backpropagation_pytorch.py

import torch

x = torch.ones(3) # Input Tensor
y = torch.zeros(2) # Expected Output Tensor

w = torch.randn(3, 2, requires_grad=True) # Weight matrix
b = torch.randn(2 , requires_grad=True) # Bias

z = torch.matmul(x, w) + b # Calculated output

loss = torch.nn.functional.binary_cross_entropy_with_logits(z, y)

loss.backward() # Computes the gradient of loss w.r.t w and b

print(w.grad)
print(b.grad)