NLData2Opt Benchmark for Optimization Modeling from Natural Language
October 19, 2026
NLData2Opt evaluates LLMs on the multi-stage reasoning required to convert text and noisy data into optimization parameters. The benchmark tests the ability to extract relevant attributes, compute missing parameters via data transformation, and predict future values from historical observations.
HOW THIS AFFECTS YOU
●
researcherYou can use this to test how well models handle real-world data noise in optimization workflows.