In the step 'Build the pip package', since we are building the binary file and not the pip package, instead run bazel build -config=opt //tensorflow:libtensorflow.so, adding -config=cuda if GPU support is desired. Be sure to follow all other steps as needed for your OS. In the step 'Prepare environment', ignore 'Install python dependencies' – these are not necessary as we are not building for Python. To build libtensorflow for TensorFlow.jl, follow the official instructions for building tensorFLow from source, except for a few minor modifications so as to build the library rather than the client. It will also be required for Mac OS X GPU support for TensorFlow versions later than 1.1. Build TensorFlow from source Building TensorFlow from sourceīuilding TensorFlow from source is recommended by Google for maximum performance, especially when running in CPU mode - on some systems, the difference can be substantial.
