-
code as follow:
```
func.func @aten_cat_test(%arg0: !torch.vtensor, %arg1: !torch.vtensor, %arg2: !torch.vtensor) -> !torch.vtensor {
%false = torch.constant.bool false
%int1 = torch.constant.…
-
While in cuDNN's [`cudnnConvolutionForward()`](https://docs.nvidia.com/deeplearning/cudnn/latest/api/cudnn-cnn-library.html#cudnnconvolutionforward) there is the `groupCount` option for grouped convol…
-
https://github.com/ml-explore/mlx/blob/03cf033f8299a5a694b31b650c6e03aa62c2a5b6/mlx/primitives.cpp#L903
-
Would you consider adding grouped capability to your convs considering that structures like ResNets can hugely benefit from it GPU computation-wise by moving some of their height into grouped width an…
ibmua updated
7 years ago
-
**Describe the bug**
I am trying to use onnx converted MobileNet-v3 on raspberry pi. However, the converted model took too much time to inference.
To find out which part delaying the whole process,…
-
I think the if-else statement is redundant? Since if the groups parameter is greater than out_channels, an error(out_channels must divisible by groups) will be raised when we define the original convo…
-
Grouped convolution is supported well in `pytorch`'s convolution layers / ops. If possible, it would a great to add that ability to `torchsparse`.
-
### 🐛 Describe the bug
Since torch 2.2.0, the performance regression of `conv_transpose1d` has occurred on CPU (#120982).
torch 2.4.0 (w/ OneDNN 3.4.2) perfectly **fixed** the regression **only when…
-
Should we implement grouped convolutions in Alexnet baseline and attention model, or only use straightforward conv layers? If using groups, how many groups should we use?
Also, should we consider usi…
-
Hello there
You have proposed two models, CARN and CARN-M.
Figures 2 and 3 of your paper, and paragraph 3 of the paper, show that the residual block in your CARN uses residual-E. But in your code, o…