.png){kind=icon}
Return
CSLLM: Code-Specific Large Language Models—A Survey
.png){kind=icon}
PRE
AI
DOI：10.1016/j.eswa.2025.130991.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• CSLLMs tackle syntax, efficiency, and security gaps in general LLMs. • Review compares Codex, CodeT5, StarCoder, and DeepSeekCoder models. • Identifier-aware pretraining and RAG enhance CSLLM performance. • CSLLMs show higher accuracy, syntax correctness, and reliability. • Ethical risks, bias, and high complexity remain open research issues.
Journal
IF:
7.5
Papers:
2.9W
Citations:
10.2W
Organization
No organization information available
Cited Papers
ShareSave
ShareSaveScaling hermeneutics: a guide to qualitative coding with LLMs for reflexive content analysis.png)
EPJ DATA SCIENCE
OAAI
EPJ DATA SCIENCEIF2.5
OAAIShareSaveShareSaveShareSave
View more