RuntimeError: memory format option is only supported by strided tensors
ljyljy1216 opened this issue · comments
When I applied the model to my own task, the following error occurred. But I have strided tensors for each one, and I still get the following error. Does anyone know why that is
Traceback (most recent call last): File "/mnt/c/Users/caica/Desktop/swin/classes/fc4/train.py", line 283, in <module> main(opt) File "/mnt/c/Users/caica/Desktop/swin/classes/fc4/train.py", line 105, in main loss = model.optimize(img, label) File "/mnt/c/Users/caica/Desktop/swin/classes/fc4/Model_FC4.py", line 40, in optimize pred = self.predict(img) File "/mnt/c/Users/caica/Desktop/swin/classes/fc4/Model_FC4.py", line 32, in predict pred, rgb, confidence = self._network(img) File "/home/liujunyi/anaconda3/envs/vmamba2/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl return self._call_impl(*args, **kwargs) File "/home/liujunyi/anaconda3/envs/vmamba2/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl return forward_call(*args, **kwargs) File "/mnt/c/Users/caica/Desktop/swin/classes/fc4/FC_4.py", line 69, in forward x = self.backbone(x) File "/home/liujunyi/anaconda3/envs/vmamba2/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl return self._call_impl(*args, **kwargs) File "/home/liujunyi/anaconda3/envs/vmamba2/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl return forward_call(*args, **kwargs) File "/mnt/c/Users/caica/Desktop/swin/classes/fc4/VMamba_main/classification/models/vmamba.py", line 1514, in forward o, x = layer_forward(layer, x) # (B, H, W, C) File "/mnt/c/Users/caica/Desktop/swin/classes/fc4/VMamba_main/classification/models/vmamba.py", line 1507, in layer_forward x = l.blocks(x) File "/home/liujunyi/anaconda3/envs/vmamba2/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl return self._call_impl(*args, **kwargs) File "/home/liujunyi/anaconda3/envs/vmamba2/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl return forward_call(*args, **kwargs) File "/home/liujunyi/anaconda3/envs/vmamba2/lib/python3.10/site-packages/torch/nn/modules/container.py", line 217, in forward input = module(input) File "/home/liujunyi/anaconda3/envs/vmamba2/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl return self._call_impl(*args, **kwargs) File "/home/liujunyi/anaconda3/envs/vmamba2/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl return forward_call(*args, **kwargs) File "/mnt/c/Users/caica/Desktop/swin/classes/fc4/VMamba_main/classification/models/vmamba.py", line 1161, in forward return self._forward(input) File "/mnt/c/Users/caica/Desktop/swin/classes/fc4/VMamba_main/classification/models/vmamba.py", line 1149, in _forward x = x + self.drop_path(self.op(self.norm(x))) File "/home/liujunyi/anaconda3/envs/vmamba2/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl return self._call_impl(*args, **kwargs) File "/home/liujunyi/anaconda3/envs/vmamba2/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl return forward_call(*args, **kwargs) File "/mnt/c/Users/caica/Desktop/swin/classes/fc4/VMamba_main/classification/models/vmamba.py", line 745, in forwardv2 y = self.forward_core(x) File "/mnt/c/Users/caica/Desktop/swin/classes/fc4/VMamba_main/classification/models/vmamba.py", line 706, in forward_corev2 ys: torch.Tensor = selective_scan( File "/mnt/c/Users/caica/Desktop/swin/classes/fc4/VMamba_main/classification/models/vmamba.py", line 598, in selective_scan return SelectiveScan.apply(u, delta, A, B, C, D, delta_bias, delta_softplus, nrows, backnrows, ssoflex) File "/home/liujunyi/anaconda3/envs/vmamba2/lib/python3.10/site-packages/torch/autograd/function.py", line 553, in apply return super().apply(*args, **kwargs) # type: ignore[misc] File "/home/liujunyi/anaconda3/envs/vmamba2/lib/python3.10/site-packages/torch/cuda/amp/autocast_mode.py", line 115, in decorate_fwd return fwd(*args, **kwargs) File "/mnt/c/Users/caica/Desktop/swin/classes/fc4/VMamba_main/classification/models/csms6s.py", line 267, in forward out, x, *rest = selective_scan_cuda_core.fwd(u, delta, A, B, C, D, delta_bias, delta_softplus, 1) RuntimeError: memory format option is only supported by strided tensors
Hello!!!
I also encountered the same problem, can you solve it please?
QAQ ^^
Are all data (u, delta, A, B, C, D, delta_bias) contiguous in your application?
Hello!!! I also encountered the same problem, can you solve it please? QAQ ^^
I updated the version of pytorch and I solved this problem.becauseI I found that I did not install the selective_scan package properly
Hello!!! I also encountered the same problem, can you solve it please? QAQ ^^
I updated the version of pytorch and I solved this problem.becauseI I found that I did not install the selective_scan package properly
Hello, can you send me the details of how you did it?