RuntimeError: Expected isFloatingType(grads[i].type().scalarType()) to be true, but got false.

[mtli77]

Hi, here

I miss this error:

Traceback (most recent call last):
  File "train.py", line 331, in <module>
    train(start_epoch)
  File "train.py", line 316, in train
    train_one_epoch()
  File "train.py", line 234, in train_one_epoch
    loss.backward()
  File "/home/ubuntu/.conda/envs/tia37/lib/python3.7/site-packages/torch/tensor.py", line 195, in backward
    torch.autograd.backward(self, gradient, retain_graph, create_graph)
  File "/home/ubuntu/.conda/envs/tia37/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.)

Could you please give me some advice about this bug? Thank you!

[iamwilliamli]

The problem is solved, when PyTorch version >= 1.3.0, it is required to add mark_non_differentiable() must be used to tell the engine if an output is not differentiable. https://pytorch.org/docs/master/notes/extending.html

Solution: https://github.com/facebookresearch/votenet/commit/c330998f04ed8cae5d4d724df26bd3af1672f6b6