{"ID":2861766,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2510.00373","arxiv_id":"2510.00373","title":"Combining Large Language Models and Gradient-Free Optimization for Automatic Control Policy Synthesis","abstract":"Large Language models (LLMs) have shown promise as generators of symbolic control policies, producing interpretable program-like representations through iterative search. However, these models are not capable of separating the functional structure of a policy from the numerical values it is parametrized by, thus making the search process slow and inefficient. We propose a hybrid approach that decouples structural synthesis from parameter optimization by introducing an additional optimization layer for local parameter search. In our method, the numerical parameters of LLM-generated programs are extracted and optimized numerically to maximize task performance. With this integration, an LLM iterates over the functional structure of programs, while a separate optimization loop is used to find a locally optimal set of parameters accompanying candidate programs. We evaluate our method on a set of control tasks, showing that it achieves higher returns and improved sample efficiency compared to purely LLM-guided search. We show that combining symbolic program synthesis with numerical optimization yields interpretable yet high-performing policies, bridging the gap between language-model-guided design and classical control tuning. Our code is available at https://sites.google.com/berkeley.edu/colmo.","short_abstract":"Large Language models (LLMs) have shown promise as generators of symbolic control policies, producing interpretable program-like representations through iterative search. However, these models are not capable of separating the functional structure of a policy from the numerical values it is parametrized by, thus making...","url_abs":"https://arxiv.org/abs/2510.00373","url_pdf":"https://arxiv.org/pdf/2510.00373v1","authors":"[\"Carlo Bosio\",\"Matteo Guarrera\",\"Alberto Sangiovanni-Vincentelli\",\"Mark W. Mueller\"]","published":"2025-10-01T00:42:15Z","proceeding":"cs.LG","tasks":"[\"cs.LG\",\"cs.AI\",\"cs.NE\",\"eess.SY\"]","methods":"[\"Large Language Model\",\"Language Model\"]","project_urls":"[\"https://sites.google.com/berkeley.edu/colmo\"]","has_code":false,"code_links":[{"ID":608833,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2861766,"paper_url":"https://arxiv.org/abs/2510.00373","paper_title":"Combining Large Language Models and Gradient-Free Optimization for Automatic Control Policy Synthesis","repo_url":"https://github.com/google/safevalues","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0},{"ID":608834,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2861766,"paper_url":"https://arxiv.org/abs/2510.00373","paper_title":"Combining Large Language Models and Gradient-Free Optimization for Automatic Control Policy Synthesis","repo_url":"https://github.com/carlobosio/colmo-code","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}