@inproceedings{zhang-etal-2026-completion,
    title = "From Completion to Editing: Unlocking Context-Aware Code Infilling via Search-and-Replace Instruction Tuning",
    author = "Zhang, Jiajun  and
      Cui, Zeyu  and
      Yang, Jiaxi  and
      Zhang, Lei  and
      Jing, Yuheng  and
      Ma, Zeyao  and
      Bai, Tianyi  and
      Wang, Zilei  and
      Liu, Qiang  and
      Wang, Liang  and
      Hui, Binyuan  and
      Lin, Junyang",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Proceedings of the 64th Annual Meeting of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl/2026.acl-long.361/",
    pages = "7955--7984",
    ISBN = "979-8-89176-390-6",
    abstract = "The dominant Fill-in-the-Middle (FIM) paradigm for code completion is constrained by its rigid inability to correct contextual errors and reliance on unaligned, insecure Base models. While Chat LLMs offer safety and Agentic workflows provide flexibility, they suffer from performance degradation and prohibitive latency, respectively. To resolve this dilemma, we propose Search-and-Replace Infilling (\textbf{SRI}), a framework that internalizes the agentic verification-and-editing mechanism into a unified, single-pass inference process. By structurally grounding edits via an explicit search phase, SRI harmonizes completion tasks with the instruction-following priors of Chat LLMs, extending the paradigm from static infilling to dynamic context-aware editing. We synthesize a high-quality dataset, \textbf{SRI-200K}, and fine-tune the \textbf{SRI-Coder} series. Extensive evaluations demonstrate that with minimal data (20k samples), SRI-Coder enables Chat models to surpass the completion performance of their Base counterparts. Crucially, unlike FIM-style tuning, SRI preserves general coding competencies and maintains inference latency comparable to standard FIM. We release our dataset and models, establishing SRI as a robust, secure, and efficient alignment recipe for next-generation interactive development."
}