Category: Neural Networks

Saving and reading network models 0

Saving and reading network models

Hits: 0First create a new network model: import torchvision vgg16 = torchvision.models.vgg16(pretrained=False) pretrained=False means that the network model is not initially trained. Way 1: save data: import torch torch.save(vgg16, 'vgg16_method1.pth') This method saves both...

pytorch – optimizer usage 0

pytorch – optimizer usage

Hits: 0Usage steps: Build an optimizer, clear each gradient of the parameter, call the backward propagation of the [loss function] to find the gradient of each parameter, and finally tune each parameter. Take Stochastic...

pytorch – loss function 0

pytorch – loss function

Hits: 0Loss is used to do two things, one is to calculate the gap between the actual output and the target, and the other is to provide a basis for our [back-propagation] update data....

Neural Networks – Nonlinear Activation 0

Neural Networks – Nonlinear Activation

Hits: 0[]Introduction to Nonlinearity The main purpose of nonlinear transformation is to add some nonlinear features to the network. Common nonlinear activations: ReLU: Indicates that the shape of the input can be arbitrary During...

Neural Networks – Max Pooling 0

Neural Networks – Max Pooling

Hits: 0[The purpose of maximum pooling is] to reduce the parameters of neural network training while retaining the original features, so that the training time is reduced. 1080p equivalent video becomes 720p An introduction...