Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2605.16299 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910244608671744 |
|---|---|
| author | Huang, Yixu Yu, Xinglei Wei, Zhongyu |
| author_facet | Huang, Yixu Yu, Xinglei Wei, Zhongyu |
| contents | Large Language Models (LLMs) excel at code generation but remain heavily reliant on large-scale annotated solutions and verification-based supervision, which constrains scalability and hinders sustained self-improvement. Recent solver--verifier frameworks exploit program execution as an automatic supervision signal, but their effectiveness degrades as solvers become moderately strong: verifier-generated tests increasingly confirm semantic correctness rather than exposing the remaining failure modes. We propose \textbf{ACE}, a self-evolving code generation framework based on a solver--adversary architecture that prioritizes active failure discovery through execution-centric supervision. A single LLM alternates between generating candidate programs and producing adversarial unit test inputs optimized to induce execution-level failures, such as runtime errors, exceptions, or non-termination. Supervision is derived solely from execution outcomes: robust programs are selected for supervised fine-tuning, while adversarial tests are optimized via Kahneman--Tversky Optimization using execution-derived preferences. Notably, the entire training loop requires no ground-truth code or external reward models. Experiments on CodeContests, MBPP, and LiveCodeBench demonstrate that ACE consistently outperforms strong solver--verifier baselines, achieving 3--7\% absolute gains in pass@1, with larger improvements on out-of-distribution benchmarks, while maintaining competitive or improved inference efficiency. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_16299 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | ACE: Self-Evolving LLM Coding Framework via Adversarial Unit Test Generation and Preference Optimization Huang, Yixu Yu, Xinglei Wei, Zhongyu Software Engineering Artificial Intelligence Large Language Models (LLMs) excel at code generation but remain heavily reliant on large-scale annotated solutions and verification-based supervision, which constrains scalability and hinders sustained self-improvement. Recent solver--verifier frameworks exploit program execution as an automatic supervision signal, but their effectiveness degrades as solvers become moderately strong: verifier-generated tests increasingly confirm semantic correctness rather than exposing the remaining failure modes. We propose \textbf{ACE}, a self-evolving code generation framework based on a solver--adversary architecture that prioritizes active failure discovery through execution-centric supervision. A single LLM alternates between generating candidate programs and producing adversarial unit test inputs optimized to induce execution-level failures, such as runtime errors, exceptions, or non-termination. Supervision is derived solely from execution outcomes: robust programs are selected for supervised fine-tuning, while adversarial tests are optimized via Kahneman--Tversky Optimization using execution-derived preferences. Notably, the entire training loop requires no ground-truth code or external reward models. Experiments on CodeContests, MBPP, and LiveCodeBench demonstrate that ACE consistently outperforms strong solver--verifier baselines, achieving 3--7\% absolute gains in pass@1, with larger improvements on out-of-distribution benchmarks, while maintaining competitive or improved inference efficiency. |
| title | ACE: Self-Evolving LLM Coding Framework via Adversarial Unit Test Generation and Preference Optimization |
| topic | Software Engineering Artificial Intelligence |
| url | https://arxiv.org/abs/2605.16299 |