Research map / Large language models / Agents, code and math
IR-OptSet: An Optimization-Sensitive Dataset for Advancing LLM-Based IR Optimizer
Zi Yang, Lei Qiu, FANG LYU, Ming Zhong, Zhilei Chai, Haojie Zhou, Huimin Cui, Xiaobing Feng
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Introduces IR-OptSet, an optimization-sensitive dataset of LLVM IR pairs for training LLMs to act as IR optimizers.
Problem
Rule-based compiler IR optimizations are hard to maintain, and LLMs' effectiveness as IR optimizers is limited by the lack of suitable training data (abstract truncated; details inferred only from title).
Approach
A dataset designed to be optimization-sensitive, intended to advance LLM-based IR optimizers; the specific construction method is not stated in the available text.
Closest papers by meaning
- Can Large Language Models Understand Intermediate Representations in Compilers? (ICML 2025 · similarity 0.88)
- QiMeng-NeuComBack: Self-Evolving Translation from IR to Assembly Code (NeurIPS 2025 · similarity 0.87)
- REASONING COMPILER: LLM-Guided Optimizations for Efficient Model Serving (NeurIPS 2025 · similarity 0.87)
- How efficient is LLM-generated code? A rigorous & high-standard benchmark (ICLR 2025 · similarity 0.87)
- Mercury: A Code Efficiency Benchmark for Code Large Language Models (NeurIPS 2024 · similarity 0.86)
- Unsupervised Evaluation of Code LLMs with Round-Trip Correctness (ICML 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
