Web3Survivor/Onlineresearch

该数据集包含跨物理、数学、生物和化学的计算密集型、自包含且明确的STEM推理问题。问题需要多步推理、符号操作、数值精度或基于模拟的验证。每个示例包含唯一标识符、领域和子领域信息、严谨的问题陈述（含LaTeX）、确定性答案以及用于模拟或验证的可选Python代码。数据集特点包括自包含、无歧义、大量使用LaTeX、需要精确计算等，旨在测试大型语言模型的深度推理能力。数据集为标准JSON格式，预期用途包括微调STEM推理模型、评估LLM计算准确性、基准测试符号和数值推理等。

This dataset contains computationally intensive, self-contained, and unambiguous STEM reasoning problems across Physics, Mathematics, Biology, and Chemistry. Problems require multi-step reasoning, symbolic manipulation, numerical accuracy, or simulation-based verification. Each example includes a unique identifier, domain and sub-domain information, a rigorous question statement (with LaTeX), a deterministic answer, and optional Python code for simulation or verification. The dataset characteristics include being self-contained, unambiguous, heavy use of LaTeX, requiring precise computation, and designed to stress-test LLM reasoning. The dataset is provided in standard JSON format and intended for uses such as fine-tuning STEM reasoning models, evaluating LLM computation accuracy, benchmarking symbolic and numeric reasoning, etc.