Fackelwurzelquadrat
#its not built in but you can get RMSE like so:
criterion = nn.MSELoss()
loss = torch.sqrt(criterion(x, y))
loss.backward()
print(x.grad)
Friendly Hawk
#its not built in but you can get RMSE like so:
criterion = nn.MSELoss()
loss = torch.sqrt(criterion(x, y))
loss.backward()
print(x.grad)