- Note: This library is deprecated for a faster and more scalable alternative (i.e. MZX),
- there will be no further changes in this library.
The
MZX
library will be released on the following link MZX
Collection of neural-net modules including both general blocks, and those introduced in latest research
The library is still under development, if you have some cool NN blocks in mind that deserves to be implemented in MZT please don't hesitate to open an issue with feature request, I'll try my best to implement those ASAP.
pip install mzt
- Convolution Blocks:
- ConvNormActivation1d
- ConvNormActivation2d
- ConvInvertedBlock1d
- ConvInvertedBlock2d
- Residual Blocks:
- ConvResidualBlock1d
- ConvResidualBlock2d
- ConvBottleNeckResidualBlock1d
- ConvBottleNeckResidualBlock2d
- ConvInvertedResidualBlock1d
- ConvInvertedResidualBlock2d
- Attention Blocks:
- SelfAttention1d
- SelfAttention2d
- MultiHeadSelfAttention1d
- MultiHeadSelfAttention2d
- MultiHeadAttention1d
- MultiHeadAttention2d
- MultiHeadAttentionLinear
- Graphical/Geometric Blocks:
- GraphConv2d