第1讲:神经网络与深度学习

  1. 课程介绍
  2. 相关工具下载及安装
  3. 深度学习:https://zh.d2l.ai/

第2讲:大语言模型及其平台

  1. 大语言模型基础理论
  2. 魔搭社区:https://www.modelscope.cn/docs/home
  3. 提示词工程:https://www.bilibili.com/video/BV1e8411o7NP

第3-7讲 深度学习与大语言模型专题论文系列

专题一:Transformers

  1. Transformer: Attention Is All You Need
  2. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
  3. ERNIE: Enhanced Language Representation with Informative Entities

专题二:GPT

  1. GPT1: Improving Language Understanding by Generative Pre-Training
  2. GPT2:Language Models are Unsupervised Multitask Learners)
  3. GPT3:Language Models are Few-Shot Learners
  4. GPT-4 Technical Report(参考资料,不用精读)

专题三:ChatGPT

  1. ChatGPT: https://openai.com/blog/chatgpt/
  2. Training language models to follow instructions with human feedback
  3. Deep Reinforcement Learning from Human Preferences

专题四:DeepSeek

  1. DeepSeek-V3 technical report
  2. DeepSeek-R1: incentivizing reasoning capability in LLMs via reinforcement learning
  3. Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
  4. TinyBERT: distilling BERT for natural language understanding

专题五:图神经网络

  1. A Gentle Introduction to Graph Neural Networks
  2. DoubleH: Twitter User Stance Detection via Bipartite Graph Neural Networks
  3. Semi-Supervised Classification with Graph Convolutional Networks
  4. Graph Attention Networks
  5. Heterogeneous Graph Attention Network

其他补充

  1. 工具推荐:https://kayzhou.github.io/2024/tools/
  2. AIGC合集:https://space.bilibili.com/3493138152819508/lists/4466965
  3. 如何阅读文章 https://www.bilibili.com/video/BV1H44y1t75x/
  4. 跟读者建立联系 https://www.bilibili.com/video/BV1hY411T7vy/
  5. 明白问题的重要性 https://www.bilibili.com/video/BV11S4y1v7S2/
  6. 如何讲好故事、故事里的论点 https://www.bilibili.com/video/BV1WB4y1v7ST
  7. 理由、论据和担保 https://www.bilibili.com/video/BV1SB4y1a75c
  8. GPU安装 https://www.bilibili.com/video/BV18K411w7Vs

考核方式任选一种:

(一)课堂论文精讲

3个同学一组,精讲一篇文章(要求50分钟左右)。

(二)实验报告(深度学习销量预测,商品多层次多标签分类,任选其一)

利用提供的数据编写代码和撰写实验报告,要求包含问题定义、实验模型、实验过程、实验结果。提交报告和代码。

(三)文献综述

围绕自己所在专题展开文献综述,要求字4000以内,引用相关文章20+篇。