This package contains experimental Vala bindings for TensorFlow.
Highly experimental Vala bindings for TensorFlow, using the C API.
As a result, current support for TensorFlow is limited by the exposed functionality in the API, which can be found in the official site.
| Feature | Python | C API | Vala |
|---|---|---|---|
| Run a predefined Graph | Yes | Yes | Yes |
| Graph construction | Yes | Yes | Yes |
| Gradients | Yes | No | No |
| Functions | Yes | No | No |
| Control Flow | Yes | No | No |
| Neural Network Library | Yes | No | No |
This step is the same as the one for the Rust bindings:
- Install SWIG and NumPy. The version from your distro's package manager should be fine for these two.
- Install Bazel, which you may need to do from source.
git clone --recurse-submodules https://github.com/tensorflow/tensorflowcd tensorflow./configurebazel build -c opt --jobs=1 tensorflow:libtensorflow_c.so
- Clone this repository and
cdinto it and:
meson build --prefix <install prefix (/usr/local by default)>
ninja -C build
ninja -C build test
sudo ninja -C build installyaourt -S tensorflow-valaWhich will drag tensorflow as a dependency.
It is worth noticing that the bindings have not been fully tested. However, it's better to try to keep up with the C API as it is being built than to do it all at once.