_version_ 1866911305982541824
author Yang, Jian
Liu, Xianglong
Lv, Weifeng
Deng, Ken
Guo, Shawn
Jing, Lin
Li, Yizhi
Liu, Shark
Luo, Xianzhen
Luo, Yuyu
Pan, Changzai
Shi, Ensheng
Tan, Yingshui
Tao, Renshuai
Wu, Jiajun
Wu, Xianjie
Wu, Zhenhe
Zan, Daoguang
Zhang, Chenchen
Zhang, Wei
Zhu, He
Zhuo, Terry Yue
Cao, Kerui
Cheng, Xianfu
Dong, Jun
Fang, Shengjie
Fei, Zhiwei
Guan, Xiangyuan
Guo, Qipeng
Han, Zhiguang
James, Joseph
Luo, Tianqi
Li, Renyuan
Li, Yuhang
Liang, Yiming
Liu, Congnan
Liu, Jiaheng
Liu, Qian
Liu, Ruitong
Loakman, Tyler
Meng, Xiangxin
Peng, Chuang
Peng, Tianhao
Shi, Jiajun
Tang, Mingjie
Wang, Boyang
Wang, Haowen
Wang, Yunli
Xu, Fanglin
Xu, Zihan
Yuan, Fei
Zhang, Ge
Zhang, Jiayi
Zhang, Xinhao
Zhou, Wangchunshu
Zhu, Hualei
Zhu, King
Dai, Bryan
Liu, Aishan
Li, Zhoujun
Lin, Chenghua
Liu, Tianyu
Peng, Chao
Shen, Kai
Qin, Libo
Song, Shuangyong
Zhan, Zizheng
Zhang, Jiajun
Zhang, Jie
Zhang, Zhaoxiang
Zheng, Bo
author_facet Yang, Jian
Liu, Xianglong
Lv, Weifeng
Deng, Ken
Guo, Shawn
Jing, Lin
Li, Yizhi
Liu, Shark
Luo, Xianzhen
Luo, Yuyu
Pan, Changzai
Shi, Ensheng
Tan, Yingshui
Tao, Renshuai
Wu, Jiajun
Wu, Xianjie
Wu, Zhenhe
Zan, Daoguang
Zhang, Chenchen
Zhang, Wei
Zhu, He
Zhuo, Terry Yue
Cao, Kerui
Cheng, Xianfu
Dong, Jun
Fang, Shengjie
Fei, Zhiwei
Guan, Xiangyuan
Guo, Qipeng
Han, Zhiguang
James, Joseph
Luo, Tianqi
Li, Renyuan
Li, Yuhang
Liang, Yiming
Liu, Congnan
Liu, Jiaheng
Liu, Qian
Liu, Ruitong
Loakman, Tyler
Meng, Xiangxin
Peng, Chuang
Peng, Tianhao
Shi, Jiajun
Tang, Mingjie
Wang, Boyang
Wang, Haowen
Wang, Yunli
Xu, Fanglin
Xu, Zihan
Yuan, Fei
Zhang, Ge
Zhang, Jiayi
Zhang, Xinhao
Zhou, Wangchunshu
Zhu, Hualei
Zhu, King
Dai, Bryan
Liu, Aishan
Li, Zhoujun
Lin, Chenghua
Liu, Tianyu
Peng, Chao
Shen, Kai
Qin, Libo
Song, Shuangyong
Zhan, Zizheng
Zhang, Jiajun
Zhang, Jie
Zhang, Zhaoxiang
Zheng, Bo
contents Large language models (LLMs) have fundamentally transformed automated software development by enabling direct translation of natural language descriptions into functional code, driving commercial adoption through tools like Github Copilot (Microsoft), Cursor (Anysphere), Trae (ByteDance), and Claude Code (Anthropic). While the field has evolved dramatically from rule-based systems to Transformer-based architectures, achieving performance improvements from single-digit to over 95\% success rates on benchmarks like HumanEval. In this work, we provide a comprehensive synthesis and practical guide (a series of analytic and probing experiments) about code LLMs, systematically examining the complete model life cycle from data curation to post-training through advanced prompting paradigms, code pre-training, supervised fine-tuning, reinforcement learning, and autonomous coding agents. We analyze the code capability of the general LLMs (GPT-4, Claude, LLaMA) and code-specialized LLMs (StarCoder, Code LLaMA, DeepSeek-Coder, and QwenCoder), critically examining the techniques, design decisions, and trade-offs. Further, we articulate the research-practice gap between academic research (e.g., benchmarks and tasks) and real-world deployment (e.g., software-related code tasks), including code correctness, security, contextual awareness of large codebases, and integration with development workflows, and map promising research directions to practical needs. Last, we conduct a series of experiments to provide a comprehensive analysis of code pre-training, supervised fine-tuning, and reinforcement learning, covering scaling law, framework selection, hyperparameter sensitivity, model architectures, and dataset comparisons.
format Preprint
id arxiv_https___arxiv_org_abs_2511_18538
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle From Code Foundation Models to Agents and Applications: A Comprehensive Survey and Practical Guide to Code Intelligence
Yang, Jian
Liu, Xianglong
Lv, Weifeng
Deng, Ken
Guo, Shawn
Jing, Lin
Li, Yizhi
Liu, Shark
Luo, Xianzhen
Luo, Yuyu
Pan, Changzai
Shi, Ensheng
Tan, Yingshui
Tao, Renshuai
Wu, Jiajun
Wu, Xianjie
Wu, Zhenhe
Zan, Daoguang
Zhang, Chenchen
Zhang, Wei
Zhu, He
Zhuo, Terry Yue
Cao, Kerui
Cheng, Xianfu
Dong, Jun
Fang, Shengjie
Fei, Zhiwei
Guan, Xiangyuan
Guo, Qipeng
Han, Zhiguang
James, Joseph
Luo, Tianqi
Li, Renyuan
Li, Yuhang
Liang, Yiming
Liu, Congnan
Liu, Jiaheng
Liu, Qian
Liu, Ruitong
Loakman, Tyler
Meng, Xiangxin
Peng, Chuang
Peng, Tianhao
Shi, Jiajun
Tang, Mingjie
Wang, Boyang
Wang, Haowen
Wang, Yunli
Xu, Fanglin
Xu, Zihan
Yuan, Fei
Zhang, Ge
Zhang, Jiayi
Zhang, Xinhao
Zhou, Wangchunshu
Zhu, Hualei
Zhu, King
Dai, Bryan
Liu, Aishan
Li, Zhoujun
Lin, Chenghua
Liu, Tianyu
Peng, Chao
Shen, Kai
Qin, Libo
Song, Shuangyong
Zhan, Zizheng
Zhang, Jiajun
Zhang, Jie
Zhang, Zhaoxiang
Zheng, Bo
Software Engineering
Computation and Language
Large language models (LLMs) have fundamentally transformed automated software development by enabling direct translation of natural language descriptions into functional code, driving commercial adoption through tools like Github Copilot (Microsoft), Cursor (Anysphere), Trae (ByteDance), and Claude Code (Anthropic). While the field has evolved dramatically from rule-based systems to Transformer-based architectures, achieving performance improvements from single-digit to over 95\% success rates on benchmarks like HumanEval. In this work, we provide a comprehensive synthesis and practical guide (a series of analytic and probing experiments) about code LLMs, systematically examining the complete model life cycle from data curation to post-training through advanced prompting paradigms, code pre-training, supervised fine-tuning, reinforcement learning, and autonomous coding agents. We analyze the code capability of the general LLMs (GPT-4, Claude, LLaMA) and code-specialized LLMs (StarCoder, Code LLaMA, DeepSeek-Coder, and QwenCoder), critically examining the techniques, design decisions, and trade-offs. Further, we articulate the research-practice gap between academic research (e.g., benchmarks and tasks) and real-world deployment (e.g., software-related code tasks), including code correctness, security, contextual awareness of large codebases, and integration with development workflows, and map promising research directions to practical needs. Last, we conduct a series of experiments to provide a comprehensive analysis of code pre-training, supervised fine-tuning, and reinforcement learning, covering scaling law, framework selection, hyperparameter sensitivity, model architectures, and dataset comparisons.
title From Code Foundation Models to Agents and Applications: A Comprehensive Survey and Practical Guide to Code Intelligence
topic Software Engineering
Computation and Language
url https://arxiv.org/abs/2511.18538