LLM (Large Language Model)
大语言模型（LLM），基于 Transformer 架构、在海量文本上训练得到的神经网络，能够生成、总结、翻译自然语言并进行推理。代表产品包括 GPT、Claude、LLaMA 等。
English Definition
"A neural network, typically based on the transformer architecture, trained on massive text corpora to generate, summarize, translate, or reason about natural language. Examples include GPT, Claude, and LLaMA."
用法说明 · 针对中文母语者
在 HCI 论文中常以缩写形式出现。中文常简称'大模型'。LLM 是 foundation model 的一种应用形态，注意与 generative AI（更广概念）、chatbot（应用形式）区分。
常见错误
- ·把 LLM 等同于 generative AI 的全部
- ·忽略 LLM 的'基础模型'属性而误认为只能做对话
- ·把 LLM 视作'搜索引擎替代品'而忽视其生成与推理能力
真实用例 · 2 条
"we engaged in a Research through Design (RtD) [30, 108] process that ultimately led to our creation and implementation of Quologue—a web application integrated with a large language model (LLM) that enables users to express themselves by using their own e-book highlights as a resource"
"We also highlight how a keyword-based, stepwise interaction model could offer interpretive flexibility in LLM interventions."