python -m unittest -v errors

[bxiang233]

Hi @nicolas-chaulet , thanks a lot for your great work. I tried to install TP3D in ubuntu. My environment settings are:
CUDA 11.1
Torch 1.9.0+cu111
GPU model: NVIDIA TITAN RTX (also tried NVIDIA GeForce RTX 2080 Ti)
python: 3.8.5

After I run "python -m unittest -v", I got:

test_kpconv (test.test_api.TestAPIEncoder) ... Model failing:
KPConvEncoder(
(down_modules): ModuleList(
(0): KPDualBlock(
Nb parameters: 3008
(blocks): ModuleList(
(0): SimpleBlock(
Nb parameters: 992; None; RadiusNeighbourFinder {'_radius': 0.05, '_max_num_neighbors': 25, '_conv_type': 'partial_dense'}
(kp_conv): KPConvLayer(InF: 4, OutF: 16, kernel_pts: 15, radius: 0.03, KP_influence: linear, Add_one: False)
(bn): FastBatchNorm1d(
(batch_norm): BatchNorm1d(16, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(activation): LeakyReLU(negative_slope=0.1)
)
(1): ResnetBBlock(
Nb parameters: 2016
(kp_conv): SimpleBlock(
Nb parameters: 976; None; RadiusNeighbourFinder {'_radius': 0.05, '_max_num_neighbors': 25, '_conv_type': 'partial_dense'}
(kp_conv): KPConvLayer(InF: 8, OutF: 8, kernel_pts: 15, radius: 0.03, KP_influence: linear, Add_one: False)
(bn): FastBatchNorm1d(
(batch_norm): BatchNorm1d(8, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(activation): LeakyReLU(negative_slope=0.1)
)
(unary_1): Sequential(
(0): Linear(in_features=16, out_features=8, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(8, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(unary_2): Sequential(
(0): Linear(in_features=8, out_features=32, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(32, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(shortcut_op): Sequential(
(0): Linear(in_features=16, out_features=32, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(32, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
)
(activation): LeakyReLU(negative_slope=0.1)
)
)
)
(1): KPDualBlock(
Nb parameters: 9312
(blocks): ModuleList(
(0): ResnetBBlock(
Nb parameters: 1568
(kp_conv): SimpleBlock(
Nb parameters: 976; GridSampling3D(grid_size=0.04, quantize_coords=False, mode=mean); RadiusNeighbourFinder {'_radius': 0.05, '_max_num_neighbors': 25, '_conv_type': 'partial_dense'}
(kp_conv): KPConvLayer(InF: 8, OutF: 8, kernel_pts: 15, radius: 0.03, KP_influence: linear, Add_one: False)
(bn): FastBatchNorm1d(
(batch_norm): BatchNorm1d(8, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(activation): LeakyReLU(negative_slope=0.1)
)
(unary_1): Sequential(
(0): Linear(in_features=32, out_features=8, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(8, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(unary_2): Sequential(
(0): Linear(in_features=8, out_features=32, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(32, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(shortcut_op): Identity()
(activation): LeakyReLU(negative_slope=0.1)
)
(1): ResnetBBlock(
Nb parameters: 7744
(kp_conv): SimpleBlock(
Nb parameters: 3872; None; RadiusNeighbourFinder {'_radius': 0.1, '_max_num_neighbors': 25, '_conv_type': 'partial_dense'}
(kp_conv): KPConvLayer(InF: 16, OutF: 16, kernel_pts: 15, radius: 0.06, KP_influence: linear, Add_one: False)
(bn): FastBatchNorm1d(
(batch_norm): BatchNorm1d(16, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(activation): LeakyReLU(negative_slope=0.1)
)
(unary_1): Sequential(
(0): Linear(in_features=32, out_features=16, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(16, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(unary_2): Sequential(
(0): Linear(in_features=16, out_features=64, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(64, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(shortcut_op): Sequential(
(0): Linear(in_features=32, out_features=64, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(64, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
)
(activation): LeakyReLU(negative_slope=0.1)
)
)
)
(2): KPDualBlock(
Nb parameters: 36416
(blocks): ModuleList(
(0): ResnetBBlock(
Nb parameters: 6080
(kp_conv): SimpleBlock(
Nb parameters: 3872; GridSampling3D(grid_size=0.08, quantize_coords=False, mode=mean); RadiusNeighbourFinder {'_radius': 0.1, '_max_num_neighbors': 25, '_conv_type': 'partial_dense'}
(kp_conv): KPConvLayer(InF: 16, OutF: 16, kernel_pts: 15, radius: 0.06, KP_influence: linear, Add_one: False)
(bn): FastBatchNorm1d(
(batch_norm): BatchNorm1d(16, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(activation): LeakyReLU(negative_slope=0.1)
)
(unary_1): Sequential(
(0): Linear(in_features=64, out_features=16, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(16, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(unary_2): Sequential(
(0): Linear(in_features=16, out_features=64, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(64, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(shortcut_op): Identity()
(activation): LeakyReLU(negative_slope=0.1)
)
(1): ResnetBBlock(
Nb parameters: 30336
(kp_conv): SimpleBlock(
Nb parameters: 15424; None; RadiusNeighbourFinder {'_radius': 0.2, '_max_num_neighbors': 25, '_conv_type': 'partial_dense'}
(kp_conv): KPConvLayer(InF: 32, OutF: 32, kernel_pts: 15, radius: 0.12, KP_influence: linear, Add_one: False)
(bn): FastBatchNorm1d(
(batch_norm): BatchNorm1d(32, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(activation): LeakyReLU(negative_slope=0.1)
)
(unary_1): Sequential(
(0): Linear(in_features=64, out_features=32, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(32, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(unary_2): Sequential(
(0): Linear(in_features=32, out_features=128, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(128, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(shortcut_op): Sequential(
(0): Linear(in_features=64, out_features=128, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(128, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
)
(activation): LeakyReLU(negative_slope=0.1)
)
)
)
(3): KPDualBlock(
Nb parameters: 144000
(blocks): ModuleList(
(0): ResnetBBlock(
Nb parameters: 23936
(kp_conv): SimpleBlock(
Nb parameters: 15424; GridSampling3D(grid_size=0.16, quantize_coords=False, mode=mean); RadiusNeighbourFinder {'_radius': 0.2, '_max_num_neighbors': 25, '_conv_type': 'partial_dense'}
(kp_conv): KPConvLayer(InF: 32, OutF: 32, kernel_pts: 15, radius: 0.12, KP_influence: linear, Add_one: False)
(bn): FastBatchNorm1d(
(batch_norm): BatchNorm1d(32, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(activation): LeakyReLU(negative_slope=0.1)
)
(unary_1): Sequential(
(0): Linear(in_features=128, out_features=32, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(32, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(unary_2): Sequential(
(0): Linear(in_features=32, out_features=128, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(128, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(shortcut_op): Identity()
(activation): LeakyReLU(negative_slope=0.1)
)
(1): ResnetBBlock(
Nb parameters: 120064
(kp_conv): SimpleBlock(
Nb parameters: 61568; None; RadiusNeighbourFinder {'_radius': 0.4, '_max_num_neighbors': 25, '_conv_type': 'partial_dense'}
(kp_conv): KPConvLayer(InF: 64, OutF: 64, kernel_pts: 15, radius: 0.24, KP_influence: linear, Add_one: False)
(bn): FastBatchNorm1d(
(batch_norm): BatchNorm1d(64, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(activation): LeakyReLU(negative_slope=0.1)
)
(unary_1): Sequential(
(0): Linear(in_features=128, out_features=64, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(64, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(unary_2): Sequential(
(0): Linear(in_features=64, out_features=256, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(256, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(shortcut_op): Sequential(
(0): Linear(in_features=128, out_features=256, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(256, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
)
(activation): LeakyReLU(negative_slope=0.1)
)
)
)
(4): KPDualBlock(
Nb parameters: 572672
(blocks): ModuleList(
(0): ResnetBBlock(
Nb parameters: 94976
(kp_conv): SimpleBlock(
Nb parameters: 61568; GridSampling3D(grid_size=0.32, quantize_coords=False, mode=mean); RadiusNeighbourFinder {'_radius': 0.4, '_max_num_neighbors': 25, '_conv_type': 'partial_dense'}
(kp_conv): KPConvLayer(InF: 64, OutF: 64, kernel_pts: 15, radius: 0.24, KP_influence: linear, Add_one: False)
(bn): FastBatchNorm1d(
(batch_norm): BatchNorm1d(64, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(activation): LeakyReLU(negative_slope=0.1)
)
(unary_1): Sequential(
(0): Linear(in_features=256, out_features=64, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(64, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(unary_2): Sequential(
(0): Linear(in_features=64, out_features=256, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(256, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(shortcut_op): Identity()
(activation): LeakyReLU(negative_slope=0.1)
)
(1): ResnetBBlock(
Nb parameters: 477696
(kp_conv): SimpleBlock(
Nb parameters: 246016; None; RadiusNeighbourFinder {'_radius': 0.8, '_max_num_neighbors': 25, 'conv_type': 'partial_dense'}
(kp_conv): KPConvLayer(InF: 128, OutF: 128, kernel_pts: 15, radius: 0.48, KP_influence: linear, Add_one: False)
(bn): FastBatchNorm1d(
(batch_norm): BatchNorm1d(128, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(activation): LeakyReLU(negative_slope=0.1)
)
(unary_1): Sequential(
(0): Linear(in_features=256, out_features=128, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(128, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(unary_2): Sequential(
(0): Linear(in_features=128, out_features=512, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(512, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(shortcut_op): Sequential(
(0): Linear(in_features=256, out_features=512, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(512, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
)
(activation): LeakyReLU(negative_slope=0.1)
)
)
)
)
(inner_modules): ModuleList(
(0): GlobalBaseModule(
(nn): Sequential(
(0): Sequential(
(0): Linear(in_features=515, out_features=512, bias=True)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.2)
)
)
)
)
(up_modules): ModuleList()
)/cluster/scratch/bxiang/envtorch19py38/lib/python3.8/site-packages/torch_scatter/scatter.py:58: UserWarning: floor_divide is deprecated, and will be removed in a future version of pytorch. It currently rounds toward 0 (like the 'trunc' function NOT 'floor'). This results in incorrect rounding for negative values.
To keep the current behavior, use torch.div(a, b, rounding_mode='trunc'), or for actual floor division, use torch.div(a, b, rounding_mode='floor'). (Triggered internally at /pytorch/aten/src/ATen/native/BinaryOps.cpp:450.)
out.floor_divide(count)
ERROR
test_minkowski (test.test_api.TestAPIEncoder) ... Minkowski API is deprecated in favor of the SparseConv3d API. It should be a simple drop in replacement (no change to the API).
/cluster/scratch/bxiang/envtorch19py38/lib/python3.8/site-packages/torch/autograd/function.py:144: DeprecationWarning: <class 'MinkowskiConvolution.MinkowskiConvolutionFunction'> should not be instantiated. Methods on autograd functionsare all static, so you should invoke them on the class itself. Instantiating an autograd function will raise an error in a future version of PyTorch.
warnings.warn(f"{cls} should not be instantiated. Methods on autograd functions"
Minkowski API is deprecated in favor of the SparseConv3d API. It should be a simple drop in replacement (no change to the API).
ok
test_pn2 (test.test_api.TestAPIEncoder) ... /cluster/scratch/bxiang/envtorch19py38/lib/python3.8/site-packages/torch/nn/functional.py:718: UserWarning: Named tensors and all their associated APIs are an experimental feature and subject to change. Please do not use them for anything important until they are released as stable. (Triggered internally at /pytorch/c10/core/TensorImpl.h:1156.)
return torch.max_pool2d(input, kernel_size, stride, padding, dilation, ceil_mode)
ok
test_rsconv (test.test_api.TestAPIEncoder) ... The down_conv_nn has a different size as radii. Make sure of have SharedRSConv
ok
test_kpconv (test.test_api.TestAPIUnet) ...
Model failing:
KPConvUnet(
(down_modules): ModuleList(
(0): KPDualBlock(
Nb parameters: 9728
(blocks): ModuleList(
(0): SimpleBlock(
Nb parameters: 1984; None; RadiusNeighbourFinder {'_radius': 0.05, '_max_num_neighbors': 25, '_conv_type': 'partial_dense'}
(kp_conv): KPConvLayer(InF: 4, OutF: 32, kernel_pts: 15, radius: 0.03, KP_influence: linear, Add_one: False)
(bn): FastBatchNorm1d(
(batch_norm): BatchNorm1d(32, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(activation): LeakyReLU(negative_slope=0.1)
)
(1): ResnetBBlock(
Nb parameters: 7744
(kp_conv): SimpleBlock(
Nb parameters: 3872; None; RadiusNeighbourFinder {'_radius': 0.05, '_max_num_neighbors': 25, '_conv_type': 'partial_dense'}
(kp_conv): KPConvLayer(InF: 16, OutF: 16, kernel_pts: 15, radius: 0.03, KP_influence: linear, Add_one: False)
(bn): FastBatchNorm1d(
(batch_norm): BatchNorm1d(16, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(activation): LeakyReLU(negative_slope=0.1)
)
(unary_1): Sequential(
(0): Linear(in_features=32, out_features=16, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(16, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(unary_2): Sequential(
(0): Linear(in_features=16, out_features=64, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(64, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(shortcut_op): Sequential(
(0): Linear(in_features=32, out_features=64, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(64, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
)
(activation): LeakyReLU(negative_slope=0.1)
)
)
)
(1): KPDualBlock(
Nb parameters: 36416
(blocks): ModuleList(
(0): ResnetBBlock(
Nb parameters: 6080
(kp_conv): SimpleBlock(
Nb parameters: 3872; GridSampling3D(grid_size=0.04, quantize_coords=False, mode=mean); RadiusNeighbourFinder {'_radius': 0.05, '_max_num_neighbors': 25, '_conv_type': 'partial_dense'}
(kp_conv): KPConvLayer(InF: 16, OutF: 16, kernel_pts: 15, radius: 0.03, KP_influence: linear, Add_one: False)
(bn): FastBatchNorm1d(
(batch_norm): BatchNorm1d(16, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(activation): LeakyReLU(negative_slope=0.1)
)
(unary_1): Sequential(
(0): Linear(in_features=64, out_features=16, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(16, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(unary_2): Sequential(
(0): Linear(in_features=16, out_features=64, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(64, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(shortcut_op): Identity()
(activation): LeakyReLU(negative_slope=0.1)
)
(1): ResnetBBlock(
Nb parameters: 30336
(kp_conv): SimpleBlock(
Nb parameters: 15424; None; RadiusNeighbourFinder {'_radius': 0.1, '_max_num_neighbors': 25, '_conv_type': 'partial_dense'}
(kp_conv): KPConvLayer(InF: 32, OutF: 32, kernel_pts: 15, radius: 0.06, KP_influence: linear, Add_one: False)
(bn): FastBatchNorm1d(
(batch_norm): BatchNorm1d(32, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(activation): LeakyReLU(negative_slope=0.1)
)
(unary_1): Sequential(
(0): Linear(in_features=64, out_features=32, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(32, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(unary_2): Sequential(
(0): Linear(in_features=32, out_features=128, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(128, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(shortcut_op): Sequential(
(0): Linear(in_features=64, out_features=128, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(128, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
)
(activation): LeakyReLU(negative_slope=0.1)
)
)
)
(2): KPDualBlock(
Nb parameters: 144000
(blocks): ModuleList(
(0): ResnetBBlock(
Nb parameters: 23936
(kp_conv): SimpleBlock(
Nb parameters: 15424; GridSampling3D(grid_size=0.08, quantize_coords=False, mode=mean); RadiusNeighbourFinder {'_radius': 0.1, '_max_num_neighbors': 25, '_conv_type': 'partial_dense'}
(kp_conv): KPConvLayer(InF: 32, OutF: 32, kernel_pts: 15, radius: 0.06, KP_influence: linear, Add_one: False)
(bn): FastBatchNorm1d(
(batch_norm): BatchNorm1d(32, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(activation): LeakyReLU(negative_slope=0.1)
)
(unary_1): Sequential(
(0): Linear(in_features=128, out_features=32, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(32, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(unary_2): Sequential(
(0): Linear(in_features=32, out_features=128, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(128, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(shortcut_op): Identity()
(activation): LeakyReLU(negative_slope=0.1)
)
(1): ResnetBBlock(
Nb parameters: 120064
(kp_conv): SimpleBlock(
Nb parameters: 61568; None; RadiusNeighbourFinder {'_radius': 0.2, '_max_num_neighbors': 25, '_conv_type': 'partial_dense'}
(kp_conv): KPConvLayer(InF: 64, OutF: 64, kernel_pts: 15, radius: 0.12, KP_influence: linear, Add_one: False)
(bn): FastBatchNorm1d(
(batch_norm): BatchNorm1d(64, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(activation): LeakyReLU(negative_slope=0.1)
)
(unary_1): Sequential(
(0): Linear(in_features=128, out_features=64, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(64, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(unary_2): Sequential(
(0): Linear(in_features=64, out_features=256, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(256, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(shortcut_op): Sequential(
(0): Linear(in_features=128, out_features=256, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(256, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
)
(activation): LeakyReLU(negative_slope=0.1)
)
)
)
(3): KPDualBlock(
Nb parameters: 572672
(blocks): ModuleList(
(0): ResnetBBlock(
Nb parameters: 94976
(kp_conv): SimpleBlock(
Nb parameters: 61568; GridSampling3D(grid_size=0.16, quantize_coords=False, mode=mean); RadiusNeighbourFinder {'_radius': 0.2, '_max_num_neighbors': 25, '_conv_type': 'partial_dense'}
(kp_conv): KPConvLayer(InF: 64, OutF: 64, kernel_pts: 15, radius: 0.12, KP_influence: linear, Add_one: False)
(bn): FastBatchNorm1d(
(batch_norm): BatchNorm1d(64, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(activation): LeakyReLU(negative_slope=0.1)
)
(unary_1): Sequential(
(0): Linear(in_features=256, out_features=64, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(64, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(unary_2): Sequential(
(0): Linear(in_features=64, out_features=256, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(256, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(shortcut_op): Identity()
(activation): LeakyReLU(negative_slope=0.1)
)
(1): ResnetBBlock(
Nb parameters: 477696
(kp_conv): SimpleBlock(
Nb parameters: 246016; None; RadiusNeighbourFinder {'_radius': 0.4, '_max_num_neighbors': 25, '_conv_type': 'partial_dense'}
(kp_conv): KPConvLayer(InF: 128, OutF: 128, kernel_pts: 15, radius: 0.24, KP_influence: linear, Add_one: False)
(bn): FastBatchNorm1d(
(batch_norm): BatchNorm1d(128, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(activation): LeakyReLU(negative_slope=0.1)
)
(unary_1): Sequential(
(0): Linear(in_features=256, out_features=128, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(128, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(unary_2): Sequential(
(0): Linear(in_features=128, out_features=512, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(512, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(shortcut_op): Sequential(
(0): Linear(in_features=256, out_features=512, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(512, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
)
(activation): LeakyReLU(negative_slope=0.1)
)
)
)
(4): KPDualBlock(
Nb parameters: 2284032
(blocks): ModuleList(
(0): ResnetBBlock(
Nb parameters: 378368
(kp_conv): SimpleBlock(
Nb parameters: 246016; GridSampling3D(grid_size=0.32, quantize_coords=False, mode=mean); RadiusNeighbourFinder {'_radius': 0.4, '_max_num_neighbors': 25, '_conv_type': 'partial_dense'}
(kp_conv): KPConvLayer(InF: 128, OutF: 128, kernel_pts: 15, radius: 0.24, KP_influence: linear, Add_one: False)
(bn): FastBatchNorm1d(
(batch_norm): BatchNorm1d(128, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(activation): LeakyReLU(negative_slope=0.1)
)
(unary_1): Sequential(
(0): Linear(in_features=512, out_features=128, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(128, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(unary_2): Sequential(
(0): Linear(in_features=128, out_features=512, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(512, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(shortcut_op): Identity()
(activation): LeakyReLU(negative_slope=0.1)
)
(1): ResnetBBlock(
Nb parameters: 1905664
(kp_conv): SimpleBlock(
Nb parameters: 983552; None; RadiusNeighbourFinder {'_radius': 0.8, '_max_num_neighbors': 25, '_conv_type': 'partial_dense'}
(kp_conv): KPConvLayer(InF: 256, OutF: 256, kernel_pts: 15, radius: 0.48, KP_influence: linear, Add_one: False)
(bn): FastBatchNorm1d(
(batch_norm): BatchNorm1d(256, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(activation): LeakyReLU(negative_slope=0.1)
)
(unary_1): Sequential(
(0): Linear(in_features=512, out_features=256, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(256, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(unary_2): Sequential(
(0): Linear(in_features=256, out_features=1024, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(1024, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.1)
)
(shortcut_op): Sequential(
(0): Linear(in_features=512, out_features=1024, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(1024, eps=1e-05, momentum=0.02, affine=True, track_running_stats=True)
)
)
(activation): LeakyReLU(negative_slope=0.1)
)
)
)
)
(inner_modules): ModuleList(
(0): Identity()
)
(up_modules): ModuleList(
(0): FPModule_PD(
(nn): Sequential(
(0): Sequential(
(0): Linear(in_features=1536, out_features=256, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(256, eps=1e-05, momentum=0.2, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.2)
)
)
)
(1): FPModule_PD(
(nn): Sequential(
(0): Sequential(
(0): Linear(in_features=512, out_features=128, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(128, eps=1e-05, momentum=0.2, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.2)
)
)
)
(2): FPModule_PD(
(nn): Sequential(
(0): Sequential(
(0): Linear(in_features=256, out_features=64, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(64, eps=1e-05, momentum=0.2, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.2)
)
)
)
(3): FPModule_PD(
(nn): Sequential(
(0): Sequential(
(0): Linear(in_features=128, out_features=32, bias=False)
(1): FastBatchNorm1d(
(batch_norm): BatchNorm1d(32, eps=1e-05, momentum=0.2, affine=True, track_running_stats=True)
)
(2): LeakyReLU(negative_slope=0.2)
)
)
)
)
)ERROR
test_pn2 (test.test_api.TestAPIUnet) ... ok
test_rsconv (test.test_api.TestAPIUnet) ... The down_conv_nn has a different size as radii. Make sure of have SharedRSConv
ok
test_sparseconv3d (test.test_api.TestAPIUnet) ... /cluster/scratch/bxiang/envtorch19py38/lib/python3.8/site-packages/torch/autograd/function.py:144: DeprecationWarning: <class 'MinkowskiConvolution.MinkowskiConvolutionTransposeFunction'> should not be instantiated. Methods on autograd functionsare all static, so you should invoke them on the class itself. Instantiating an autograd function will raise an error in a future version of PyTorch.
warnings.warn(f"{cls} should not be instantiated. Methods on autograd functions"
ok
test_get_sample (test.test_basedataset.TestBatchCollate) ... FAIL
test_num_batches (test.test_basedataset.TestBatchCollate) ... ok
test_add_weights (test.test_basedataset.TestDataset) ... ok
test_empty_dataset (test.test_basedataset.TestDataset) ... ok
test_get_by_name (test.test_basedataset.TestDataset) ... ok
test_multiple_test_datasets (test.test_basedataset.TestDataset) ... ok
test_normal (test.test_basedataset.TestDataset) ... ok
test_simple_datasets (test.test_basedataset.TestDataset) ... ok
test_accumulated_gradient (test.test_basemodel.TestBaseModel) ... /cluster/scratch/bxiang/envtorch19py38/lib/python3.8/site-packages/omegaconf/omegaconf.py:572: UserWarning: update() merge flag is is not specified, defaulting to False.
For more details, see omry/omegaconf#367
warnings.warn(
ok
test_enable_dropout_eval (test.test_basemodel.TestBaseModel) ... ok
test_getinput (test.test_basemodel.TestBaseModel) ... ok
test_load_pretrained_model (test.test_basemodel.TestBaseModel)
test load_state_dict_with_same_shape ... ok
test_validatedata (test.test_basemodel.TestBaseModel) ... ok
test_fromlist (test.test_batch.TestSimpleBatch) ... ok
test_scheduler (test.test_bn_scheduler.TestBNMomentumScheduler) ... ok
test_evaldetection (test.test_boxstuff.TestAP) ...
tensor([0.1033, 0.0699, 0.1033, 0.1033, 0.1033, 0.1033, 0.1033, 0.1033, 0.1033,
0.1033])
/cluster/shadow/.lsbatch/1632731258.185285678: line 8: 33270 Segmentation fault (core dumped) python -m unittest -v

Do you have any idea how to fix it? I tried install it by pip and poetry. Both did not solve it... Look forwad to someone's help. Thanks a lot!

[bxiang233]

Build torch-sparse and torch-scatter from source solved my problems.