SWE-Dev: Building Software Engineering Agents with Training and Inference Scaling (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) have advanced rapidly from conversational problem solving to addressing real-world tasks involving tool use, such as software engineering (SWE). |
| Approach: | They propose to build an LLM-based software engineering agent that synthesizes test cases and scales up agent trajectories to build training data. |
| Outcome: | The proposed model outperforms state-of-the-art models on the SWE-bench-Verified benchmark. |
Similar Papers
SWE-MERA: A Dynamic Benchmark for Agenticly Evaluating Large Language Models on Software Engineering Tasks (2025.emnlp-demos)
Copied to clipboard
Adamenko Pavel, Ivanov Mikhail, Aidar Valeev, Rodion Levichev, Pavel Zadorozhny, Ivan Lopatin, Dmitrii Babaev, Alena Fenogenova, Valentin Malykh
| Challenge: | SWE-bench is a static benchmark that collects only once and never updates. |
| Approach: | They propose a dynamic, continuously updated benchmark to address data contamination issues by collecting real-world GitHub issues and rigorous quality validation. |
| Outcome: | The proposed benchmarks are based on a dataset of 2,294 GitHub issues and their corresponding pull requests (PRs) the static nature of the benchmarks makes it hard to distinguish meaningful progress. |
SWE-QA-Pro: A Representative Benchmark and Scalable Training Recipe for Repository-Level Code Understanding (2026.findings-acl)
Copied to clipboard
Songcheng Cai, Zhiheng Lyu, Yuansheng Ni, Xiangchao Chen, Baichuan Zhou, Shenzhe Zhu, Yi Lu, Haozhe Wang, Chi Ruan, Benjamin Schneider, Weixu Zhang, Xiang Li, Andy Zheng, Yuyu Zhang, Ping Nie, Wenhu Chen
| Challenge: | Existing benchmarks for agentic repository-level code understanding overlook long tail topics and rely on memorized knowledge. |
| Approach: | They propose a repository-level agentic code understanding benchmark that uses long-tail repositories with executable environments to enforce topical balance. |
| Outcome: | Empirically, a Qwen3-8B model trained with the proposed benchmark outperforms GPT-4o by 2.3 points. |
AgentGym: Evaluating and Training Large Language Model-based Agents across Diverse Environments (2025.acl-long)
Copied to clipboard
Zhiheng Xi, Yiwen Ding, Wenxiang Chen, Boyang Hong, Honglin Guo, Junzhe Wang, Xin Guo, Dingwen Yang, Chenyang Liao, Wei He, Songyang Gao, Lu Chen, Rui Zheng, Yicheng Zou, Tao Gui, Qi Zhang, Xipeng Qiu, Xuanjing Huang, Zuxuan Wu, Yu-Gang Jiang
| Challenge: | Large language models (LLMs) are promising foundations to build generally-capable agents . however, the community lacks a unified interactive framework that covers diverse environments for comprehensive evaluation of agents. |
| Approach: | They propose a framework that features 7 real-world scenarios, 14 environments, and 89 tasks for unified, real-time, and concurrent agent interaction. |
| Outcome: | The proposed framework features 7 real-world scenarios, 14 environments, and 89 tasks for unified, real-time, and concurrent agent interaction. |
SWE-Mutation: Can LLMs Generate Reliable Test Suites in Software Engineering? (2026.findings-acl)
Copied to clipboard
Yuxuan Sun, Yuze Zhao, Yufeng Wang, Yao Du, Zhiyuan Ma, Jinbo Wang, Mengdi Zhang, Kai Zhang, Zhenya Huang
| Challenge: | Evaluating software engineering capabilities is a core component of large language models (LLMs). |
| Approach: | They propose a benchmark to evaluate LLM-generated test suites that introduces mutated solutions that attempt to "fool" them. |
| Outcome: | The proposed test suites are based on 2,636 mutated variants derived from 800 original instances and include a multilingual subset spanning nine programming languages. |
Agent-FLAN: Designing Data and Methods of Effective Agent Tuning for Large Language Models (2024.findings-acl)
Copied to clipboard
| Challenge: | Existing studies focus on prompt engineering or framework scheduling of one/multiple LLMs. |
| Approach: | They propose to integrate LLMs as agents into their training corpus by decomposition and redesigning the training corpu . they propose to use LLM-FLAN to effectively fine-tune LANguage models for Agents by reducing hallucinations. |
| Outcome: | The proposed model outperforms prior best models by 3.5% across agent evaluation datasets. |
SWE-AGILE: A Software Agent Framework for Efficiently Managing Dynamic Reasoning Context (2026.findings-acl)
Copied to clipboard
| Challenge: | representative ReAct-style approaches lack explicit System-2 reasoning for deep analysis and handling complex edge cases. |
| Approach: | They propose a software agent framework that preserves full reasoning history while compressing historical reasoning content into concise Reasoning Digests. |
| Outcome: | Empirically, the proposed framework sets a new standard for 7B-8B models on SWE-Bench-Verified using only 2.2k trajectories and 896 tasks. |
SWE-Fixer: Training Open-Source LLMs for Effective and Efficient GitHub Issue Resolution (2025.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models excel in code generation benchmarks, but these benchmarks focus on single-file scenarios with constrained context scope. |
| Approach: | They propose an open-source framework to effectively resolve GitHub issues using a code file retrieval module and a model-based code editing module. |
| Outcome: | The proposed approach achieves state-of-the-art performance on two GitHub benchmarks. |
UTBoost: Rigorous Evaluation of Coding Agents on SWE-Bench (2025.acl-long)
Copied to clipboard
| Challenge: | Large language models have enabled the development of coding agents for real-world code generation. |
| Approach: | They propose a novel LLM-driven test case generator that analyzes codebases and dependencies to generate test cases for real-world Python projects. |
| Outcome: | The proposed framework improves the performance of SWE-Bench by analyzing codebases and dependencies. |
SWE-Swiss: A Multi-Task Fine-Tuning and RL Recipe for High-Performance Issue Resolution (2026.findings-acl)
Copied to clipboard
Zhenyu He, Qingping Yang, Wei Shen, Xiaojian Zhong, Kechi Zhang, Chenxin An, Wenlei Shi, Tianle Cai, Di He, Jiaze Chen, Jingjing Xu
| Challenge: | SWE-Swiss-32B demonstrates strong generalization to other common LLM benchmarks. |
| Approach: | They propose a two-phase training recipe that decomposes issue resolution into three core skills: Localization, Repair, and Unit Test Generation. |
| Outcome: | The proposed model achieves a 60.2% score on the SWE-bench Verified benchmark and is in the top-tier performance bracket of much larger models. |
Agent Lumos: Unified and Modular Training for Open-Source Language Agents (2024.acl-long)
Copied to clipboard
Da Yin, Faeze Brahman, Abhilasha Ravichander, Khyathi Chandu, Kai-Wei Chang, Yejin Choi, Bill Yuchen Lin
| Challenge: | Lumos is a framework for training open-source agents on complex interactive tasks. |
| Approach: | They propose a framework for training open-source LLM-based agents called Lumos . Lumos features a learnable, unified and modular architecture with a planning module that learns high-level subgoal generation and a grounding module trained to translate these into the actions using various tools in the execution module. |
| Outcome: | The framework outperforms open-source agents on QA and web tasks. |