output=model(inputs) #calcualte output using input

loss=criterion(output,targets) #calculate loss
loss.backward() #Backpropagate loss

optimiser.step() #optimiser