Transformer principle and code elaboration

Course link ：​ ​​https://edu.51cto.com/course/30124.html​​

Transformer From NLP（ natural language processing ）, And cross-border application to CV（ Computer vision ） field . At present, it has become a new paradigm of deep learning , Great influence and application prospects .

This course is right Transformer Principle and PyTorch And TensorFlow 2 Elaborate the code , To help you master its detailed principle and specific implementation .

Principle elaboration Part includes ： Attention mechanism and self attention mechanism 、Transformer An overview of the architecture 、Encoder Long attention （Multi-Head Attention）、Encoder The location code of （Positional Encoding）、 Residual link （Residual Connection）、 Layer normalization （Layer Normalization）、FFN（Feed Forward Network）、Transformer Training and performance of 、Transformer Machine translation workflow . 

Code elaboration Some use Jupyter Notebook Yes Transformer Of PyTorch And TensorFlow 2 The implementation code is interpreted line by line , Include ： install PyTorch/TensorFlow、Transformer Data set loading and preprocessing code interpretation 、Transformer Position coding and multi head attention code interpretation 、Transformer Of Transformer Class code interpretation 、Transformer Optimizer and loss function code interpretation 、Transformer Interpretation of training code 、Transformer Reasoning and weight saving code interpretation .