MILES: Modular Instruction Memory with Learnable Selection for Self-Improving LLM Reasoning

Ruilin Tong and a coauthor submitted a paper to arXiv on July 8, 2026, titled "MILES: Modular Instruction Memory with Learnable Selection for Self-Improving LLM Reasoning." The paper is categorized under cs.CL and cs.LG. A PDF and TeX source are available through arXiv, which issued a DOI via DataCite. The work addresses methods for improving LLM reasoning through modular instruction memory with learnable selection.