Getting the following error when trying to run RainSynAll100CUDA_VISIBLE_DEVICES=0 bash ./tools/dist_test.sh configs/derainers/RainMamba/RainSynAll100.py "RainSynAll100.pth" 1 --save-path '../results/RainSynAll100/'
File "/RainMamba/code/RainSynAll100/modules/mambablock.py", line 19, in forward
x = x.transpose(1, 2).contiguous().reshape(B, C, nf, H, W)
RuntimeError: shape '[1, 256, 3, 64, 64]' is invalid for input of size 1310720
Was able to train the model without any problems.
By the way, is it possible to make the pre-trained models publicly available? Currently they require approval.
The reason for the error is that the size of the input image and the order of the curve are inconsistent.
You can use the newly revised code below
RainMamba/code/RainSynAll100/mmedit/models/derainers/DrainNet. py.
Thank you for your reminder! I used to think they were already accessible because they could be accessed normally on my device. I will fix this issue immediately.
Getting the following error when trying to run
RainSynAll100
CUDA_VISIBLE_DEVICES=0 bash ./tools/dist_test.sh configs/derainers/RainMamba/RainSynAll100.py "RainSynAll100.pth" 1 --save-path '../results/RainSynAll100/'
Was able to train the model without any problems.
By the way, is it possible to make the pre-trained models publicly available? Currently they require approval.
Thanks!