RetailOpt-190 is a solver-validated benchmark for evaluating semantic reliability in text-to-optimization. It tests whether LLM-based agents can reconstruct the intended optimization structure—not just produce runnable code.
RetailOpt-190 contains 190 retail supply chain optimization instances designed to test compositional consistency in LLM-generated optimization code. Each… See the full description on the dataset page:
https://huggingface.co/datasets/Jacoblian/RetailOpt-190.
