A detailed description of the changes that are made in each update. Please note that during the alpha stage, backwards and forwards compatibility are not guaranteed!
NeuralFit 0.2.2 (11-02-2023)
- Fixed epoch status printing in Jupyter and Colab notebooks
- Fixed step time value
NeuralFit 0.2.1 (11-02-2023)
numpyversion to oldest-supported-numpy
- Made tensorflow optional, it can now be installed via
pip install neuralfit[keras]if necessary
NeuralFit 0.2.0 (05-02-2023)
- Fixed odd population sizes leading to crashes.
- Fixed bug within evolution algorithm that would lead to degraded performance.
- Better and more verbose checking of array shapes.
- Added recurrent neural networks!
NeuralFit 0.1.6 (14-01-2023)
- Enabled server certificate authentication to improve security.
- Switched to static linking of third party libraries to avoid shipping certificate .so’s and .dll’s.
- Removed support for
NeuralFit 0.1.5 (10-01-2023)
- Switched to manylinux2014 wheels to support more Linux versions.
NeuralFit 0.1.4 (07-01-2023)
- Updated issues tracker page on PyPI.
NeuralFit 0.1.3 (07-01-2023)
- Fixed double inclusion of .so files in manylinux builds.
NeuralFit 0.1.2 (06-01-2023)
- Fixed DLL import error on more Windows machines.
NeuralFit 0.1.1 (06-01-2023)
- Fixed DLL import error on Windows machines.
NeuralFit 0.1.0 (06-01-2023)
The initial alpha release! Visit the documentation for a detailed overview of all the features.