Hello,
I am trying to reproduce VoxelNeXt's results for varying sparsity pruning ratios on the NuScenes Mini dataset. However, I am having an issue with the weights loading properly (see image showing loaded 512/542). I was wondering if there was a pre-trained model available for the SPS 3D backbone (https://github.com/dvlab-research/VoxelNeXt/blob/master/pcdet/models/backbones_3d/spconv_backbone_voxelnext_sps.py) or if there is perhaps something I am doing wrong that would fix my issue.
Thank you so much and I appreciate your work!
