NEW PROMISE DATASET

This dataset extends the PROMISE requirement classification corpus to include sustainability-aware metrics for Large Language Model (LLM) inference. Each record captures a software requirement labeled as functional or non-functional, along with metadata on prompt type, token length, inference time, energy consumption (J), and carbon emissions (gCO\u2082eq) measured using the CodeCarbon library. The dataset supports research in \u201ccarbon prompting\u201d \u2014 the study of how prompt engineering strategies influence environmental footprint and model efficiency. It enables benchmarking of LLM performance, energy efficiency, and prompt optimization methods for sustainable AI applications.