New course announcement ✨
We're teaching an in-person LLM bootcamp in the SF Bay Area on November 14, 2023. Come join us if you want to see the most up-to-date materials building LLM-powered products and learn in a hands-on environment.
[ Ссылка ]
Hope to see some of you there!
--------------------------------------------------------------------------------------------- In this video, we will dive deep into Recurrent Neural Networks (RNNs).
First, we will briefly look at the sequence problems. Then, we will introduce the RNN architecture. Then, we will discuss the problems with RNNs and solutions to address them. We will review a case study on Machine Translation at Google. We will also explore the CTC loss function that will be used in lab 3. Afterward, we will dissect the pros and cons of RNNs. Finally, we will preview a non-recurrent sequence model (a topic for next week!).
00:00 - Introduction
01:34 - Sequence Problems
06:28 - Review of RNNs
22:00 - Vanishing Gradient Issue
27:52 - LSTMs and Its Variants
34:10 - Bidirectionality and Attention from Google's Neural Machine Translation
46:38 - CTC Loss
52:12 - Pros and Cons of Encoder-Decoder LSTM Architectures
54:55 - WaveNet
![](https://s2.save4k.ru/pic/2b0TPDmzoaQ/maxresdefault.jpg)