XINJIE ZHANG 《大模型技术综述与智能导购 Agent 实践》A Technical Review of LLMs and a Furniture Shopping Agent in Practice
北京大学暑期课程《大模型：从基础到实战》的课程考核成果，包含一份大模型知识综述和一个智能导购 Agent 的实践部分。 The assessed deliverable for Peking University’s summer course “Large Language Models: From Foundations to Practice” — a written review of LLM topics plus a hands-on furniture shopping agent.
为什么做Why I made it
课程考核要求把学到的大模型知识系统整理，并落到一个实践项目上。 The assessment asked us to organise what we had learned about large language models and apply it to a hands-on project.
我做了什么What I did
我用自己的语言梳理了大模型、NLP、词向量、Transformer、数据生命周期、对齐与安全、模型评测、语言 / 多模态 / 具身模型、Agent 等主题；并与 AI 协作完成智能导购实践部分。 In my own words I worked through topics including large language models, NLP, word vectors, Transformers, the data lifecycle, alignment and safety, model evaluation, language / multimodal / embodied models, and agents; and I built the shopping-agent practice section together with AI.
AI 帮了什么How AI helped
AI 辅助实践部分的编程与页面实现。 AI assisted with the coding and page implementation of the practice section.
我负责的判断What I decided
我判断每个主题的表述是否准确、结构是否清楚，以及实践部分的结果是否可用。 I judged whether each topic was described accurately, whether the structure was clear, and whether the practice results were usable.