justchenhao
Memory error
I get a memory error for both BAM and PAM at line 169 of train.py. Does it really need that much memory or am I missing something?
Traceback (most recent call last):
File "train.py", line 169, in
I get a memory error for both BAM and PAM at line 169 of train.py. Does it really need that much memory or am I missing something?
Traceback (most recent call last): File "train.py", line 169, in miou_current = val(opt, model) File "train.py", line 86, in val score = model.test(val=True) # run inference File "/home/supervisedCD/STANet/models/CDFA_model.py", line 72, in test self.forward() File "/home/supervisedCD/STANet/models/CDFA_model.py", line 90, in forward self.feat_A, self.feat_B = self.netA(self.feat_A,self.feat_B) File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, **kwargs) File "/home/supervisedCD/STANet/models/backbone.py", line 46, in forward x = self.Self_Att(x) File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, **kwargs) File "/home/supervisedCD/STANet/models/BAM.py", line 37, in forward energy = torch.bmm(proj_query, proj_key) # transpose check RuntimeError: CUDA out of memory. Tried to allocate 256.00 GiB (GPU 0; 23.70 GiB total capacity; 825.29 MiB already allocated; 18.55 GiB free; 3.90 GiB reserved in total by PyTorch)
I also encountered the same problem in the same place recently. Have you solved it now?
I get a memory error for both BAM and PAM at line 169 of train.py. Does it really need that much memory or am I missing something?
Traceback (most recent call last): File "train.py", line 169, in miou_current = val(opt, model) File "train.py", line 86, in val score = model.test(val=True) # run inference File "/home/supervisedCD/STANet/models/CDFA_model.py", line 72, in test self.forward() File "/home/supervisedCD/STANet/models/CDFA_model.py", line 90, in forward self.feat_A, self.feat_B = self.netA(self.feat_A,self.feat_B) File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, **kwargs) File "/home/supervisedCD/STANet/models/backbone.py", line 46, in forward x = self.Self_Att(x) File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl result = self.forward(*input, **kwargs) File "/home/supervisedCD/STANet/models/BAM.py", line 37, in forward energy = torch.bmm(proj_query, proj_key) # transpose check RuntimeError: CUDA out of memory. Tried to allocate 256.00 GiB (GPU 0; 23.70 GiB total capacity; 825.29 MiB already allocated; 18.55 GiB free; 3.90 GiB reserved in total by PyTorch)
I have good news for you and share it after I post a comment message. I can run it！ There will be a verification process automatically after one round of training. But there is no cropped code in the verification process, you need to crop the val picture to 256 by yourself before training.
你可以看我的仓库，我对图片裁剪进行了复现，裁剪后的图片存放在与原数据文件夹相同的子目录中。https://github.com/kangpeilun/utils-for-img-process/tree/main
你可以看我的仓库，我对图片裁剪进行了复现，裁剪后的图片存放在与原数据文件夹相同的子目录中。https://github.com/kangpeilun/utils-for-img-process/tree/main
That's cool.Thank you.
