Taking together Stanford CS224n course

Course page: link
YouTube Videos: link

• Spring 2020: link
  
• Summer 2019: link
  
• Fall 2018: link
  

How to submit homework:

1. Fork repository
2. Commit finished assignment and push it
3. Make a pull request
4. Wait until 1 or 2 reviewers approve your solution
5. Merge your pull request

• Video lecture
• Additional materials:
  • Word2Vec Tutorial - The Skip-Gram Model
  • Efficient Estimation of Word Representations in Vector Space - Word2Vec paper
  • Distributed Representations of Words and Phrases and their Compositionality - negative sampling paper
  • A Corrected CBOW Implementation
  • Bias in Word Embeddings

• Video lecture
• Enriching Word Vectors with Subword Information
• Additional materials:
  • GloVe paper
  • FastText talk
  • CS231n: Backpropagation
• Optional materials:
  • A Latent Variable Model Approach to PMI-based Word Embeddings

• Video lecture
• Matrix calculus review, Differential calculus review
• Derivatives, Backprop, and Vectorization
• Learning representations by back-propagating errors
• CS231n notes on network architectures and backprop
• Additional materials:
  • Yes you should understand backprop
• Optional materials:
  • NLP (Almost) from Scratch - optional, but highly recommended

• Initialization, Normalization video lecture
  • Lecture notes 1
  • Lecture notes 2
• Optimization video lecture
  • Lecture notes 1

• Video lecture
• Lecture notes
• The Unreasonable Effectiveness of Recurrent Neural Networks

• Video lecture
• Lecture notes
• Sequence modeling: Recurrent and Recursive Neural Networks
• On the difficulty of training Recurrent Neural Networks
• Vanishing gradient Jupyter notebook
• Understanding LSTM

Assignment 1
deadline: 16/07/21

Assignment 2 explanation
Assignment 2 code
deadline: 30/07/21

Assignment 3 explanation (Skipping)
Assignment 3 code (Skipping)
deadline: 7/09/21

• Adversarial attacks on NLP models (https://arxiv.org/pdf/1911.03677.pdf)
• NLP models (BERT, RoBERT, ALBERT, etc.) in continual or multitask learning setting