frugally-deep
Header-only library for using Keras (TensorFlow) models in C++.
MIT License
Bot releases are hidden (Show)
- Updates TensorFlow from
2.15to2.16, which resulted in some non-minor changes. - Dropped support for
Bidirectional,GRU,LSTM, and stateful models. They might be re-added again in the future though.
Published by Dobiasd 8 months ago
- Simplified the internal implementation of
sum_tensors - Removed some redundant internal stuff
- Auto-formatted all code
Published by Dobiasd 10 months ago
- Added support for
MultiHeadAttentionlayers
Published by Dobiasd 10 months ago
- Removed the
-p0suffix from our version numbers
Published by Dobiasd 10 months ago
- Improved the install process using CMake on Fedora Linux
Published by Dobiasd 11 months ago
- Added support for
UnitNormalizationlayers.
Published by Dobiasd 11 months ago
- Added support for
LayerNormalizationlayer. - Some improvements on
BatchNormalization.
Published by Dobiasd 12 months ago
- Improved performance of softmax calculation (https://github.com/Dobiasd/frugally-deep/issues/405)
Published by Dobiasd about 1 year ago
- Updated dependency FunctionalPlus to version
v0.2.20-p0
Published by Dobiasd about 1 year ago
Published by Dobiasd about 1 year ago
- Added support for
AdditiveAttentionlayer - Added support for
use_scaleandscore_modeconcatinAttentionlayer - Fixed softmax implementation
- Fixed compiler errors with Visual Studio 2022 (C++20)
- Avoid unnecessarily compiling the TensorFlow model when loading it for conversion
Published by Dobiasd about 1 year ago
- Added support for CategoryEncoding layer with output_mode one_hot
- Added support for Attention layer
- Fixed zero variance in a Normalization layer
- Added support for keepdims in global pooling layers
Published by Dobiasd over 1 year ago
- Improved the performance of the
Denselayer - Fixed variance-size assertion in the
Normalizationlayer - Added support for new layer types:
ActivityRegularizationCategoryEncodingCenterCropCropping3DDotMinimumResizingZeroPadding3D
Published by Dobiasd about 2 years ago
- Fixes interpretation of axis value in
Normalizationlayer (issue 357)
Published by Dobiasd over 2 years ago
- Improved performance (with
strides==(1, 1)) ofConv2D,DepthwiseConv2D, andSeparableConv2D. - JSON export: Reduced memory usage and output.file size.
Published by Dobiasd over 2 years ago
- Added support for the
Rescalinglayer.
Published by Dobiasd over 2 years ago
- Fixed missing
inlinekeyword.
Published by Dobiasd over 2 years ago
- Improved performance of SeparableConv2D and DepthwiseConv2D layers
Published by Dobiasd over 2 years ago
- improved checks and docs
- added support for
FixedDropout(noop) - added support for tensor expansion in
Multiplylayer
Published by Dobiasd almost 3 years ago
- Added support for
Normalizationlayer.
