@inproceedings{lin-etal-2026-coeffi,
title = "{C}o{E}ffi: Enhance Efficient Code Generation via Dynamic Collaborative Decoding",
author = "Lin, Jiahui and
Zhang, Jingrui and
Zhang, Zixuan and
Li, Meilin and
Fu, Man and
Yu, Siyao and
Sun, Weifeng and
Yan, Meng",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.797/",
pages = "16886--16905",
ISBN = "979-8-89176-462-0",
abstract = "Code LLMs exhibit remarkable capabilities in automated code generation, but the generated code often suffers from low computational efficiency. Existing solutions have two limitations: (1) direct fine-tuning lacks transferability across model scales; (2) The efficiency optimization method for iterative generation-evaluation loops incurs high overhead. We propose CoEffi, which adopts inference-side intervention that directly operates on the decoding process without modifying base model parameters, enabling real-time efficiency improvement across different model scales while avoiding iterative reevaluation. CoEffi uses a small efficiency expert model for proposal and the base model to evaluate its acceptability, additionally introducing a region-aware dynamic acceptance threshold mechanism that maintains a sliding window to record historical acceptance rates and adjusts the dynamic threshold accordingly. We extensively evaluate CoEffi on four mainstream code LLM families and three efficiency benchmarks. Compared with vanilla instruction-tuned models, CoEffi improves average Beyond@1 by 9.12{\%}{--}38.62{\%} across the evaluated model families. It also improves Pass@1 on Mercury and ENAMEL, while EffiBench exhibits model-dependent correctness gains and regressions. A family specific trained expert can be reused across compatible base model scales within the same model family."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="lin-etal-2026-coeffi">
<titleInfo>
<title>CoEffi: Enhance Efficient Code Generation via Dynamic Collaborative Decoding</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jiahui</namePart>
<namePart type="family">Lin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jingrui</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zixuan</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Meilin</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Man</namePart>
<namePart type="family">Fu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Siyao</namePart>
<namePart type="family">Yu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Weifeng</namePart>
<namePart type="family">Sun</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Meng</namePart>
<namePart type="family">Yan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Code LLMs exhibit remarkable capabilities in automated code generation, but the generated code often suffers from low computational efficiency. Existing solutions have two limitations: (1) direct fine-tuning lacks transferability across model scales; (2) The efficiency optimization method for iterative generation-evaluation loops incurs high overhead. We propose CoEffi, which adopts inference-side intervention that directly operates on the decoding process without modifying base model parameters, enabling real-time efficiency improvement across different model scales while avoiding iterative reevaluation. CoEffi uses a small efficiency expert model for proposal and the base model to evaluate its acceptability, additionally introducing a region-aware dynamic acceptance threshold mechanism that maintains a sliding window to record historical acceptance rates and adjusts the dynamic threshold accordingly. We extensively evaluate CoEffi on four mainstream code LLM families and three efficiency benchmarks. Compared with vanilla instruction-tuned models, CoEffi improves average Beyond@1 by 9.12%–38.62% across the evaluated model families. It also improves Pass@1 on Mercury and ENAMEL, while EffiBench exhibits model-dependent correctness gains and regressions. A family specific trained expert can be reused across compatible base model scales within the same model family.</abstract>
<identifier type="citekey">lin-etal-2026-coeffi</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.797/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>16886</start>
<end>16905</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T CoEffi: Enhance Efficient Code Generation via Dynamic Collaborative Decoding
%A Lin, Jiahui
%A Zhang, Jingrui
%A Zhang, Zixuan
%A Li, Meilin
%A Fu, Man
%A Yu, Siyao
%A Sun, Weifeng
%A Yan, Meng
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F lin-etal-2026-coeffi
%X Code LLMs exhibit remarkable capabilities in automated code generation, but the generated code often suffers from low computational efficiency. Existing solutions have two limitations: (1) direct fine-tuning lacks transferability across model scales; (2) The efficiency optimization method for iterative generation-evaluation loops incurs high overhead. We propose CoEffi, which adopts inference-side intervention that directly operates on the decoding process without modifying base model parameters, enabling real-time efficiency improvement across different model scales while avoiding iterative reevaluation. CoEffi uses a small efficiency expert model for proposal and the base model to evaluate its acceptability, additionally introducing a region-aware dynamic acceptance threshold mechanism that maintains a sliding window to record historical acceptance rates and adjusts the dynamic threshold accordingly. We extensively evaluate CoEffi on four mainstream code LLM families and three efficiency benchmarks. Compared with vanilla instruction-tuned models, CoEffi improves average Beyond@1 by 9.12%–38.62% across the evaluated model families. It also improves Pass@1 on Mercury and ENAMEL, while EffiBench exhibits model-dependent correctness gains and regressions. A family specific trained expert can be reused across compatible base model scales within the same model family.
%U https://aclanthology.org/2026.findings-emnlp.797/
%P 16886-16905
Markdown (Informal)
[CoEffi: Enhance Efficient Code Generation via Dynamic Collaborative Decoding](https://aclanthology.org/2026.findings-emnlp.797/) (Lin et al., Findings 2026)
ACL
- Jiahui Lin, Jingrui Zhang, Zixuan Zhang, Meilin Li, Man Fu, Siyao Yu, Weifeng Sun, and Meng Yan. 2026. CoEffi: Enhance Efficient Code Generation via Dynamic Collaborative Decoding. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 16886–16905, Budapest, Hungary. Association for Computational Linguistics.