Data to 7 Oct 2026

Training Software Engineering Agents and Verifiers with SWE-Gym

arXiv 2412.21139 ↗First posted 30 Dec 2024cs.SE cs.CLcited by 300

On the map: LLM agents and tool use → Software engineering coding agents

Abstract (arXiv)

We present SWE-Gym, the first environment for training real-world software engineering (SWE) agents. SWE-Gym contains 2,438 real-world Python task instances, each comprising a codebase with an executable runtime environment, unit tests, and a task specified in natural language. We use SWE-Gym to train language model based SWE agents, achieving up to 19% absolute gains in resolve rate on the popular SWE-Bench Verified and Lite test sets. We also experiment with inference-time scaling through verifiers trained on agent trajectories sampled from SWE-Gym. When combined with our fine-tuned SWE agents, we achieve 32.0% and 26.0% on SWE-Bench Verified and Lite, respectively, reflecting a new state-of-the-art for open-weight SWE agents. To facilitate further research, we publicly release SWE-Gym, models, and agent trajectories.

Key facts. Cited by 300 papers in the map (data through 7 Oct 2026). Placed in Software engineering coding agents, one of 16 directions in LLM agents and tool use. Builds on 35 of its 51 references that are in the map. First posted on arXiv 30 Dec 2024.

Builds on

35 of 51 references.

Closest ancestors · 6

Same direction, newest first

Other references · 29

2022Training language models to follow instructions with human feedbackcited by 15,712
2017Proximal Policy Optimization Algorithmscited by 14,441
2021Training Verifiers to Solve Math Word Problemscited by 9,743
2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 8,851
2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,355
2021Evaluating Large Language Models Trained on Codecited by 8,053
2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,203
2023Let's Verify Step by Stepcited by 3,598
2022Competition-Level Code Generation with AlphaCodecited by 1,427
2024Qwen2.5-Coder Technical Reportcited by 1,151
2023SGLang: Efficient Execution of Structured Language Model Programscited by 1,107
2024Large Language Monkeys: Scaling Inference Compute with Repeated Samplingcited by 946

17 more

2023Math-Shepherd: Verify and Reinforce LLMs Step-by-step without Human Annotationscited by 895
2021Measuring Coding Challenge Competence With APPScited by 879
2017Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithmcited by 723
2019Solving Rubik's Cube with a Robot Handcited by 699
2023AgentTuning: Enabling Generalized Agent Abilities for LLMscited by 319
2024AutoCodeRover: Autonomous Program Improvementcited by 272
2023FireAct: Toward Language Agent Fine-tuningcited by 247
2024OpenCodeInterpreter: Integrating Code Generation with Execution and Refinementcited by 242
2024ArCHer: Training Language Model Agents via Hierarchical Multi-Turn RLcited by 212
2024DigiRL: Training In-The-Wild Device-Control Agents with Autonomous Reinforcement Learningcited by 210
2024Advancing LLM Reasoning Generalists with Preference Treescited by 191
2024Fine-Tuning Large Vision-Language Models as Decision-Making Agents via Reinforcement Learningcited by 158
2022Offline RL for Natural Language Generation with Implicit Language Q Learningcited by 137
2024Autonomous Evaluation and Refinement of Digital Agentscited by 136
2024AgentGym: Evolving Large Language Model-based Agents across Diverse Environmentscited by 120
2024DART-Math: Difficulty-Aware Rejection Tuning for Mathematical Problem-Solvingcited by 92
2024CodeR: Issue Resolving with Multi-Agent and Task Graphscited by 89

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 109; 2026: 191.

300 papers.

Sort:

By area:

2025SWE-smith: Scaling Data for Software Engineering Agentscited by 255
2025SWE-RL: Advancing LLM Reasoning via Reinforcement Learning on Open Software Evolutioncited by 246
2025UI-TARS-2 Technical Report: Advancing GUI Agent with Multi-Turn Reinforcement Learningcited by 205
2025Survey on Evaluation of LLM-based Agentscited by 190
2025A Survey of Reinforcement Learning for Large Reasoning Modelscited by 175
2025APIGen-MT: Agentic Pipeline for Multi-Turn Data Generation via Simulated Agent-Human Interplaycited by 158
2025From LLM Reasoning to Autonomous AI Agents: A Comprehensive Reviewcited by 146
2025R2E-Gym: Procedural Environments and Hybrid Verifiers for Scaling Open-Weights SWE Agentscited by 140
2025Scaling Long-Horizon LLM Agent via Context-Foldingcited by 136
2025SWE-rebench: An Automated Pipeline for Task Collection and Decontaminated Evaluation of Software Engineering Agentscited by 127
2026Agentic Harness Engineering: Observability-Driven Automatic Evolution of Coding-Agent Harnessescited by 119
2025SWEET-RL: Training Multi-Turn LLM Agents on Collaborative Reasoning Taskscited by 108

38 more

2025Large Language Models Often Know When They Are Being Evaluatedcited by 105
2025TRAIL: Trace Reasoning and Agentic Issue Localizationcited by 98
2025S*: Test Time Scaling for Code Generationcited by 95
2025Reinforcement Learning Optimization for Large-Scale Learning: An Efficient and User-Friendly Scaling Librarycited by 94
2025Agent KB: Leveraging Cross-Domain Experience for Agentic Problem Solvingcited by 90
2025CWM: An Open-Weights LLM for Research on Code Generation with World Modelscited by 87
2025Nemotron 3 Nano: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoningcited by 85
2026CaP-X: A Framework for Benchmarking and Improving Coding Agents for Robot Manipulationcited by 72
2025CyberGym: Evaluating AI Agents' Real-World Cybersecurity Capabilities at Scalecited by 71
2025SEC-bench: Automated Benchmarking of LLM Agents on Real-World Software Security Taskscited by 69
2025WebGen-Bench: Evaluating LLMs on Generating Interactive and Functional Websites from Scratchcited by 66
2025SWE-Exp: Experience-Driven Software Issue Resolutioncited by 64
2026Qwen3-Coder-Next Technical Reportcited by 60
2025Self-Challenging Language Model Agentscited by 56
2025SWE-Debate: Competitive Multi-Agent Debate for Software Issue Resolutioncited by 52
2025SWE-EVO: Benchmarking Coding Agents in Long-Horizon Software Evolution Scenarioscited by 49
2025The SWE-Bench Illusion: When State-of-the-Art LLMs Remember Instead of Reasoncited by 47
2025SkyRL-Agent: Efficient RL Training for Multi-turn LLM Agentcited by 44
2026Endless Terminals: Scaling RL Environments for Terminal Agentscited by 43
2025Training Proactive and Personalized LLM Agentscited by 42
2025Agent-RLVR: Training Software Engineering Agents via Guidance and Environment Rewardscited by 41
2025The Complexity Trap: Simple Observation Masking Is as Efficient as LLM Summarization for Agent Context Managementcited by 40
2025Repo2Run: Automated Building Executable Environment for Code Repository at Scalecited by 39
2025LitBench: A Benchmark and Dataset for Reliable Evaluation of Creative Writingcited by 39
2025MLE-Dojo: Interactive Environments for Empowering LLM Agents in Machine Learning Engineeringcited by 38
2025Large Language Models for Planning: A Comprehensive and Systematic Surveycited by 38
2025Context as a Tool: Context Management for Long-Horizon SWE-Agentscited by 38
2025Toward Training Superintelligent Software Agents through Self-Play SWE-RLcited by 37
2025Kimi-Dev: Agentless Training as Skill Prior for SWE-Agentscited by 35
2025Training Long-Context, Multi-Turn Software Engineering Agents with Reinforcement Learningcited by 34
2026Nemotron 3 Super: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoningcited by 33
2026Nemotron 3 Ultra: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoningcited by 32
2026SWE-Lego: Pushing the Limits of Supervised Fine-tuning for Software Issue Resolvingcited by 30
2026FeatureBench: Benchmarking Agentic Coding for Complex Feature Developmentcited by 30
2025BugPilot: Complex Bug Generation for Efficient Learning of SWE Skillscited by 28
2025SoRFT: Issue Resolving with Subtask-oriented Reinforced Fine-Tuningcited by 27
2025Why Are Web AI Agents More Vulnerable Than Standalone LLMs? A Security Analysiscited by 27
2025AutoPDL: Automatic Prompt Optimization for LLM Agentscited by 27

50 most cited of 300 All 300 (JSON)

Similar, not cited

Close in wording, no citation either way