Application: Practicing Pytorch fundamentals

At this point you should have a reasonable understanding of the fundementals of neural networks and PyTorch; we can reduce down what we have learnt to the following…

• Data in Pytorch is stored in arrays of n-dimensions, that we call tensors.
• Tensors make effcient use of CPU and GPU, allowing rapid calculations.
• Neural networks are fundementally a series of mathmatical opperations we represent in ‘computational graph’
• The movement of data through these opperations is the forward pass that creates a loss value.
• Loss metrics are used to measure gradients of loss, with respect to weights, using the chain rule.
• Loss gradients are used by optimisation algorithms to adjust weights and biases in order to improve the accuracy of network outputs.

Now is a good time to consolidate your new knowledge and get build our understanding of PyTorch and Lightening. We have built a jupyter notebook with the basic tensor opperations; from...