vinsis/understanding-neuralnetworks-pytorch

Understanding nuts and bolts of neural networks with PyTorch

Understanding nuts and bolts of neural networks with PyTorch

(Work in progress)

This is a series of articles in an attempt to understand the witchcraft that is the neural networks (using PyTorch).

Basic

• Gradient descent and various ways to implement it in PyTorch

• Dropout

• Batch normalization

• Backpropagation and detach

• Losses in PyTorch

• register_buffer

Intermediate

• Groups in convolution

• Hooks

• Weight normalization

• Einstein summation

• Instance and layer normalizations