A multi-label prompt classification corpus designed for training lightweight, CPU-efficient machine learning models (such as CatBoost, LightGBM, and FastText) for prompt complexity estimation, task intent classification, output token length forecasting, and dynamic LLM routing.
