When I swapped the 3D network for my own, some problems occurred

[Jaywxy]

When I swapped the 3D network for my own, some problems occurred：

  File "/mnt/Mhd_4T/2.code/2DPASS/network/2dpass-repvgg3d.py", line 183, in forward
    data_dict = self.fusion(data_dict)
  File "/home/wxy/anaconda3/envs/spvnas/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl
    return forward_call(*input, **kwargs)
  File "/mnt/Mhd_4T/2.code/2DPASS/network/2dpass-repvgg3d.py", line 142, in forward
    singlescale_loss, fuse_feat = self.fusion_to_single_KD(data_dict, idx)
  File "/mnt/Mhd_4T/2.code/2DPASS/network/2dpass-repvgg3d.py", line 123, in fusion_to_single_KD
    seg_loss_3d = self.seg_loss(pts_pred_full, pts_label_full)
  File "/mnt/Mhd_4T/2.code/2DPASS/network/2dpass-repvgg3d.py", line 92, in seg_loss
    ce_loss = self.ce_loss(logits, labels)
  File "/home/wxy/anaconda3/envs/spvnas/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1110, in _call_impl
    return forward_call(*input, **kwargs)
  File "/home/wxy/anaconda3/envs/spvnas/lib/python3.7/site-packages/torch/nn/modules/loss.py", line 1165, in forward
    label_smoothing=self.label_smoothing)
  File "/home/wxy/anaconda3/envs/spvnas/lib/python3.7/site-packages/torch/nn/functional.py", line 2996, in cross_entropy
    return torch._C._nn.cross_entropy_loss(input, target, weight, _Reduction.get_enum(reduction), ignore_index, label_smoothing)
ValueError: Expected input batch_size (26166) to match target batch_size (26167).

This seems to be the reason for the inconsistency in size when seg_loss_3d = self.seg_loss(pts_pred_full, pts_label_full), but I don't understand why this happens, the grid division part and scales are both using your code, only modifying the 3D network part, hoping to get your reply,thanks!