xiaoxuanNLP

Translates natural language requests into evalscope CLI commands for LLM evaluation (eval) and performance benchmarking (perf). Discovers benchmarks via evalscope benchmark-info CLI with tag-based filtering (e.g. --tag Math Coding) and detailed metadata queries. Also supports result visualization via evalscope app. Use when the user wants to evaluate model capabilities, run performance/stress tests, find or filter benchmarks by capability tags, get benchmark details, or view evaluation results.