NLP语言模型发展一览深度学习阶段：encoder-decoder/word2vec/attention预训练模型阶段：transformer/bert/GPT-2/GPT-3，pre- training+fine-tuning大语言模型阶段：GPT3.5/GPT4，instruction-tuning/prompt-train

一 推荐场景Xu 等 - 2024 - Prompting Large Language Models for Recommender Sy.pdf 1 LLM作为推荐模型1）判别式推荐discriminative recommendation item scoring re-ranking tasks click-through rate prediction [tips]https://git

When Large Language Model based Agent Meets User Behavior Analysis: A Novel User Simulation Paradigma profiling module, a memory module and an action module 用户配置文件包括ID、姓名、性别、年龄、特征、职业、兴趣和行为特征。 （Featu