REVIEW 3 major objections 5 minor 67 references
SWE-Bench ProMax: Benchmarking Agents on Large-Scale Multilingual Code Refactoring
T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash
Pith's one-line read A new benchmark gives coding agents large-scale refactoring tasks across seven languages, and the best agent solves only 41.2% of them.
desk verdict A harder, multilingual refactoring benchmark with a credible empirical story, but the curation claims need an audit trail before the 41.2% headline is interpretable. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The central mechanism is the three-stage curation pipeline: automated mining of refactoring commits from high-star GitHub repositories, environment construction with Docker validation that the gold patch passes its own test suite, and expert-driven filtering and rewriting. The expert stage is the load-bearing part: commit analysis, quality filtering to remove single-file or low-complexity tasks and to delete overly narrow and overly broad tests, and rewriting of every issue description from scratch so that each description is a necessary and sufficient condition for the gold patch.
What would settle it
A concrete check is to independently audit the 170 instances with fresh annotators following the same protocol and measure the agreement rate on (a) which tests are overly narrow or overly broad, and (b) whether each rewritten description is a necessary and sufficient specification of the gold patch. A second concrete check is to run a simple retrieval baseline that copies the gold patch verbatim from the public commit history: if that baseline passes the test suite on a large fraction of instances, the benchmark is contaminated rather than unsaturated.
Extended reading notes
Core claim
The paper claims that large-scale multilingual code refactoring is a distinct, currently underserved benchmark domain, and that SWE-Bench ProMax is a valid instantiation of it: 170 expert-curated instances from real refactoring commits, spanning Python, Java, TypeScript, Go, C, C++, and Rust, with an average gold patch of 11.4 files and 261.6 lines of code. The benchmark's evaluation quality is the load-bearing claim: every problem statement is rewritten to be a necessary and sufficient specification of the gold patch, and test suites are manually pruned of overly narrow tests that reject valid solutions and overly broad tests that check unstated requirements. On this benchmark, under two agent scaffolds, the best model reaches a 41.2% resolve rate, and the dominant failure mode is incomplete refactoring—agents modify fewer files than the gold patch requires and consume more interaction rounds when they fail.
Load-bearing premise
The manual expert review that prunes bad tests and rewrites problem statements is assumed to be complete and unbiased, yet no inter-annotator agreement, second audit, or post-release test-suite changes are provided to back it up.
Editorial extensions
If this is right
- If the curation claim holds, SWE-Bench ProMax offers an execution-based evaluation where passing tests means the refactoring was applied correctly, in contrast to prior refactoring studies that used code-smell counts, compilability, or alignment scores.
- The 41.2% ceiling means the benchmark is not saturated, so it can separate future agent improvements that existing benchmarks cannot.
- The failure-mode analysis implies that improving cross-file coordination—not reasoning per se—is the critical bottleneck, giving a concrete target for agent design.
- The cost analysis implies that open-weight models can approach frontier resolve rates at a small fraction of the API cost, so affordability alone does not explain frontier performance gaps.
- The multilingual design allows per-language comparison, so it can reveal how type systems, memory models, and build ecosystems affect agent behavior.
Reading between the lines
- If the benchmark is adopted widely, the label 'SWE-bench ProMax' could come to mean 'instance size in files and LOC' as the proxy for difficulty, which may push future benchmarks toward even larger patches without verifying that the added size is behaviorally relevant.
- The finding that agents under-modify files suggests a testable extension: measuring whether explicit file-coverage planning, such as asking the agent to enumerate all call sites before editing, improves resolve rate without changing model weights.
- The curation pipeline could be applied as a filter to existing bug-fix benchmarks, providing a cheap way to re-estimate how much of the reported 'saturation' on SWE-bench Verified is a test-quality artifact rather than genuine model capability.
- Because the gold patch can be reproduced verbatim from public commit data, the benchmark's resistance to contamination may degrade over time unless the dataset or evaluation adds dynamic holdout mechanisms.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper introduces SWE-Bench ProMax, a benchmark of 170 code-refactoring instances mined from real commits across seven programming languages and 70 repositories. Each instance ships with a Docker environment at the pre-refactoring commit, a rewritten problem statement, a manually reviewed test suite, and the gold patch. The authors claim a three-stage curation pipeline that removes overly narrow and overly broad tests and verifies each description as a necessary and sufficient specification of the gold patch, so an instance is resolved if and only if all tests pass. They evaluate six frontier models under two agent scaffolds, reporting a best resolve rate of 41.2%, per-language rates, cost and step statistics, and a trajectory analysis identifying incomplete cross-file refactoring as the dominant failure mode. The dataset is released publicly on Hugging Face.
Significance. If the curation claims hold, the benchmark fills a genuine gap: it targets large-scale multilingual refactoring (averaging 11.4 source files and 261.6 LOC per instance), a domain that existing benchmarks cover only in single-language or small-scale settings. The public dataset with per-instance Docker environments and the comparison of proprietary versus open-weight models are useful empirical contributions. The finding that open-weight models approach proprietary performance at a fraction of the cost and the analysis of incomplete refactoring as the main failure mode are informative. However, the significance is conditional on the transparency of the curation process and the reproducibility of the evaluation, both of which are currently incomplete.
major comments (3)
- [Section 3.2, Stage 3 (steps 3 and 4)] The paper's central evaluation-quality claim is asserted but not evidenced. Section 3.2 states that each rewritten description is "verified to serve as both a necessary and sufficient condition for the gold patch" and that "the tests must pass if and only if the refactoring is correctly applied." No protocol, inter-annotator agreement, list of removed tests, or second audit is provided, and the released artifact appears to include only the dataset, not the evaluation harness or a commit hash for the evaluation code. Because the headline 41.2% resolve rate is meaningful only if passing tests truly indicates correct refactoring, the authors should either release a transparent audit trail (e.g., annotated test-suite changes, independent verification records) or substantially weaken the necessary-and-sufficient wording, and they should provide a reproducible evaluation harness with pinned versions.
- [Appendix B, Figure 6; abstract and Section 1] The paper presents SWE-Bench ProMax as a behavior-preserving refactoring benchmark (abstract: "coordinated, behavior-preserving changes"; Section 1: "restructuring existing code without changing its external behavior"), yet the authors' own multi-label classification in Figure 6 labels 41.2% of instances as Bug Fix and 43.5% as New Feature, and the text states that every instance involves at least two categories. This is not a formal contradiction, because a commit can fix a latent bug or enable a feature without its message saying "bug fix," but it makes the evaluation claim concrete: if tests assert new or changed behavior, then resolving an instance may require implementing behaviors beyond a pure refactoring, and the resolve rate measures commit reconstruction rather than refactoring ability. The authors should either restrict the benchmark to genuinely behavior-preserving commits or explicitly reframe the contribution as general repository-level code maintenance, and the claims about refactoring-specific difficulty and failure modes in Sections 1 and 5.2 should be adjusted accordingly.
- [Section 5.1 and Table 3] All experimental results are single runs with no variance estimates. With 170 instances, differences such as GLM-5 at 36.5% versus GPT-5.2 at 41.2% (a difference of 8 instances) are within plausible sampling noise, yet the paper draws conclusions from these small gaps, including the cost-efficiency comparison in Section 5.3. The authors should report multiple runs (or bootstrap confidence intervals) for each model-scaffold combination and release the exact model checkpoints, scaffold versions, and evaluation harness commit so that the results are reproducible and the comparisons are statistically grounded.
minor comments (5)
- [Section 3.2, Stage 1] Stage 1 filters for commit messages containing "refactor" and not "bug fix," but Appendix B classifies 41.2% of instances as Bug Fix. Please add a sentence explaining that this reflects bugs fixed without the phrase "bug fix" in the message, or adjust the filter description to avoid apparent inconsistency.
- [Appendix B, Figures 6–8] The label "Documentation Nl" appears to be a typo for "Documentation"; the same stray "Nl" may appear in the co-occurrence figures. Please correct the labels.
- [Section 3.2, Stage 3(3)] The phrase "necessary and sufficient condition for the gold patch" is stronger than manual review can establish. Consider wording such as "manually verified for mutual consistency" unless a formal verification protocol is supplied.
- [Section 4.1] No URL or version identifier is given for mini-swe-agent, and the exact versions of the evaluated models are not listed. Please include these, and state how many instances hit the 300-step or $10 cost caps, since the trajectory analysis in Section 5.2 depends on distinguishing unproductive exploration from budget truncation.
- [Section 3.3 and Table 2] Table 2 reports "Gold Patch (source)" average files, while Table 5 reports "Avg. #Non-test" files; please align the terminology so the two tables are directly comparable.
Circularity Check
No significant circularity: the benchmark's resolve-rate claim is a direct empirical measurement on a curated test suite, not a quantity fitted from or defined by the models being evaluated.
full rationale
The paper's central claim is that SWE-Bench ProMax is an unsaturated, expert-curated refactoring benchmark, supported by the measured 41.2% best resolve rate. This is an empirical measurement against a fixed test suite, not a derivation from fitted parameters, and the resolve-rate definition ('pass every test in the suite') is an evaluation protocol, not a circular construction. The curation stage rewrites problem statements with knowledge of the gold patch and manually reviews tests for over-narrowness and over-broadness; this is standard benchmark construction and, while it carries quality-assurance risk, it does not make the difficulty result equivalent to its inputs. The paper's many self-citations appear in background and related-work discussions and are not load-bearing for the benchmark's evaluation or results. The only substantive concerns—unverified inter-annotator agreement, possible category overlap with 'bug fix' despite the commit-message filter, and reliance on manual verification of 'necessary and sufficient' descriptions—are correctness or validation risks, not circular reasoning. No equation or fitted value is renamed as a prediction, and no uniqueness theorem or prior self-derived constraint is used to force the reported outcome.
Assumptions & free parameters
assumptions (4)
- domain assumption Commit messages containing 'refactor' and not 'bug fix', from repositories with at least 500 stars and a primary language comprising at least 80% of the codebase, identify genuine refactoring commits.
- domain assumption The manually reviewed test suites, after removal of overly narrow and overly broad tests, pass if and only if the refactoring is correctly applied.
- domain assumption Docker environments built with SWE-Factory faithfully reproduce the pre-refactoring repository state with dependencies installed.
- domain assumption The gold patch from the original developer is the ground truth for the intended refactoring.
Cite this review
Pith. "Pith review of SWE-Bench ProMax: Benchmarking Agents on Large-Scale Multilingual Code Refactoring." pith.science (2026). https://pith.science/paper/CXIIWTYT
@misc{pith2026260809802,
author = {Pith},
title = {Pith review of: SWE-Bench ProMax: Benchmarking Agents on Large-Scale Multilingual Code Refactoring},
year = {2026},
howpublished = {\url{https://pith.science/paper/CXIIWTYT}},
note = {Machine review of arXiv:2608.09802}
}
read the original abstract
As AI coding agents take on increasingly complex, long-horizon software engineering tasks, existing benchmarks are rapidly saturating and their evaluation quality has come under serious scrutiny: a recent audit found that nearly 60% of unsolved SWE-bench Verified instances contain flawed tests -- either overly narrow tests that reject correct solutions or overly broad tests that check unstated requirements -- and that frontier models can verbatim reproduce gold patches from training data. Code refactoring, which requires coordinated, behavior-preserving changes across many files, offers a substantially harder and more realistic test of agent capability, yet remains underserved by current benchmarks. We introduce SWE-Bench ProMax, an expert-curated, multilingual code refactoring benchmark of 170 instances drawn from real commits across seven programming languages (Python, Java, TypeScript, Go, C, C++, and Rust). Every instance undergoes rigorous, multi-stage curation that directly addresses the quality problems identified in prior benchmarks: issue descriptions are rewritten from scratch to provide precise, unambiguous specifications, and test suites are manually reviewed to remove overly narrow and overly broad tests. Tasks with insufficient complexity or limited cross-file scope are filtered out, yielding a benchmark of challenging, large-scale refactoring tasks that average 11.4 modified files and 261.6 lines of code per instance, substantially exceeding the scale of existing benchmarks. Experiments with frontier models under two agent scaffolds show that the best model achieves only 41.2% resolve rate, confirming that SWE-Bench ProMax presents a meaningful and unsaturated challenge for current AI coding agents. Our benchmark is available at https://huggingface.co/datasets/swe-bench-promax/SWE-Bench-ProMax.
Reference graph
Works this paper leans on
-
[1]
Advances in neural information processing systems , volume=
Attention is all you need , author=. Advances in neural information processing systems , volume=
-
[2]
arXiv preprint arXiv:2107.03374 , year=
Evaluating large language models trained on code , author=. arXiv preprint arXiv:2107.03374 , year=
-
[3]
arXiv preprint arXiv:2105.09938 , year=
Measuring coding challenge competence with apps , author=. arXiv preprint arXiv:2105.09938 , year=
-
[4]
arXiv preprint arXiv:2108.07732 , year=
Program synthesis with large language models , author=. arXiv preprint arXiv:2108.07732 , year=
-
[5]
arXiv preprint arXiv:2403.07974 , year=
Livecodebench: Holistic and contamination free evaluation of large language models for code , author=. arXiv preprint arXiv:2403.07974 , year=
-
[6]
International Conference on Learning Representations , volume=
Swe-bench: Can language models resolve real-world github issues? , author=. International Conference on Learning Representations , volume=
-
[7]
arXiv preprint arXiv:2410.03859 , year=
Swe-bench multimodal: Do ai systems generalize to visual software domains? , author=. arXiv preprint arXiv:2410.03859 , year=
-
[8]
arXiv preprint arXiv:2509.16941 , year=
Swe-bench pro: Can ai agents solve long-horizon software engineering tasks? , author=. arXiv preprint arXiv:2509.16941 , year=
Show all 67 references
-
[9]
arXiv preprint arXiv:2512.18470 , year=
SWE-EVO: Benchmarking coding agents in long-horizon software evolution scenarios , author=. arXiv preprint arXiv:2512.18470 , year=
-
[10]
arXiv preprint arXiv:2505.20411 , year=
Swe-rebench: An automated pipeline for task collection and decontaminated evaluation of software engineering agents , author=. arXiv preprint arXiv:2505.20411 , year=
-
[11]
Advances in Neural Information Processing Systems , volume=
Swe-bench goes live! , author=. Advances in Neural Information Processing Systems , volume=
-
[12]
Advances in Neural Information Processing Systems , volume=
Multi-swe-bench: A multilingual benchmark for issue resolving , author=. Advances in Neural Information Processing Systems , volume=
-
[13]
arXiv preprint arXiv:2506.10954 , year=
Swe-factory: Your automated factory for issue resolution training data and evaluation benchmarks , author=. arXiv preprint arXiv:2506.10954 , year=
-
[14]
Findings of the Association for Computational Linguistics: ACL 2026 , pages=
Multi-Docker-Eval: A ‘Shovel of the Gold Rush’Benchmark on Automatic Environment Building for Software Engineering , author=. Findings of the Association for Computational Linguistics: ACL 2026 , pages=
2026
-
[15]
arXiv preprint arXiv:2601.11868 , year=
Terminal-bench: Benchmarking agents on hard, realistic tasks in command line interfaces , author=. arXiv preprint arXiv:2601.11868 , year=
-
[16]
arXiv preprint arXiv:2503.07832 , year=
Refactorbench: Evaluating stateful reasoning in language agents through code , author=. arXiv preprint arXiv:2503.07832 , year=
-
[17]
arXiv preprint arXiv:2602.03712 , year=
SWE-Refactor: A Repository-Level Benchmark for Real-World LLM-Based Code Refactoring , author=. arXiv preprint arXiv:2602.03712 , year=
-
[18]
arXiv preprint arXiv:2511.04824 , year=
Agentic Refactoring: An Empirical Study of AI Coding Agents , author=. arXiv preprint arXiv:2511.04824 , year=
-
[19]
IEEE Transactions on Software Engineering , volume=
An automated approach to discovering software refactorings by comparing successive versions , author=. IEEE Transactions on Software Engineering , volume=. 2025 , publisher=
2025
-
[20]
IEEE Transactions on Software Engineering , year=
An Empirical Study of Software Refactorings in Real-World Open-Source Java Projects , author=. IEEE Transactions on Software Engineering , year=
-
[21]
Advances in neural information processing systems , volume=
Judging llm-as-a-judge with mt-bench and chatbot arena , author=. Advances in neural information processing systems , volume=
-
[22]
arXiv preprint arXiv:2403.04132 , year=
Chatbot arena: An open platform for evaluating llms by human preference , author=. arXiv preprint arXiv:2403.04132 , year=
-
[23]
arXiv preprint arXiv:2410.12784 , year=
Judgebench: A benchmark for evaluating llm-based judges , author=. arXiv preprint arXiv:2410.12784 , year=
-
[24]
arXiv preprint arXiv:2502.16614 , year=
Codecriticbench: A holistic code critique benchmark for large language models , author=. arXiv preprint arXiv:2502.16614 , year=
-
[25]
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=
Codejudgebench: Benchmarking llm-as-a-judge for coding tasks , author=. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=
-
[26]
Proceedings of the 31st International Conference on Computational Linguistics , pages=
CodeJudge-eval: Can large language models be good judges in code understanding? , author=. Proceedings of the 31st International Conference on Computational Linguistics , pages=
-
[27]
2025 , howpublished =
2025
-
[28]
2025 , howpublished =
Introducing Upgrades to. 2025 , howpublished =
2025
-
[29]
2026 , howpublished =
Why. 2026 , howpublished =
2026
-
[30]
How We Compare Model Quality at
Naman Jain , year =. How We Compare Model Quality at
-
[31]
arXiv preprint arXiv:2503.14499 , volume=
Measuring ai ability to complete long tasks , author=. arXiv preprint arXiv:2503.14499 , volume=. 2025 , publisher=
2025 arXiv
-
[32]
Findings of the Association for Computational Linguistics: ACL 2026 , pages=
Longcli-bench: A preliminary benchmark and study for long-horizon agentic programming in command-line interfaces , author=. Findings of the Association for Computational Linguistics: ACL 2026 , pages=
2026
-
[33]
arXiv preprint arXiv:2603.00729 , year=
Qwen3-coder-next technical report , author=. arXiv preprint arXiv:2603.00729 , year=
-
[34]
arXiv preprint arXiv:2406.11931 , year=
Deepseek-coder-v2: Breaking the barrier of closed-source models in code intelligence , author=. arXiv preprint arXiv:2406.11931 , year=
-
[35]
arXiv preprint arXiv:2602.15763 , year=
Glm-5: from vibe coding to agentic engineering , author=. arXiv preprint arXiv:2602.15763 , year=
-
[36]
arXiv preprint arXiv:2507.20534 , year=
Kimi k2: Open agentic intelligence , author=. arXiv preprint arXiv:2507.20534 , year=
-
[37]
arXiv preprint arXiv:2601.03267 , year=
Openai gpt-5 system card , author=. arXiv preprint arXiv:2601.03267 , year=
-
[38]
2025 , howpublished =
Gemini 3 Pro , author =. 2025 , howpublished =
2025
-
[39]
2026 , howpublished =
Claude Sonnet 4.6 , author =. 2026 , howpublished =
2026
-
[40]
5: Visual Agentic Intelligence , author=
Kimi K2. 5: Visual Agentic Intelligence , author=. arXiv preprint arXiv:2602.02276 , year=
-
[41]
The Thirty-eighth Annual Conference on Neural Information Processing Systems , year=
Swe-agent: Agent-computer interfaces enable automated software engineering , author=. The Thirty-eighth Annual Conference on Neural Information Processing Systems , year=
-
[42]
International Conference on Learning Representations , volume=
Openhands: An open platform for ai software developers as generalist agents , author=. International Conference on Learning Representations , volume=
-
[43]
arXiv preprint arXiv:2407.01489 , year=
Agentless: Demystifying llm-based software engineering agents , author=. arXiv preprint arXiv:2407.01489 , year=
-
[44]
Forty-third International Conference on Machine Learning , year=
CodeTaste: Can LLMs Generate Human-Level Code Refactorings? , author=. Forty-third International Conference on Machine Learning , year=
-
[45]
ACM Transactions on Software Engineering and Methodology , year=
An empirical study on the code refactoring capability of large language models , author=. ACM Transactions on Software Engineering and Methodology , year=
-
[46]
arXiv preprint arXiv:2511.21788 , year=
Code Refactoring with LLM: A Comprehensive Evaluation With Few-Shot Settings , author=. arXiv preprint arXiv:2511.21788 , year=
-
[47]
IEEE Transactions on Software Engineering , volume=
RefactoringMiner 2.0 , author=. IEEE Transactions on Software Engineering , volume=. 2020 , publisher=
2020
-
[48]
arXiv preprint arXiv:2504.08703 , year=
Swe-polybench: A multi-language benchmark for repository level evaluation of coding agents , author=. arXiv preprint arXiv:2504.08703 , year=
-
[49]
arXiv preprint arXiv:2508.05988 , year=
Pruning the unsurprising: Efficient code reasoning via first-token surprisal , author=. arXiv preprint arXiv:2508.05988 , year=
-
[50]
Proceedings of the ACM on Software Engineering , volume=
In line with context: Repository-level code generation via context inlining , author=. Proceedings of the ACM on Software Engineering , volume=. 2026 , publisher=
2026
-
[51]
arXiv preprint arXiv:2601.05110 , year=
Glimprouter: Efficient collaborative inference by glimpsing one token of thoughts , author=. arXiv preprint arXiv:2601.05110 , year=
-
[52]
arXiv preprint arXiv:2606.28436 , year=
Dockerless: Environment-Free Program Verifier for Coding Agents , author=. arXiv preprint arXiv:2606.28436 , year=
-
[53]
arXiv preprint arXiv:2606.28434 , year=
SWE-MeM: Learning Adaptive Memory Management for Long-Horizon Coding Agents , author=. arXiv preprint arXiv:2606.28434 , year=
-
[54]
2026 56th Annual IEEE International Conference on Dependable Systems and Networks (DSN) , pages=
Zero-shot vulnerability detection in low-resource smart contracts through solidity-only training , author=. 2026 56th Annual IEEE International Conference on Dependable Systems and Networks (DSN) , pages=. 2026 , organization=
2026
-
[55]
, author=
FlowMalTrans: Unsupervised Binary Code Translation for Malware Detection Using Flow-Adapter Architecture. , author=. EMNLP (Findings) , pages=
-
[56]
arXiv preprint arXiv:2607.07946 , year=
DeepSWE: Measuring Frontier Coding Agents on Original, Long-Horizon Engineering Tasks , author=. arXiv preprint arXiv:2607.07946 , year=
-
[57]
arXiv preprint arXiv:2602.09892 , year=
Immersion in the github universe: Scaling coding agents to mastery , author=. arXiv preprint arXiv:2602.09892 , year=
-
[58]
arXiv preprint arXiv:2603.13023 , year=
davinci-env: Open swe environment synthesis at scale , author=. arXiv preprint arXiv:2603.13023 , year=
-
[59]
arXiv preprint arXiv:2606.07682 , year=
SWE-Marathon: Can Agents Autonomously Complete Ultra-Long-Horizon Software Work? , author=. arXiv preprint arXiv:2606.07682 , year=
-
[60]
2025 IEEE/ACM 47th International Conference on Software Engineering (ICSE 2025) , year=
Between lines of code: Unraveling the distinct patterns of machine and human programmers , author=. 2025 IEEE/ACM 47th International Conference on Software Engineering (ICSE 2025) , year=
2025
-
[61]
2026 IEEE/ACM 48th International Conference on Software Engineering (ICSE 2026) , year=
From code to correctness: Closing the last mile of code generation with hierarchical debugging , author=. 2026 IEEE/ACM 48th International Conference on Software Engineering (ICSE 2026) , year=
2026
-
[62]
arXiv preprint arXiv:2509.14635 , year=
SWE-QA: Can Language Models Answer Repository-level Code Questions? , author=. arXiv preprint arXiv:2509.14635 , year=
-
[63]
2025 IEEE/ACM 40th International Conference on Automated Software Engineering (ASE) , year=
LongCodeZip: Compress Long Context for Code Language Models , author=. 2025 IEEE/ACM 40th International Conference on Automated Software Engineering (ASE) , year=
2025
-
[64]
arXiv preprint arXiv:2602.01785 , year=
CodeOCR: On the Effectiveness of Vision Language Models in Code Understanding , author=. arXiv preprint arXiv:2602.01785 , year=
-
[65]
arXiv preprint arXiv:2507.23361 , year=
SWE-Exp: Experience-driven software issue resolution , author=. arXiv preprint arXiv:2507.23361 , year=
-
[66]
arXiv preprint arXiv:2601.16746 , year=
Swe-pruner: Self-adaptive context pruning for coding agents , author=. arXiv preprint arXiv:2601.16746 , year=
-
[67]
arXiv preprint arXiv:2507.23348 , year=
Swe-debate: Competitive multi-agent debate for software issue resolution , author=. arXiv preprint arXiv:2507.23348 , year=
Reviewed August 11, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.