Search papers, labs, and topics across Lattice.
The paper introduces "Learning to Commit," a framework that enhances LLM-based coding agents' ability to generate organic pull requests by learning from a repository's commit history. The agent uses supervised contrastive reflection on past commits to learn coding style, API usage, and architectural invariants specific to the project. By conditioning code generation on these learned skills, the agent produces pull requests that better align with the project's conventions, leading to improved organicity scores on held-out future tasks.
LLMs can learn to generate more "organic" pull requests by distilling coding style, API usage, and architectural invariants from a project's commit history, leading to better acceptance rates.
Large language model (LLM)-based coding agents achieve impressive results on controlled benchmarks yet routinely produce pull requests that real maintainers reject. The root cause is not functional incorrectness but a lack of organicity: generated code ignores project-specific conventions, duplicates functionality already provided by internal APIs, and violates implicit architectural constraints accumulated over years of development. Simply exposing an agent to the latest repository snapshot is not enough: the snapshot reveals the final state of the codebase, but not the repository-specific change patterns by which that state was reached. We introduce Learning to Commit, a framework that closes this gap through Online Repository Memory. Given a repository with a strict chronological split, the agent performs supervised contrastive reflection on earlier commits: it blindly attempts to resolve each historical issue, compares its prediction against the oracle diff, and distils the gap into a continuously growing set of skills-reusable patterns capturing coding style, internal API usage, and architectural invariants. When a new PR description arrives, the agent conditions its generation on these accumulated skills, producing changes grounded in the project's own evolution rather than generic pretraining priors. Evaluation is conducted on genuinely future, merged pull requests that could not have been seen during the skill-building phase, and spans multiple dimensions including functional correctness, code-style consistency, internal API reuse rate, and modified-region plausibility. Experiments on an expert-maintained repository with rich commit history show that Online Repository Memory effectively improves organicity scores on held-out future tasks.
Tsinghua AI