罗 先 镇

xzluo@ir.hit.edu.cn (+86) 188 4556 0861 Luowaterbi.github.io Google Scholar 15290843908

教育背景

哈尔滨工业大学 · 博士研究生 · 代码大模型 2022.09 – 至今
导师：车万翔 教授 · 朱庆福 副教授
荣誉：国家奖学金 · 🏆 ACL 2025 Outstanding Paper (Top 0.3%) · 🏆 ICML 2026 Oral (Top 0.7%)
哈尔滨工程大学 · 本科 · 计算机科学与技术 2018.09 – 2022.06
荣誉：国家奖学金 · 2 × ICPC 区域赛银牌 · 2 × ICPC EC-Final 铜牌

实习经历

快手科技 · KStar 研究实习生 2025.08 – 2026.03 · 北京
阶跃星辰 · 预训练研究实习生 2024.12 – 2025.07 · 北京
度小满 · 校企合作研究员 2023.11 – 2024.09 · 北京

一作论文

🏆 ICML 2026 Oral CVE-Factory: Scaling Expert-Level Agentic Tasks for Code Security Vulnerability
罗先镇, 张景远, 周士祺, 黄金阳, 肖川, 朱庆福, 马志远, 岳星, 岳洋, 曾文聪, 车万翔
ACL 2026 Scaling Laws for Code: A More Data-Hungry Regime
罗先镇*, 郑文镇*, 朱庆福, 张荣燚, 李厚意, 黄思明, 范元涛, 车万翔
ICLR 2026 How Many Code and Test Cases Are Enough? Evaluating Test Cases Generation from a Binary-Matrix Perspective
罗先镇*, 黄金阳*, 郑文镇, 朱庆福, 徐明拯, 徐逸恒, 范元涛, 覃立波, 车万翔
🏆 ACL 2025 Outstanding Paper Turning Trash into Treasure: Accelerating Inference of Large Language Models with Token Recycling
罗先镇, 王一轩, 朱庆福, 张致铭, 张玄昱, 杨青, 许冬亮
ACL 2025 ChartCoder: Advancing Multimodal Large Language Model for Chart-to-Code Generation
赵烜乐*, 罗先镇*, 施琦, 陈驰, 王硕, 车万翔, 刘知远, 孙茂松
EMNLP 2024 Python is Not Always the Best Choice: Embracing Multilingual Program of Thoughts
罗先镇, 朱庆福, 张致铭, 覃立波, 张玄昱, 杨青, 许冬亮, 车万翔
EMNLP 2024 Make Some Noise: Unlocking Language Model Parallel Inference Capability through Noisy Training
王一轩*, 罗先镇*, 魏福烜, 刘议骏, 朱庆福, 张玄昱, 杨青, 许冬亮, 车万翔
Arxiv Is Compression Really Linear with Code Intelligence?
徐杨世杰*, 罗先镇*, 初征, 李厚意, 黄思明, 王秋锋, 车万翔, 朱庆福, 周水庚
Arxiv Success is in the Details: Evaluate and Enhance Details Sensitivity of Code LLMs through Counterfactuals
罗先镇, 朱庆福, 张致铭, 徐明拯, 程天豪, 王一轩, 初征, 徐杨世杰, 马志远, 范元涛, 车万翔
Arxiv Semi-Instruct: Bridging Natural-Instruct and Self-Instruct for Code Large Language Models
罗先镇, 朱庆福, 张致铭, 王旭, 杨青, 许冬亮, 车万翔

其他论文

ICML 2026 Memoria-Bench: A Comprehensive Benchmark for Evaluating Memory in Long-Horizon Autonomous Agents
王秋锋, ..., 黄思明, 罗先镇, ..., 张前文
ACL 2026 (Findings) Format-Adapter: Improving Reasoning Capability of LLMs by Adapting Suitable Format
王丁子睿, 张玄靓, 曹荣禹, 窦隆绪, 罗先镇, 马迎伟, 朱庆福, 车万翔, 黎槟华, 黄飞, 李永彬
Survey From Code Foundation Models to Agents and Applications: A Practical Guide to Code Intelligence
核心贡献者
Tech Report Step-3 is Large yet Affordable: Model-system Co-design for Cost-effective Decoding
核心贡献者
ACL 2025 📢 Oral OpenCoder: The Open Cookbook for Top-Tier Code Large Language Models
黄思明, 程天豪, ..., 罗先镇, ..., 汪自力
KDD 2025 Advancing Tool-Augmented Large Language Models via Meta-Verification and Reflection Learning
马志远, 刘嘉聿, 罗先镇, 黄振亚, 朱庆福, 车万翔
ACL 2025 (Findings) ChartEdit: How Far Are MLLMs From Automating Chart Analysis? Evaluating MLLMs' Capability via Chart Editing
赵烜乐*, 刘学鑫*, 杨昊玥*, 罗先镇, 曾繁虎, 李建玲, 施琦, 陈驰
COLING 2024 A Survey on Natural Language Processing for Programming
朱庆福, 罗先镇, 刘芳, 高翠芸, 车万翔
ACL 2022 (Findings) Inverse is Better! Fast and Accurate Prompt for Few-shot Slot Tagging
侯宇泰, 陈成, 罗先镇, 李博涵, 车万翔
AI Open 2022 Augmented and Challenging Datasets with Multi-step Reasoning and Multi-span Questions for Chinese Judicial Reading Comprehension
孟庆烨, 王梓岳, 陈航, 罗先镇, 王宝鑫, 陈志鹏, 崔一鸣, 吴大勇, 陈志刚, 王士进
Arxiv Automated Snippet-Alignment Data Augmentation for Code Translation
张致铭, 朱庆福, 罗先镇, 王一轩, 李博涵, 车万翔

项目经历

活字：开源通用大语言模型 12 200 2023.03 – 2023.05
Abacus：轻量级代码大语言模型 2 47 2023.10 – 2024.09