how to modify tensor for U-Net decoder?

[poodarchu]

I build a sparse U-Net (without skip connection) for 3d point cloud segmentation.

Suppose the input is of shape 16000x3, the output of encoder is 1000x3. I want to check the individual value of the encoder output.

enc_out: feats: 1000x128 coords: 1000x4 (B, X, Y, Z) spatial_range: (64x64x8)

Here I want to create a new SparseTensor with new_out: feats: enc_out.feats[10:11, :], coords: enc_out.coords[10:11, ] spatial_range: enc_out.spatial_range stride: enc_out.stride.

But it fails when I try to pass the new_out to U-net decoder.

File "/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl return forward_call(*args, *kwargs) File "/lib/python3.8/site-packages/torch/nn/modules/container.py", line 217, in forward input = module(input) File "/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl return forward_call(args, **kwargs) File "/torchsparse/torchsparse/nn/modules/conv.py", line 98, in forward return F.conv3d( File "/torchsparse/torchsparse/nn/functional/conv/conv.py", line 138, in conv3d kmap = F.transpose_kernel_map( File "/torchsparse/torchsparse/nn/functional/conv/kmap/build_kmap.py", line 233, in transpose_kernel_map kmap["out_in_map"], make_divisible(kmap["sizes"][0], cta_M) TypeError: 'NoneType' object is not subscriptable

How should I adjust the values in cmap, kmap and hashmaps to achieve my goal?