问HN:向大型语言模型(LLM)报告数据的工具应该采用什么样的格式比较好?
对于命令行工具,人们喜欢视觉上有组织的文本,可能带有颜色或文本用户界面(TUI)。而机器则喜欢以制表符分隔的表格数据,或者 JSON 及类似格式。<p>那么,对于大型语言模型(LLMs)来说,处理表格或结构化数据的等价物是什么?是什么让它们能够轻松高效地理解这些数据?
查看原文
For CLI tools, humans like visually organized text possibly with colors or TUIs. Machines likes tabs-separated tabular data, or json and similar.<p>What's the equivalent for LLMs for tabular or structured data? What makes them grasp the data easily and efficiently?