927d2b319bc7a6c31c63ad64e6124477c2a5b583,examples/pytorch/ogb/ogbn-products/graphsage/main.py,SAGE,inference,#SAGE#Any#Any#Any#Any#,56

Before Change

                h = x[input_nodes].to(device)
                h_dst = h[:block.number_of_dst_nodes()]
                h = layer(block, (h, h_dst))
                if l != len(self.layers) - 1:
                    h = self.activation(h)
                    h = self.dropout(h)

                y[output_nodes] = h.cpu()

            x = y
        return y

After Change

    
    return (th.argmax(pred, dim=1) == labels).float().sum() / len(pred)

def evaluate(model, g, nfeat, labels, val_nid, test_nid, device):
    
    Evaluate the model on the validation set specified by ``val_mask``.
    g : The entire graph.
    inputs : The features of all the nodes.
    labels : The labels of all the nodes.
    val_mask : A 0-1 mask indicating which nodes do we actually compute the accuracy for.
    device : The GPU device to evaluate on.
    
    model.eval()

Instances