Code-A1: Adversarial Evolving of Code LLM and Test LLM via Reinforcement Learning
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914398830854144 |
|---|---|
| author | Wang, Aozhe Yan, Yuchen Zhou, Nan Lu, Zhengxi Lu, Weiming Xiao, Jun Zhuang, Yueting Shen, Yongliang |
| author_facet | Wang, Aozhe Yan, Yuchen Zhou, Nan Lu, Zhengxi Lu, Weiming Xiao, Jun Zhuang, Yueting Shen, Yongliang |
| contents | Reinforcement learning for code generation relies on verifiable rewards from unit test pass rates. Yet high-quality test suites are scarce, existing datasets offer limited coverage, and static rewards fail to adapt as models improve. Recent self-play methods unify code and test generation in a single model, but face a inherent dilemma: white-box access leads to self-collusion where the model produces trivial tests for easy rewards, yet black-box restriction yields generic tests that miss implementation-specific bugs. We introduce Code-A1, an adversarial co-evolution framework that jointly optimizes a Code LLM and a Test LLM with opposing objectives. The Code LLM is rewarded for passing more tests, while the Test LLM is rewarded for exposing more defects. This architectural separation eliminates self-collusion risks and safely enables white-box test generation, where the Test LLM can inspect candidate code to craft targeted adversarial tests. We further introduce a Mistake Book mechanism for experience replay and a composite reward balancing test validity with adversarial difficulty. Experiments on Qwen2.5-Coder models demonstrate that Code-A1 achieves code generation performance matching or exceeding models trained on human-annotated tests, while significantly improving test generation capability. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_15611 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Code-A1: Adversarial Evolving of Code LLM and Test LLM via Reinforcement Learning Wang, Aozhe Yan, Yuchen Zhou, Nan Lu, Zhengxi Lu, Weiming Xiao, Jun Zhuang, Yueting Shen, Yongliang Computation and Language Reinforcement learning for code generation relies on verifiable rewards from unit test pass rates. Yet high-quality test suites are scarce, existing datasets offer limited coverage, and static rewards fail to adapt as models improve. Recent self-play methods unify code and test generation in a single model, but face a inherent dilemma: white-box access leads to self-collusion where the model produces trivial tests for easy rewards, yet black-box restriction yields generic tests that miss implementation-specific bugs. We introduce Code-A1, an adversarial co-evolution framework that jointly optimizes a Code LLM and a Test LLM with opposing objectives. The Code LLM is rewarded for passing more tests, while the Test LLM is rewarded for exposing more defects. This architectural separation eliminates self-collusion risks and safely enables white-box test generation, where the Test LLM can inspect candidate code to craft targeted adversarial tests. We further introduce a Mistake Book mechanism for experience replay and a composite reward balancing test validity with adversarial difficulty. Experiments on Qwen2.5-Coder models demonstrate that Code-A1 achieves code generation performance matching or exceeding models trained on human-annotated tests, while significantly improving test generation capability. |
| title | Code-A1: Adversarial Evolving of Code LLM and Test LLM via Reinforcement Learning |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2603.15611 |