Pytorch fundamentals

Set of simple Pytorch scritps for better understanding pytorch fundamentals: tensors, loss functins, modules.

Some of examples are taken from https://pytorch.org/tutorials/

Examples

• autograd.py - simple example showing how autograd mechanism works

• cuda_device_info.py - how to check if cuda is available

• tensors.py- how to create tensor

• shapes.py - how to check tensor shape and reshapes to another dimension using 'view' function

• loss_functions.py - how differnet loss function works (L1, mse, crossentrophy etc)

• dorpout.py - how to apply dropout and why it scale output

• nn_blocks.py - building blocks

• lstm_input.py- how to prepare input for lstm, how to read output from lstm layer

• loss_functions.py- how loss functions work and how theri output look like

• loss_functions_combined.py- how to combine valued from more then one loss function,

• softmax_accuracy.py- how compute softmax for multilabel multivalue output and how to compute accuracy based on softmax output