Jonathan Pascoe

Where would num_features be applied on the [point cnn example](https://github.com/rusty1s/pytorch_geometric/blob/master/benchmark/points/point_cnn.py)?

Thanks for the pointer @rusty1s. The following still yields an error in validation. ``` class Net(torch.nn.Module): def __init__(self, num_classes, num_features=3): super().__init__() self.conv1 = XConv(num_features, 48, dim=3, kernel_size=8, hidden_channels=32) self.conv2 =...

hmm... It's pretty small for an ALS dataset. Maybe around 6.5 million points. The counts are definitely changing from start to end of the forward pass. After the first line...

Perfect! This seems to get me to what I'm looking for. Thanks for the assist. ```python import torch import torch.nn.functional as F from torch_geometric.nn import XConv class Net(torch.nn.Module): def __init__(self,...