From 5ecc680acb5af70c54a34d7d87f28740dc87b42a Mon Sep 17 00:00:00 2001 From: Yuhao Chen Date: Mon, 25 Nov 2024 21:49:44 -0500 Subject: [PATCH] modify llm course --- docs/大语言与深度生成模型/CS194.en.md | 23 +++++++++++++++++++++++ docs/大语言与深度生成模型/CS194.md | 22 ++++++++++++++++++++++ docs/大语言与深度生成模型/CS50.en.md | 22 ---------------------- docs/大语言与深度生成模型/CS50.md | 17 ----------------- 4 files changed, 45 insertions(+), 39 deletions(-) create mode 100644 docs/大语言与深度生成模型/CS194.en.md create mode 100644 docs/大语言与深度生成模型/CS194.md delete mode 100644 docs/大语言与深度生成模型/CS50.en.md delete mode 100644 docs/大语言与深度生成模型/CS50.md diff --git a/docs/大语言与深度生成模型/CS194.en.md b/docs/大语言与深度生成模型/CS194.en.md new file mode 100644 index 00000000..d393b925 --- /dev/null +++ b/docs/大语言与深度生成模型/CS194.en.md @@ -0,0 +1,23 @@ +# UC Berkeley CS 194/294-267: Understanding Large Language Models: Foundations and Safety + +## Course Introduction + +- **University**: UC Berkeley +- **Prerequisites**: CS 182/282A Deep Neural Networks or equivalent, with hands-on deep learning experience. +- **Course Difficulty**: 🌟🌟🌟🌟🌟🌟 +- **Course Website**: [Understanding Large Language Models](http://rdi.berkeley.edu/understanding_llms/s24) + +"Understanding Large Language Models: Foundations and Safety" is a Spring 2024 course at UC Berkeley, co-taught by **Professor Dawn Song** and **Dan Hendrycks**, with **Yu Gai** as the GSI. This course explores the foundational principles, interpretability, scaling laws, and risks associated with large language models (LLMs) such as ChatGPT. + +The course provides a rigorous introduction to LLMs, discussing their emergence, limitations, and potential risks, as well as methods for safer and more beneficial applications. Topics covered include: + +- Foundations of LLMs +- Interpretability +- Scaling laws +- Adversarial robustness +- AI alignment and governance +- Privacy, watermarking, and Trojans +- Agency, emergence, reasoning, and mathematics +- Evaluation and benchmarking + + diff --git a/docs/大语言与深度生成模型/CS194.md b/docs/大语言与深度生成模型/CS194.md new file mode 100644 index 00000000..363d29a8 --- /dev/null +++ b/docs/大语言与深度生成模型/CS194.md @@ -0,0 +1,22 @@ +# UC Berkeley CS 194/294-267： Understanding Large Language Models: Foundations and Safety + +## 课程简介 + +- **所属大学**：UC Berkeley +- **先修要求**：CS 182/282A 深度神经网络课程或等效课程，并具有深度学习的实际操作经验。 +- **课程难度**：🌟🌟🌟🌟🌟🌟 +- **课程网站**： + +《理解大型语言模型：基础与安全》是 UC Berkeley 在 2024 年春季开设的一门课程，由 **Dawn Song 教授** 和 **Dan Hendrycks** 联合授课，助教为 **Yu Gai**。本课程重点探讨大型语言模型（如 ChatGPT）的基础原理、可解释性、扩展定律以及相关风险。 + +课程内容包括： + +- 大型语言模型的基础 +- 可解释性 +- 扩展定律 +- 对抗鲁棒性 +- 人工智能对齐与治理 +- 隐私、水印与木马 +- 代理性、涌现性、推理与数学 +- 评估与基准测试 + diff --git a/docs/大语言与深度生成模型/CS50.en.md b/docs/大语言与深度生成模型/CS50.en.md deleted file mode 100644 index f46b891f..00000000 --- a/docs/大语言与深度生成模型/CS50.en.md +++ /dev/null @@ -1,22 +0,0 @@ -# Harvard's CS50: Introduction to AI with Python - -## Descriptions - -- Offered by: Harvard University -- Prerequisites: Basic knowledge of probability theory and Python -- Programming Languages: Python -- Difficulty: 🌟🌟🌟 -- Class Hour: 30 - -A very basic introductory AI course, what makes it stand out is the 12 well-designed programming assignments, all of which will use the learned knowledge to implement a simple game AI, such as using reinforcement learning to play Nim game, using max-min search with alpha-beta pruning to sweep mines, and so on. It's perfect for newbies to get started or bigwigs to relax. - -## Course Resources - -- Course Website: -- Recordings: -- Textbooks: No textbook is needed in this course. -- Assignments: with 12 programming labs of high quality mentioned above. - -## Personal Resources - -All the resources and assignments used by @PKUFlyingPig in this course are maintained in [PKUFlyingPig/cs50_ai - GitHub](https://github.com/PKUFlyingPig/cs50_ai). diff --git a/docs/大语言与深度生成模型/CS50.md b/docs/大语言与深度生成模型/CS50.md deleted file mode 100644 index 388ce293..00000000 --- a/docs/大语言与深度生成模型/CS50.md +++ /dev/null @@ -1,17 +0,0 @@ -# CS324 - Large Language Models - -## 课程简介 - -- 所属大学：Stanford -- 先修要求：深度学习 -- 编程语言：Python -- 课程难度：🌟🌟🌟🌟🌟 -- 预计学时：30 小时 - -一门非常基础的 AI 入门课，让人眼前一亮的是 12 个设计精巧的编程作业，都会用学到的 AI 知识去实现一个简易的游戏 AI，比如用强化学习训练一个 Nim 游戏的 AI，用 alpha-beta 剪枝去扫雷等等，非常适合新手入门或者大佬休闲。 - -## 课程资源 - -- 课程网站： - -