A new survey paper explores the application of large language models (LLMs) to software engineering tasks, focusing on how reasoning techniques can enhance performance. The paper categorizes and examines various code-specific reasoning methods, including those that leverage structural information and execution feedback. It also discusses the development of SWE agents that combine planning, tool use, and multi-step interactions, highlighting open challenges and future research directions in this evolving field. AI
IMPACT Provides a structured overview of LLM reasoning techniques applicable to software engineering, guiding future research and development in AI-powered coding tools.
RANK_REASON This is a survey paper on AI techniques applied to software engineering. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
