A Reproducible Semantic Benchmark for Multivendor DSM-to-CLI Translation (opens in new tab)

Translating high-level network intents into correct multivendor configurations remains a central challenge in network automation, as syntactically valid outputs may still violate the intended operational state. Despite recent advances in Large Language Models (LLMs), the field still lacks reproducible semantic benchmarks for rigorous cross-vendor evaluation. This paper presents a reproducible DSM-to-CLI semantic benchmark covering five cloud L...