Paper

Code2LoRA: Hypernetwork-Generated Adapters for Code Language Models under Software Evolution

Liliana Hotsko, Yinxi Li, Yuntian Deng, Pengyu Nie

First arXiv submission: June 4, 2026 · arXiv

Trains a hypernetwork to generate repository-specific LoRA adapters from source code and update them as commits arrive.

Read paper ↗

What it does

Code2LoRA-Static maps a repository snapshot to an adapter. Code2LoRA-Evo updates a GRU state with each code diff and generates the adapter from that state.

Conditioning

A repository snapshot; the evolution variant also receives successive commit diffs.

Program form

Generated LoRA matrices for a frozen code language model.

Execution

The code model uses the generated adapter without adding repository context tokens to each request.

Related work

Code2LoRA and PAW train models to generate reusable weights. Code2LoRA puts repository knowledge into adapters; PAW compiles function specifications into neural programs with a compiler–interpreter interface.

Limitations

The paper evaluates Python test-assertion completion with Qwen2.5-Coder-1.5B.

Related terms: Code2LoRA, repository-conditioned adapters, commit-streaming adapters, hypernetworks, weight-space programming.

Downloads

JSON record · Markdown · Find in the full list