Req2XML-CoT-SFT Dataset

This dataset contains 2,000 samples designed for supervised fine-tuning (SFT) of large language models on the task of translating natural language software and system requirements into formal Abstract Syntax Tree (AST) XML representations. Each sample includes a natural language requirement as input, a structured four-step Chain-of-Thought reasoning process, and a corresponding ground truth AST XML output. The dataset was synthetically generated using Qwen-max across multiple engineering domains including autonomous driving and industrial control systems. It is intended to support research on requirement formalization, structured output generation, and reasoning-enhanced fine-tuning of language models.