Build WasmEdge With WasmEdge-Tensorflow Plug-in

The WasmEdge-TensorFlow plug-in is a software component that extends the functionality of the WasmEdge runtime. It allows developers to perform TensorFlow model inference with similar APIs to Python. The plug-in is designed for Rust to WebAssembly applications and depends on the TensorFlow C library for its operations.

Prerequisites

The prerequisites of the WasmEdge-Tensorflow plug-in is the same as the WasmEdge building environment on the Linux platforms or MacOS platforms.

Build WasmEdge with WasmEdge-Tensorflow Plug-in

To enable the WasmEdge WasmEdge-Tensorflow, developers need to building the WasmEdge from source with the cmake option -DWASMEDGE_PLUGIN_TENSORFLOW=On.

cd <path/to/your/wasmedge/source/folder>
cmake -GNinja -Bbuild -DCMAKE_BUILD_TYPE=Release -DWASMEDGE_PLUGIN_TENSORFLOW=On
cmake --build build
# For the WasmEdge-Tensorflow plug-in, you should install this project.
cmake --install build

备注

If the built wasmedge CLI tool cannot find the WasmEdge-Tensorflow plug-in, you can set the WASMEDGE_PLUGIN_PATH environment variable to the plug-in installation path (such as /usr/local/lib/wasmedge/, or the built plug-in path build/plugins/wasmedge_tensorflow/) to try to fix this issue.

Then you will have an executable wasmedge runtime under /usr/local/bin and the WasmEdge-Tensorflow plug-in under /usr/local/lib/wasmedge/libwasmedgePluginWasmEdgeTensorflow.so after installation.

Install the TensorFlow Dependency

Installing the necessary libtensorflow_cc.so and libtensorflow_framework.so on both Linux and MacOS platforms, we recommend the following commands:

curl -s -L -O --remote-name-all https://github.com/second-state/WasmEdge-tensorflow-deps/releases/download/TF-2.12.0-CC/WasmEdge-tensorflow-deps-TF-TF-2.12.0-CC-manylinux2014_x86_64.tar.gz
# For the Linux aarch64 platforms, please use the `WasmEdge-tensorflow-deps-TF-TF-2.12.0-CC-manylinux2014_aarch64.tar.gz`.
# For the MacOS x86_64 platforms, please use the `WasmEdge-tensorflow-deps-TF-TF-2.12.0-CC-darwin_x86_64.tar.gz`.
# For the MacOS arm64 platforms, please use the `WasmEdge-tensorflow-deps-TF-TF-2.12.0-CC-darwin_arm64.tar.gz`.
tar -zxf WasmEdge-tensorflow-deps-TF-TF-2.12.0-CC-manylinux2014_x86_64.tar.gz
rm -f WasmEdge-tensorflow-deps-TF-TF-2.12.0-CC-manylinux2014_x86_64.tar.gz

The shared library will be extracted in the current directory ./libtensorflow_cc.so.2.12.0 and ./libtensorflow_framework.so.2.12.0 on Linux platforms, or ./libtensorflow_cc.2.12.0.dylib and ./libtensorflow_framework.2.12.0.dylib on MacOS platforms.

备注

After building the plug-in, you can also find these shared libraries under the build/_deps/wasmedge_tensorflow_lib_tf-src/ directory.

Then you can move the library to the installation path and create the symbolic link:

mv libtensorflow_cc.so.2.12.0 /usr/local/lib
mv libtensorflow_framework.so.2.12.0 /usr/local/lib
ln -s libtensorflow_cc.so.2.12.0 /usr/local/lib/libtensorflow_cc.so.2
ln -s libtensorflow_cc.so.2 /usr/local/lib/libtensorflow_cc.so
ln -s libtensorflow_framework.so.2.12.0 /usr/local/lib/libtensorflow_framework.so.2
ln -s libtensorflow_framework.so.2 /usr/local/lib/libtensorflow_framework.so

If on MacOS platforms:

mv libtensorflow_cc.2.12.0.dylib /usr/local/lib
mv libtensorflow_framework.2.12.0.dylib /usr/local/lib
ln -s libtensorflow_cc.2.12.0.dylib /usr/local/lib/libtensorflow_cc.2.dylib
ln -s libtensorflow_cc.2.dylib /usr/local/lib/libtensorflow_cc.dylib
ln -s libtensorflow_framework.2.12.0.dylib /usr/local/lib/libtensorflow_framework.2.dylib
ln -s libtensorflow_framework.2.dylib /usr/local/lib/libtensorflow_framework.dylib

Or create the symbolic link in the current directory and set the environment variable export LD_LIBRARY_PATH=$(pwd):${LD_LIBRARY_PATH}.

For more information, you can refer to the GitHub repository.