RuntimeError: expected backend CPU and dtype Float but got backend CPU and dtype Long

[PRAYALANKAR]

File "torchcv-master/examples/fpnssd/train.py", line 37, in
box_coder = FPNSSDBoxCoder()
File "/torchcv-master/torchcv/models/fpnssd/box_coder.py", line 15, in init
self.anchor_boxes = self._get_anchor_boxes(input_size=torch.tensor([512.,512.]))
File "torchcv-master/torchcv/models/fpnssd/box_coder.py", line 57, in _get_anchor_boxes
xy = (xy * grid_size)
RuntimeError: expected backend CPU and dtype Float but got backend CPU and dtype Long
i am getting this any suggestion or anyone faced the issue and solved it.

[kritiksoman]

I am also having the same issue.

[PRAYALANKAR]

Hi
I solved the issue by making the following changes.
in meshgrid.py make following changes into the meshgrid function so that return value of the function is also float.
def meshgrid(x, y, row_major=True):
aa = x * 1.
bb = y * 1.
a = torch.arange(0,aa)
b = torch.arange(0,bb)
xx = a.repeat(y).view(-1,1)
yy = b.view(-1,1).repeat(1,x).view(-1,1)
return torch.cat([xx,yy],1) if row_major else torch.cat([yy,xx],1)

[deepzlk]

grid_size=grid_size.long()
floattensor->longtensor