RuntimeError: Expected isFloatingType(grads[i].type().scalarType()) to be true, but got false. (Could this error message be improved? If so, please report an enhancement request to PyTorch.)

[Yan haixu]

我在python train.py的时候，发生这个报错

File "train.py", line 375, in
train(start_epoch)
File "train.py", line 352, in train
train_one_epoch()
File "train.py", line 247, in train_one_epoch
loss.backward()
File "/home/hova/anaconda3/lib/python3.7/site-packages/torch/tensor.py", line 166, in backward
torch.autograd.backward(self, gradient, retain_graph, create_graph)
File "/home/hova/anaconda3/lib/python3.7/site-packages/torch/autograd/init.py", line 99, in backward
allow_unreachable=True) # allow_unreachable flag
RuntimeError: Expected isFloatingType(grads[i].type().scalarType()) to be true, but got false. (Could this error message be improved? If so, please report an enhancement request to PyTorch.)

哥哥救我Orz

[Yan haixu]

✅[Solved]

THE SAME ISSUE AS THIS

The pointnet2/* applied here is an earlier version.
Add ctx.mark_non_differentiable(new_ind) at the end of forward function of FurthestPointSampling in pointnet2/pointnet2_utils.py

OR replace the whole `pointnet2* folder with THIS