Learning Day 6: Pytorch unsqueeze, squeeze, transpose and broadcasting

Continue from day 5 on unsqueeze

• Actual scenario on adding bias term to image data:

import torch

b = torch.rand(32)  #  bias…

Learning Day 10: Finding gradients for activation and loss functions in Pytorch

Activation functions

Learning Day 13: GPU with Pytorch and Visdom

GPU with Pytorch

Putting models and data…

Learning Day 4: Pytorch Datatype and MNIST

Continue from day 3 on Pytorch Datatype

Tensor dimension:

• Dim 0: scalar (a number like 2.3) — Use case: loss
• Dim 1: 1D tensor, vector (a list…

Learning Day 7: Pytorch merging, splitting and math operations

Pytorch merging and splitting

torch.cat()

• concatenate/join multiple tensors. dim=x where x is the dimension to join

Learning Day 2: Bias, variance and gradient descent

Bias vs Variance

Learning Day 1: Overview of ML and Regression

What is ML: Automatic function searching

What is model/architecture: Giving the computer a range for function searching (linear, non-linear, neural network)

Homework summary for different…

Learning Day 14: Pytorch overfitting and validation set, k-fold concept

Overfitting

• When the model is fitted too well to the training data and cannot generalise to the test data
• In real-world scenario, test data is not…

Learning Day 3: Gradient descent and Pytorch basic

Gradient descent

Code implementation of gradient descent for a simple linear regression problem. A few take-aways:

• We are trying to find the minima for the loss function, not the…

Learning Day 12: Pytorch for fully connected NN

This is using Pytorch’s convenient functions to substitute the hard work done one day 10

• In day 10, we need to define our own w and b and know exactly what their…