Рет қаралды 9,235
Transformers are the State of the Art Models nowadays, but how do they work?
This video explains and demystifies the novel neural network architecture in an intuitive manner with step by step explanation and illustrations on how transformers work.
This video also explains what is the intuition behind using QUERY, KEY and VALUE terminology in Attention Mechanism.
Chapters
0:00 Introduction
0:28 High Level Working Overview of Encoder & Decoder
1:34 Encoder - Decoder Flow
3:03 Can we have ONLY encoder or ONLY decoder based architectures?
5:22 The ENCODER Components
6:06 Why Self-Attention?
6:48 How to compute a Self-Attention Mechanism?
9:31 Intuition behind Query, Key and Value Terminology
11:04 Feed-Forward Layer
11:53 Layer Normalization
13:13 Positional Embeddings
14:48 Classification Head
15:41 The Decoder
#transformers #datascience #machinelearning #encoder #decoder #neuralnetwork