Terminology
- word vector: 词向量
- Word2vec: word to vector
- objective function gradients: 目标函数梯度
- optimization basics: 优化基础
- effective modern methods for deep learning: 深度学习有效的现代方法
**Key learning today(今天的学习重点):
The (really surprising!) result is that word meaning can be represented rather well by a (high-dimensional) vector of real numbers.
真的令人惊讶的结果表明,用高维的实数向量可以很好地表示单词的意思。**
rather well: 相当好
real numbers: 实数
Lecture 1: Introduction and Word Vectors (80mins in total)
The course(12.5%)
Course logistics in brief 课程组织工作简介
- Instructor: Christopher Manning (aka. Chris Manning)
- Head TA: Anna Goldie
- Coordinator(协调人,班主任): Amelie Byun
- TAs: Many wonderful people! See website
- Time: Tu/Th 3:15-4:45 Pacific time, Zoom U.(->video)
- We’ve put a lot of other important information on the class webpage. Please read it!
http://cs224n.stanford.edu/
a.k.a.,http://www.stanford.edu/class/cs224n/ - TAs, syllabus (教学大纲), help sessions(帮助会话)/office hours, Ed(for all course questions/discussion)
- Office hours start Thursday evening!
- Python/numpy and then PyTorch tutorials: First two Fridays 1:30-2:30 Pacific time on Zoom U.
- Slide PDFs uploaded before each lecture
What do we hope to teach? (A.k.a. “learning goals”) 我们想要教什么?也就是,学习目标。
- The foundations of the (effective modern methods for deep learning) applied to NLP (应用于自然语言处理的有效的现代深度学习方法的基础) Basics first, then key methods used in NLP: Word vectors, feed-forward networks, recurrent networks, attention, encoder-decoder models, transformers, etc. 首先是基础,然后是在NLP中使用的关键方法,如:词向量,前馈网络,循环网络,注意力,编码器-解码器模型,transformers,等。
- A big picture understanding of human languages and the difficulties in understanding and producing them via computers. (对人类语言的全局理解,通过计算机理解和生成人类语言的困难)
- An understanding of and ability to build systems (in pytorch) for some of the major problems in NLP: Word meaning, dependency parsing, machine translation, question answering. 对于在NLP中的一些主要的问题,对系统的理解和使用pytorch构建该系统的能力。主要问题有:词的意思,依存分析,机器翻译,问答系统。
- Course work and grading policy
- 5x1-week Assignments: 6%+4x12%: 54%
- HW1 is released today! Due next Tuesday! At 3:15 p.m.
- Submitted to Gradescope in Canvas (i.e., using @stanford.edu email for your Gradescope account)
- Final Default or Custom Course Project (1-3 people): 43%
- Project proposal: 5%, milestone: 5%, poster or web summary: 3%, report: 30%
- Participation: 3%
- Guest lecture reactions, Ed, (???) course evals, karma - see website!
- Late day policy
- 6 free late days; afterwards, 1% off course grade per day late
- assignments not accepted more than 3 days late per assignment unless given permission in advance
- Collaboration policy: Please read the website and the Honor Code! Understand allowed collaboration and how to document it: Don’t take code off the web; acknowledge working with other students; write your own assignment solutions.