f-dangel / unfoldNd

(N=1,2,3)-dimensional unfold (im2col) and fold (col2im) in PyTorch
MIT License
82 stars 6 forks source link

[onnx]Unsupported: ONNX export of convolution for kernel of unknown shape #36

Open Liupei1101 opened 5 months ago

Liupei1101 commented 5 months ago

pth to onnx failed when using unfoldNd() to instead unfold()

f-dangel commented 5 months ago

Hi, could you provide details on what you are trying to achieve, as well as an example that reproduces the behavior you are getting? Also, could this be related to #35?

Liupei1101 commented 5 months ago

Hi, could you provide details on what you are trying to achieve, as well as an example that reproduces the behavior you are getting? Also, could this be related to #35?

raise errors.SymbolicValueError( torch.onnx.errors.SymbolicValueError: Unsupported: ONNX export of convolution for kernel of unknown shape. [Caused by the value 'input.423 defined in (%input.423 : Float(, , , , strides=[1866240, 14580, 162, 1], requires_grad=1, device=cpu) = onnx::Reshape[allowzero=0](%3549, %3550),

Liupei1101 commented 5 months ago

in unfoldNd function failed ,the error is Unsupported: ONNX export of convolution for kernel of unknown shape

Liupei1101 commented 5 months ago

image