ORLM: A Customizable Framework in Training Large Models for Automated Optimization Modeling

Optimization modeling and solving play a critical role in the application of Operations Research (OR) tools to address real-world problems, yet they pose challenges and require extensive expertise from OR experts. With the advent of large language models (LLMs), new opportunities have emerged to streamline and automate these tasks. However, current research predominantly relies on closed-source LLMs such as GPT-4, along with extensive prompt engineering techniques. This reliance stems from the scarcity of high-quality training datasets for optimization modeling, resulting in elevated costs, prolonged processing times, and privacy concerns. To address these challenges, our work is the first to propose a viable path for training open-source LLMs that are capable of optimization modeling as well as developing and executing solver codes, eventually leading to a superior ability for automating optimization modeling and solving. Particularly, we introduce a semi-automated data synthesis framework designed for optimization modeling issues, named OR-Instruct. This framework merges the training data requirements of large models with the unique characteristics of optimization modeling problems, and allows for customizable enhancements tailored to specific scenarios or modeling types. To evaluate the performance of our proposed framework, we present the IndustryOR benchmark, the inaugural industrial standard for evaluating LLMs in solving practical OR problems. Utilizing data synthesized through OR-Instruct, we train various open-source LLMs with a capacity of 7 billion parameters (dubbed ORLMs). The resulting model demonstrates significantly enhanced optimization modeling capabilities, achieving state-of-the-art performance across the NL4OPT, MAMO, and IndustryOR benchmarks. Our code and data are available at \url{https://github.com/Cardinal-Operations/ORLM}.