Sizes of tensors must match except in dimension 2. Got 60 and 120 (The offending index is 0)
Getting a shape mismatch for directories despite the same exact image characteristics across all images. It also seems sporadic as other folders with similar creation methods do not have this issue. Any help identifying the cause would be helpful.
- I have also confirmed that every single image in the folder works when using correct_batch.
- I think this issue was raised in this Issue, however the delta is much larger than they observed
Traceback (most recent call last):
File "experiments/correct_batch.py", line 62, in <module>
y_r, cb_r, cr_r = net(y_q, y, c_q, cb, cr)
File "/opt/conda/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
result = self.forward(*input, **kwargs)
File "/data1/vinayak/quantization-guided-ac/model/deartifacting_network.py", line 88, in forward
r_cb = self.color_net(q_y, q_c, restored, cb)
File "/opt/conda/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
result = self.forward(*input, **kwargs)
File "/data1/vinayak/quantization-guided-ac/model/color_restore.py", line 42, in forward
d = self.block_enhancer2(torch.cat([d, y], dim=1))
RuntimeError: Sizes of tensors must match except in dimension 2. Got 60 and 120 (The offending index is 0)
root@7743ed28fae9:/data1/vinayak/quantization-guided-ac#
P.S Great work on the code though. Its well structured and mostly easy to use and I found your paper very interesting.