Code-A1: Adversarial Evolving of Code LLM and Test LLM via Reinforcement Learning
This work solves the problem of generating reliable code and tests for developers and researchers in AI and software engineering, representing a novel method for a known bottleneck rather than an incremental improvement.
The paper tackled the problem of reinforcement learning for code generation by addressing the scarcity of high-quality test suites and the limitations of existing methods, introducing Code-A1, an adversarial co-evolution framework that jointly optimizes a Code LLM and a Test LLM, which achieved code generation performance matching or exceeding models trained on human-annotated tests and significantly improved test generation capability.
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.