Reinforcement Learning with Negative Tests as Completeness Signal for Formal Specification Synthesis
This addresses the problem of incomplete feedback in specification synthesis for program verification, enabling more robust modular reasoning in codebases, though it is incremental as it builds on existing RL and testing approaches.
The authors tackled the challenge of generating strong program specifications by introducing SpecRL, a reinforcement learning framework that uses negative tests as a completeness signal, improving specification strength and verification success over baseline methods and generalizing to out-of-distribution benchmarks.
The specification synthesis task aims to automatically generate specifications, together with any necessary auxiliary verification annotations, for existing programs. This task is important because such specifications serve as behavioral contracts that support modular reasoning and reusable verification across a codebase. At the same time, it remains challenging because verifier-only feedback is fundamentally incomplete: passing verification establishes soundness, but cannot distinguish weak specifications from strong ones. What is missing is a fine-grained signal for specification completeness. We present SpecRL, a reinforcement learning framework for specification synthesis in Dafny. SpecRL introduces a self-contained pipeline that generates negative tests, i.e., input-output pairs that can never be produced by the program. We use the fraction of these negative tests rejected by a candidate specification as a signal of specification completeness, which is integrated into the reward for RL training. Experiments across four model sizes show that SpecRL improves both specification strength and verification success over SFT and RL with a binary specification-strength reward, generalizes to an out-of-distribution benchmark, and remains competitive on that unseen benchmark compared to much larger general-purpose LLMs.