A wrapper around TensorFlow, a popular open source machine learning framework from Google.
Why use TensorFlow.jl?
See a list of advantages over the Python API.
What's changed recently?
using TensorFlow sess = TensorFlow.Session() x = TensorFlow.constant(Float64[1,2]) y = TensorFlow.Variable(Float64[3,4]) z = TensorFlow.placeholder(Float64) w = exp(x + z + -y) run(sess, TensorFlow.global_variables_initializer()) res = run(sess, w, Dict(z=>Float64[1,2])) Base.Test.@test res ≈ exp(-1)
To enable support for GPU usage on Linux, set an environment variable
TF_USE_GPU to "1" and then rebuild the package. eg
ENV["TF_USE_GPU"] = "1" Pkg.build("TensorFlow")
CUDA 8.0 and cudnn are required for GPU usage. If you need to use a different version of CUDA, or if you want GPU support on Mac OS X, you can compile libtensorflow from source.
Initial precompilation (eg, the first time you type
using TensorFlow) can take around five minutes, so please be patient. Subsequent load times will only be a few seconds.
Installation via Docker
docker run -it malmaud/julia:tf to open a Julia REPL that already
has TensorFlow installed:
julia> using TensorFlow julia>
For a version of TensorFlow.jl that utilizes GPUs, use
nvidia-docker run -it malmaud/julia:tf_gpu.
Download nvidia-docker if you don't
already have it.
Logistic regression example
Realistic demonstration of using variable scopes and advanced optimizers
using Distributions # Generate some synthetic data x = randn(100, 50) w = randn(50, 10) y_prob = exp(x*w) y_prob ./= sum(y_prob,2) function draw(probs) y = zeros(size(probs)) for i in 1:size(probs, 1) idx = rand(Categorical(probs[i, :])) y[i, idx] = 1 end return y end y = draw(y_prob) # Build the model sess = Session(Graph()) X = placeholder(Float64) Y_obs = placeholder(Float64) variable_scope("logistic_model", initializer=Normal(0, .001)) do global W = get_variable("weights", [50, 10], Float64) global B = get_variable("bias", , Float64) end Y=nn.softmax(X*W + B) Loss = -reduce_sum(log(Y).*Y_obs) optimizer = train.AdamOptimizer() minimize_op = train.minimize(optimizer, Loss) saver = train.Saver() # Run training run(sess, global_variables_initializer()) checkpoint_path = mktempdir() info("Checkpoint files saved in $checkpoint_path") for epoch in 1:100 cur_loss, _ = run(sess, [Loss, minimize_op], Dict(X=>x, Y_obs=>y)) println(@sprintf("Current loss is %.2f.", cur_loss)) train.save(saver, sess, joinpath(checkpoint_path, "logistic"), global_step=epoch) end
If you see issues from the ccall or python interop, try updating TensorFlow both in Julia and in the global python install:
$ pip install --upgrade tensorflow
Optional: Using a custom TensorFlow binary
To build TensorFlow from source, or if you already have a TensorFlow binary that you wish to use, follow the instructions at https://malmaud.github.io/TensorFlow.jl/latest/build_from_source/. This is recommended by Google for maximum performance, and is currently needed for Mac OS X GPU support.
For Linux users, a convenience script is included to use Docker to easily build the library. Just install docker and run
julia build_libtensorflow.so from the "deps" directory of the TensorFlow.jl package. Note that this method may not link to all libraries available on the target system such as Intel MKL.