcement learning (RL) with compiler feedback for exploring the output space of LLMs to enhance code generation quality. However, the lengthy code generated by LLMs in response to complex human requirements makes RL exploration a challenge. Also, since the unit tests may not cover the complicated code, optimizing LLMs by using these unexecuted code snippets is ineffective. To tackle these challenges, we introduce StepCoder, a novel RL framework for code generation, consisting of two main components: CCCS addresses the exploration challenge by breaking the long sequences code generation task into a Curriculum of Code Completion Subtasks, while FGO only optimizes the model by masking the unexecuted code segments to provide Fine-Grained Optimization. In addition, we furthermore construct the APPS+ dataset for RL training, which is manually verified to ensure the correctness of unit tests. Experimental results show that our method improves the ability to explore the output space and outperforms state-of-the-art approaches in corresponding benchmarks. Our dataset APPS+ and StepCoder are available online.
Forward citations
Cited by 20 Pith papers
-
Outcome-Based RL Provably Leads Transformers to Reason, but Only With the Right Data
Outcome-based policy gradient provably learns step-by-step chain traversal in a single-layer Transformer only when the training distribution has non-vanishing mass on simple (few-step) examples.
-
Training Language Models to Generate Quality Code with Program Analysis Feedback
REAL trains LLMs with rewards from program analysis and unit tests, improving joint functionality and code quality over prior methods.
-
DHRCL:Training Code LLMs with Dense Hierarchical Rewards and Curriculum Learning
DHRCL stages syntax, execution, test-pass, and AST-similarity rewards during code-LLM RL and redistributes token credit by stage, yielding about a 1-point average Pass@1 gain across Qwen3 scales.
-
AR$^2$: Adversarial Reinforcement Learning for Abstract Reasoning in Large Language Models
Adversarial teacher-student RL on computationally equivalent narrative rewrites of coding problems yields modest but consistent pass@1 gains on competitive programming benchmarks.
-
Repair-R1: Better Test Before Repair
Training a code model to produce a failing test before the patch, with reinforcement learning, improves repair success on four Python benchmarks compared with vanilla and SFT baselines.
-
D-LiFT: Improving LLM-based Decompiler Backend via Code Quality-driven Fine-tuning
D-LIFT uses a composite quality score, D-Score, as a reinforcement-learning reward to fine-tune LLMs that rewrite decompiled code, claiming large gains in readability without losing accuracy.
-
Afterburner: Reinforcement Learning Facilitates Self-Improving Code Efficiency Optimization
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.
-
Integrating Symbolic Execution into the Fine-Tuning of Code-Generating LLMs
Adding symbolic-execution-generated tests to APPS improves reward model accuracy for code generation, but barely changes final code-generation quality relative to CodeRL.
-
ACECODER: Acing Coder RL via Automated Test-Case Synthesis
ACECODER shows that automatically synthesized test cases can serve as scalable reward signals for code RL, yielding consistent gains on HumanEval, MBPP, BigCodeBench, and LiveCodeBench.
-
Preference Optimization for Reasoning with Pseudo Feedback
Generating pseudo test cases from frontier models or self-consistency and using them as DPO preference signals improves 7B LLMs on math and code benchmarks, sometimes beating 72B models.
-
ParaStudent: Generating and Evaluating Realistic Student Code by Teaching LLMs to Struggle
Fine-tuning a 7B code model on real student submissions produces code that better matches student error patterns, style, and incremental revision trajectories than prompting-only models across two temporal resolutions.
-
Improving LLM-Generated Code Quality with GRPO
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.
-
CRScore++: Reinforcement Learning with Verifiable Tool and AI Feedback for Code Review
A two-stage SFT+DPO pipeline using linter and code-smell tool outputs plus an LLM judge improves generated code review comments and appears to transfer from Python to Java and JavaScript, though the judge-based evalua...
-
Themisto: Jupyter-Based Runtime Benchmark
A new benchmark for Jupyter runtime-aware code prediction shows five frontier LLMs get under 19% exact match on output prediction and under 14% on next cell prediction, with runtime context not helping.
-
Distilling Desired Comments for Enhanced Code Review with Large Language Models
Desiview identifies desired review comments from code review datasets using the perplexity difference of the actual fix with and without each comment, and the distilled data improves LLaMA-based code review models.
-
DSTC: Direct Preference Learning with Only Self-Generated Tests and Code to Improve Code LMs
DSTC builds preference pairs by selecting the hardest self-generated test that the best self-generated code passes, then fine-tunes code LMs with DPO or KTO to improve pass@1 accuracy.
-
SysTemp: A Multi-Agent System for Template-Based Generation of SysML v2
A multi-agent LLM pipeline with a rule-based template generator and parser feedback improves syntactic correctness of generated SysML v2 models in 4 of 5 test scenarios.
-
SPA-RL: Reinforcing LLM Agents via Stepwise Progress Attribution
SPA-RL attributes a single final task reward to individual agent steps and uses the attributions plus an executability signal as dense rewards for PPO, yielding modest success-rate gains on WebShop, ALFWorld, and VirtualHome.
-
CRPE: Expanding The Reasoning Capability of Large Language Model for Code Generation
A three-stage training pipeline (synthetic problems, multi-agent chain-of-thought data, step-DPO self-improvement) raises LiveCodeBench pass@1 for Qwen2.5-Coder bases to 21.88 (7B) and 35.08 (32B).
-
Trading Devil RL: Backdoor attack via Stock market, Bayesian Optimization and Reinforcement Learning
A data-poisoning backdoor attack on audio transformers is claimed with 100 percent success on TIMIT, but the paper provides no reproducible derivation or evaluation.
Discussion (0). Continue with ORCID to comment.