Hi!
This is a page of the course "Natural Language Processing (2023-2024)"!
Here you can find all materials presented during lectures and seminars.
If you have any questions, please contact me via mail (sorokin.semen2020@gmail.com) or via Telegram @SemenSorok1n.
Sheets with Assessment
Nearly all materials are made by BobaZooba for the MS Deep Learning course. Thank you a lot for your experience and information sharing!
Week topic | Date | Video link | Homework | Useful links |
---|---|---|---|---|
About course. ML vs DL. Linear regression (start). | 16.09.2022 | Lec#1 | 1. Intro Linear Regression. In General 2. How to calculate regression using MSE 3. All about linear transformation (watch at least 1 week) |
|
MLP. Non-linear transfomation. | Lec#2 | Logistic regression general Logistic regression details part #1 Log likelihood in log regression |
||
Backpropagation for simple neural networks. Gradient Descent | Lec#3 | Derivatives Partial derivarives (en) Partial derivatives (ru) |
||
Gradient Descent (part 2) | Lec#4 | HW #0 | Gradient Multivariable chain rule |
|
Word Embeddings | Lec#5 | HW #1 | ||
RNN (part 1) | Lec#6 | |||
RNN (part 2) and CNN for texts | Lec#7 | HW #2 | ||
Language models and transfer learning based on RNN | Lec#8 | UlmFit paper | ||
Elmo. Attention in NMT | Lec#9 | ELMO overview ELMO paper Highway layer |
||
Attention in NMT | Lec#10 | |||
Transformer and BPE | Lec#11 | |||
BERT | Lec#12 | |||
GPT, LLama and other models from zoo. | Lec#13 | |||
Promt engineering and LLM finetuning: PEFT | Lec#14 |
HW | Deadline | GoogleForm |
---|---|---|
HW#0 (folder) | due to | GoogleForm Link for submit |
HW#1 (folder) | due to | GoogleForm Link for submit |
HW#2 (folder) | due to | GoogleForm Link for submit |
HW | Deadline | GoogleForm |
---|---|---|
Form #1 | due to | GoogleForm Link |