[FIXED] 'NoneType' object has no attribute 'to' in mps mode(for apple m1 computer chip) for pytorch

Issue

um_epochs = 5
device = torch.device("mps")
d2l.train_ch3(net, train_iter, test_iter, loss, num_epochs, batch_size, None, None, optimizer).to(device)

Then the terminal told me that:

 1 num_epochs = 5
      2 device = torch.device("mps")
----> 3 d2l.train_ch3(net, train_iter, test_iter, loss, num_epochs, batch_size, None, None, optimizer).to(device)

AttributeError: 'NoneType' object has no attribute 'to'

Here is the function:

def train_ch3(net, train_iter, test_iter, loss, num_epochs, batch_size,
              params=None, lr=None, optimizer=None):
    for epoch in range(num_epochs):
        train_l_sum, train_acc_sum, n = 0.0, 0.0, 0
        for X, y in train_iter:
            y_hat = net(X)
            l = loss(y_hat, y).sum()
            
            # Gradient zeroing
            if optimizer is not None:
                optimizer.zero_grad()
            elif params is not None and params[0].grad is not None:
                for param in params:
                    param.grad.data.zero_()
            
            l.backward()
            if optimizer is None:
                sgd(params, lr, batch_size)
            else:
                optimizer.step() 
            
            
            train_l_sum += l.item()
            train_acc_sum += (y_hat.argmax(dim=1) == y).sum().item()
            n += y.shape[0]
        test_acc = evaluate_accuracy(test_iter, net)
        print('epoch %d, loss %.4f, train acc %.3f, test acc %.3f'
              % (epoch + 1, train_l_sum / n, train_acc_sum / n, test_acc))

In the process of searching, I noticed that some alike problems are owing to the overloading mechanism for "=" and "-". But in this function I didn't find any similar slice...

So what's the matter with this NoneType?

Solution

The function train_ch3 does not return anything explicitly, which means that it returns NonType object. And for sure, NoneType object has not attribute named 'to'

I don't know what type of class dl3 is, but if you want to load a tensor or pytorch module onto a specific device, you should use the method to of that tensor or module like: net.to(device) or X.to(device), y.to(device)

Answered By - mjung