-
TFLite uses int8 per-channel weight quantization for transposed convolutions.
While XNNPACK includes a fast transposed convolution operation it only supports per-tensor weight quantization (i.e. a si…
-
run python test_timeit.py:
compare time with nn.Conv3d()
nn.Conv3D() forward time is :0.000066
SparseConv3d() forward time is :0.001239
nn.Conv3D() backward time is :0.000870
SparseConv3d() backw…
-
I'd like to add some semantic segmentation models.
Possible candidates:
- [ ] [Wider or Deeper: Revisiting the ResNet Model for Visual Recognition](https://arxiv.org/abs/1611.10080)
- code: h…
-
According to my understanding on the paper, in order for both images to be able to perform cross input neighborhood differences, the first two convolution layers must be tied in which their trained fi…
-
Hello,
I am new to deep neural network. I have started using matconvnet. I am using vgg-f pre-trained network for extracting features. This architecture has 5 convolutional layer and followed by 3 fu…
-
How can I modify the model to add a convolution block before the first conv layer?
-
- https://arxiv.org/abs/2107.13034
- 2021
機械学習アルゴリズムの有効性は、大量のデータから有用な特徴を抽出できるかどうかで決まる。
モデルやデータセットのサイズが大きくなるにつれて、大規模なデータセットを大幅に小さくしながらも性能の高いデータセットに圧縮するデータセット蒸留法は、学習効率や有用な特徴抽出の点で価値が高くなる。
この目的のために、…
e4exp updated
3 years ago
-
I can't find a suitable 3d version of DCN/DCNv2 on the Internet. Could you update this PyTorch version with a 3D DCN with support for large images?
For example: (N, 3, 128, 128, 128)
-
When I ran the code, I got the error :
2 root error(s) found.
(0) Unknown: Failed to get convolution algorithm. This is probably because cuDNN failed to initialize, so try looking to see if a war…
-
Thanks for your good work!
Group conv is used in a lot of ConvNets.
Will you plan to support group conv in the future version of dcn?