Custom deep learning code for TensorFlow
Project description
DeepSaki
Welcome AI enthuisiasts to DeepSaki, a collection of reusable machine learning code. :muscle::robot::metal:
The ML framework used is tensorflow and the entire code is suitable to run Google's TPUs.
Installation
Git
git clone https://github.com/sascha-kirch/DeepSaki.git
Pip
pip install DeepSaki
Content
- activations
- ComplexActivation
- initializer
- HeAlphaNormal
- HeAlphaUniform
- helper
- MakeInitializerComplex
- layers
- GlobalSumPooling2D
- ReflectionPadding (suitable for TPU)
- FourierConvolution2D
- rFFTPooling2D
- FourierFilter2D
- FourierPooling2D
- FFT2D
- iFFT2D
- Conv2DBlock
- Conv2DSplitted
- DenseBlock
- DownSampleBlock
- UpSampleBlock
- ResidualIdentityBlock
- ResBlockDown
- ResBlockUp
- ScaleLayer
- ScalarGatedSelfAttention
- Encoder
- Bottleneck
- Decoder
- helper
- GetInitializer
- pad_func
- dropout_func
- PlotLayer
- loss
- PixelDistanceLoss
- StructuralSimilarityLoss
- models
- LayoutContentDiscriminator
- PatchDiscriminator
- ResNet
- UNet
- UNetDiscriminator
- optimizer
- SWATS_ADAM
- SWATS_NADAM
- regularization
- CutMix
- CutOut
- GetMask
- utils
- DetectHw
- EnableXlaAcceleration
- EnableMixedPrecision
Repo Stats
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
DeepSaki-0.1.2.tar.gz
(112.9 kB
view hashes)
