Data and checkpoints for: Behavior and Representation in Open-Weight Large Language Models for Combinatorial Optimization

This archive accompanies the paper: Behavior and Representation in Open-Weight Large Language Models for Combinatorial Optimization: From Feature Extraction to Algorithm Selection accepted for publication in Computers & Operations Research. It provides a frozen snapshot of the code and data used to produce the results reported in the paper, to support transparency, reproducibility, and reuse of the analysis workflow. Unless otherwise noted below, the code, data, and other materials included in this archive are released under the Creative Commons Attribution (CC BY) 4.0 International licence. Some of the data included in this archive originate from the MATILDA dataset, provided by the University of Melbourne. These data retain their original licensing terms: Usage rights: Problem data instances are licensed under a Creative Commons Attribution-NonCommercial (CC BY-NC) 3.0 Australia licence, and they are copyrighted © University of Melbourne 2018.