TensorFlow in Go
Construct and execute TensorFlow graphs in Go.
WARNING: The API defined in this package is not stable and can change without notice. The same goes for the awkward package path (
github.com/tensorflow/tensorflow/tensorflow/go
).
Quickstart
-
Download and extract the TensorFlow C library, preferably into
/usr/local
. GPU-enabled versions require CUDA 8.0 and cuDNN 5.1. For other versions, the TensorFlow C library will have to be built from source (see below).- Linux: CPU-only, GPU-enabled
- OS X CPU-only, GPU-enabled
The following shell snippet downloads and extracts into
/usr/local
:TF_TYPE="cpu" # Set to "gpu" for GPU support curl -L \ "https://storage.googleapis.com/tensorflow/libtensorflow/libtensorflow-${TF_TYPE}-$(go env GOOS)-x86_64-1.0.0.tar.gz" | sudo tar -C /usr/local -xz
-
go get
this package (and run tests):go get github.com/tensorflow/tensorflow/tensorflow/go go test github.com/tensorflow/tensorflow/tensorflow/go
-
Done!
Installing into locations other than/usr/local
The TensorFlow C library (libtensorflow.so
) needs to be available at build time (e.g.,go build
) and run time (go test
or executing binaries). If the library has not been extracted into/usr/local
, then it needs to be made available through theLIBRARY_PATH
environment variable at build time and theLD_LIBRARY_PATH
environment variable (DYLD_LIBRARY_PATH
on OS X) at run time.
For example, if the TensorFlow C library was extracted into/dir
, then:
export LIBRARY_PATH=/dir/lib
export LD_LIBRARY_PATH=/dir/lib # For Linux
export DYLD_LIBRARY_PATH=/dir/lib # For OS X
Building the TensorFlow C library from source
If the "Quickstart" instructions above do not work (perhaps the release archives are not available for your operating system or architecture, or you're using a different version of CUDA/cuDNN), then the TensorFlow C library must be built from source.
Prerequisites
- bazel
- Environment to build TensorFlow from source code (Linux or OS X). If you don't need GPU support, then try the following:
sh # Linux sudoapt-get install python swig python-numpy # OS X with homebrew brew installswig
Build
-
Download the source code
go get -d github.com/tensorflow/tensorflow/tensorflow/go
-
Build the TensorFlow C library:
cd ${GOPATH}/src/github.com/tensorflow/tensorflow ./configure bazel build --config opt //tensorflow:libtensorflow.so
This can take a while (tens of minutes, more if also building for GPU).
-
Make
libtensorflow.so
available to the linker. This can be done by either:a. Copying it to a system location, e.g.,
sudo cp ${GOPATH}/src/github.com/tensorflow/tensorflow/bazel-bin/tensorflow/libtensorflow.so /usr/local/lib
OR
b. Setting environment variables:
export LIBRARY_PATH=${GOPATH}/src/github.com/tensorflow/tensorflow/bazel-bin/tensorflow # Linux export LD_LIBRARY_PATH=${GOPATH}/src/github.com/tensorflow/tensorflow/bazel-bin/tensorflow # OS X export DYLD_LIBRARY_PATH=${GOPATH}/src/github.com/tensorflow/tensorflow/bazel-bin/tensorflow
-
Build and test:
go test github.com/tensorflow/tensorflow/tensorflow/go
Generate wrapper functions for ops
Go functions corresponding to TensorFlow operations are generated inop/wrappers.go
. To regenerate them:
Prerequisites:
- Protocol buffer compiler (protoc) 3.x
- The TensorFlow repository under GOPATH
go generate github.com/tensorflow/tensorflow/tensorflow/go/op
Support
Use stackoverflow and/or Github issues.
Contributions
Contributions are welcome. If making any signification changes, probably best to discuss on a Github issue before investing too much time. Github pull requests are used for contributions.