Yorko/dl_in_nlp_deeppavlov_cs224n_spring2020

"Deep Learning in Natural Language Processing" - a course by DeepPavlov built on top of Stanford's cs224n

"DL in NLP" by DeepPavlov. Spring 2020

The course is held in Russian. Main platform is this Telegram channel. Lead by Alex Sorokin (@alexeysorokin89) & Vasiliy Konovalov (@letmypeoplego), both affiliated with MIPT. Supported by OpenDataScience, unoffical discussions are held in the #class_cs224n Slack channel.

Main links

• official course repository
• cs224n lectures 2019, playlist
• cs224n lecture notes
• Lectures and seminars by DeepPavlov, playlist 🇷🇺
• Telegram channel "DL in NLP Spring 2020" 🇷🇺
• DeepPavlov forum 🇷🇺
• Task grading FAQ 🇷🇺
• A list of useful resources from spring 2019 course session

Week 1. Word Vector Representations

1. Lectures:

   • Word Vector Representations (word2vec)
   • Word Vectors and Word Senses (0:00-38:40, 58:00-1:20:00)
   • cs231n lecture
   • cs224n lecture notes
   • FastText paper
   • GloVe paper
   • cs224n lecture on GloVe by Richard Socher (2nd author)
   • a lecture on FastText by Facebook
   • A particuliarly good tutorial on word2vec by Chris McCormick: part1, part2

2. Quiz 1 due on Feb 10th, 9am Moscow time.

3. Quiz 2 due on Feb 17th, 9am Moscow time.

4. ** Assignment 1.** Github Classroom invitation link. Due on Feb 18th, 00:00 MST.

5. Assignment 2. Studying word embeddings and implemeting word2vec. Github Classroom invitation link. Due on March 20th (extended), 23:59 Moscow time.

Weeks 2-4. Basics of neural networks

1. Materials:

   • cs231n lecture
   • cs231n lecture notes: part 1 and part 2

2. Quiz 3 due on Feb 24th, 9am Moscow time.

3. Quiz 4 due on March 2nd, 9am Moscow time.

Week 5. RNNs and language models

1. Materials:

   • cs231n lecture 6
   • cs231n lecture 7
   • cs231n lecture notes
   • lecture on ngram models 🇷🇺
   • "The Unreasonable Effectiveness of Recurrent Neural Networks", a post by A. Karpathy
   • "Understanding LSTM Networks", a post by Christopher Olah
   • On gradients in RNNs, a paper

2. Quiz 5 due on March 9th, 9am Moscow time.

3. Quiz 6 due on March 17th, 10am Moscow time.

Now that the official course repo is constantly updated, better to follow that one.