CUDA out of memory

[sumit2k4]

I preproceesed the data in matlab and generated the model file. Now in python i am getting this error:

############################################################

Video Super Resolution - Pytorch implementation

by Thang Vu (thangvubk@gmail.com

############################################################

-------YOUR SETTINGS_________
model: VRES
model_path: VRES_x3.pt
scale: 3
test_set: IndMya

Contructing dataset...
Testing...
Traceback (most recent call last):
File "/media/sumit/New Volume1/RnD/video-super-resolution-master/test.py", line 93, in
main()
File "/media/sumit/New Volume1/RnD/video-super-resolution-master/test.py", line 88, in main
stats, outputs = solver.test(train_dataset, args.model_path)
File "/media/sumit/New Volume1/RnD/video-super-resolution-master/solver.py", line 252, in test
_, _, stats, outputs = self._check_PSNR(dataset, is_test=True)
File "/media/sumit/New Volume1/RnD/video-super-resolution-master/solver.py", line 151, in _check_PSNR
output_batch = self.model(input_batch)
File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 493, in call
result = self.forward(*input, **kwargs)
File "/media/sumit/New Volume1/RnD/video-super-resolution-master/model.py", line 81, in forward
out = self.residual_layer(out)
File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 493, in call
result = self.forward(*input, **kwargs)
File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/container.py", line 92, in forward
input = module(input)
File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 493, in call
result = self.forward(*input, **kwargs)
File "/media/sumit/New Volume1/RnD/video-super-resolution-master/model.py", line 94, in forward
return self.relu(self.conv(x))
File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 493, in call
result = self.forward(*input, **kwargs)
File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/conv.py", line 338, in forward
self.padding, self.dilation, self.groups)
RuntimeError: CUDA out of memory. Tried to allocate 128.00 MiB (GPU 0; 3.94 GiB total capacity; 3.12 GiB already allocated; 7.38 MiB free; 48.66 MiB cached)

My GPU is 4GB. Though I changed the batch size to 1 and number of workers to 0, I am getting out of memory error while performing both training and testing. May I know the GPU requirement please? or is there any way that I can process it with my current GPU?

[thangvubk]

I trained the networks on a 12GB GPU. You may reduce batch size if your GPU mem is not large enough. Check here