TensorFlow.jl

TensorFlow.jl: An Idiomatic Julia Front End for TensorFlow - Published in JOSS (2018)

https://github.com/malmaud/tensorflow.jl

Science Score: 95.0%

This score indicates how likely this project is to be science-related based on various indicators:

○
CITATION.cff file
✓
codemeta.json file
Found codemeta.json file
✓
.zenodo.json file
Found .zenodo.json file
✓
DOI references
Found 6 DOI reference(s) in README and JOSS metadata
✓
Academic publication links
Links to: joss.theoj.org
✓
Committers with academic emails
5 of 38 committers (13.2%) from academic institutions
○
Institutional organization owner
✓
JOSS paper metadata
Published in Journal of Open Source Software

Keywords

gpu julia machine-learning tensorflow wrapper

Keywords from Contributors

the-human-brain flux unconstrained-optimization unconstrained-optimisation optimisation optim streaming-data stochastic-approximation onlinestats online-algorithms

Scientific Fields

Earth and Environmental Sciences Physical Sciences - 40% confidence

Last synced: 4 months ago · JSON representation

Repository

A Julia wrapper for TensorFlow

Basic Info

Host: GitHub
Owner: malmaud
License: other
Language: Julia
Default Branch: master
Homepage:
Size: 3.55 MB

Statistics

Stars: 883
Watchers: 41
Forks: 110
Open Issues: 78
Releases: 27

Topics

gpu julia machine-learning tensorflow wrapper

Created about 10 years ago · Last pushed over 1 year ago

Metadata Files

Readme Changelog License Codeowners

TensorFlow

A wrapper around TensorFlow, a popular open source machine learning framework from Google.

| ⚠️ Notice ⚠️ | | --- | | Tensorflow.jl is in minimal maintenance mode | | While it works, it is not receiving new features, and is bound to an old version, 1.13.1, of libtensorflow.| | The authors recommend Flux.jl for new users. |

Documentation

Other resources:

Why use TensorFlow.jl?

See a list of advantages over the Python API.

What's changed recently?

See NEWS.

Basic usage

```julia

using TensorFlow using Test

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.globalvariablesinitializer()) res = run(sess, w, Dict(z=>Float64[1,2])) @test res[1] ≈ exp(-1) ```

Installation

Install via

julia Pkg.add("TensorFlow")

To enable support for GPU usage on Linux, set an environment variable TF_USE_GPU to "1" and then rebuild the package. eg

julia 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

Simply run docker run -it malmaud/julia:tf to open a Julia REPL that already has TensorFlow installed:

julia 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

```julia using TensorFlow using Distributions using Printf

Generate some synthetic data

x = randn(100, 50) w = randn(50, 10) yprob = exp.(x*w) yprob ./= sum(y_prob,dims=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, shape=[-1, 50]) Y_obs = placeholder(Float64, shape=[-1, 10])

variablescope("logisiticmodel"; initializer=Normal(0, .001)) do global W = getvariable("W", [50, 10], Float64) global B = getvariable("B", [10], Float64) end

Y=nn.softmax(X*W + B)

Loss = -reducesum(log(Y).*Yobs) optimizer = train.AdamOptimizer() minimize_op = train.minimize(optimizer, Loss) saver = train.Saver()

Run training

run(sess, globalvariablesinitializer()) checkpointpath = mktempdir() @info("Checkpoint files saved in $checkpointpath") for epoch in 1:100 curloss, _ = run(sess, [Loss, minimizeop], Dict(X=>x, Yobs=>y)) println(@sprintf("Current loss is %.2f.", curloss)) train.save(saver, sess, joinpath(checkpointpath, "logistic"), globalstep=epoch) end ```

Troubleshooting

If you see issues from the ccall or python interop, try updating TensorFlow both in Julia and in the global python install:

julia julia> Pkg.build("TensorFlow")

bash $ pip install --upgrade tensorflow

Citing

If you use this software in your research, we would really appreciate if you cite us.

Malmaud, J. & White, L. (2018). TensorFlow.jl: An Idiomatic Julia Front End for TensorFlow. Journal of Open Source Software, 3(31), 1002, https://doi.org/10.21105/joss.01002

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 these instructions. 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.

Owner

Name: Jonathan Malmaud
Login: malmaud
Kind: user

Website: malmaud.com
Repositories: 26
Profile: https://github.com/malmaud

Google

JOSS Publication

TensorFlow.jl: An Idiomatic Julia Front End for TensorFlow

Published

November 01, 2018

DOI

10.21105/joss.01002

Volume 3, Issue 31, Page 1002

Authors

Jonathan Malmaud

Massachusetts Institute of Technology

Lyndon White

The University of Western Australia

Editor

Ariel Rokem

GitHub Events

Total

Issues event: 1
Watch event: 6
Fork event: 1

Last Year

Issues event: 1
Watch event: 6
Fork event: 1

Committers

Last synced: 5 months ago

All Time

Total Commits: 1,086
Total Committers: 38
Avg Commits per committer: 28.579
Development Distribution Score (DDS): 0.44

Past Year

Commits: 0
Committers: 0
Avg Commits per committer: 0.0
Development Distribution Score (DDS): 0.0

Top Committers

Name	Email	Commits
Jon Malmaud	m**d@g**m	608
Lyndon White	o**x@u**u	190
Lyndon White	l**e@r**u	80
Katie Hyatt	k**t@p**u	40
adam	a**i@i**l	30
Johan Gustafsson	j**n@2**i	21
Mike J Innes	m**s@g**m	18
yeesian	n**n@g**m	14
alex.terenin	a**n@p**m	8
pevnak	p**k@g**m	7
Fredrik Bagge Carlson	c**b@u**g	7
Tony Kelman	t**y@k**t	6
Elliot Saba	s**t@g**m	6
Olivier Thill	o**l@e**k	5
Asim Shankar	a**r@g**m	4
kevin-w-li	k**i@o**m	4
Keno Fischer	k**o@j**m	3
Lars Mescheder	L**r@g**t	3
byronbest	4****t	3
Stephen Smith	3****9	3
karenl7	k**7@s**u	2
femtocleaner[bot]	f****]	2
WooKyoung Noh	w**h@g**m	2
Andreas Noack	a**n@g**m	2
Colin Phillips	c**p@d**a	2
Guido Kraemer	g****r	2
Henry Poitras	k**u@g**m	2
Páll Haraldsson	P**n@g**m	2
Alexander Barth	b**r@g**m	1
Alexander Terenin	a****n	1
and 8 more...

Committer Domains (Top 20 + Academic)

uab.cat: 1 mit.edu: 1 kingston.net: 1 discovery.co.za: 1 stanford.edu: 1 gmx.net: 1 juliacomputing.com: 1 google.com: 1 essex.ac.uk: 1 kelman.net: 1 ulund.org: 1 petuum.com: 1 247.ai: 1 imgw.pl: 1 physics.ucsb.edu: 1 research.uwa.edu.au: 1 ucc.asn.au: 1

Issues and Pull Requests

Last synced: 4 months ago

All Time

Total issues: 58
Total pull requests: 44
Average time to close issues: 6 months
Average time to close pull requests: about 2 months
Total issue authors: 43
Total pull request authors: 15
Average comments per issue: 2.93
Average comments per pull request: 3.34
Merged pull requests: 28
Bot issues: 0
Bot pull requests: 0

Past Year

Issues: 0
Pull requests: 0
Average time to close issues: N/A
Average time to close pull requests: N/A
Issue authors: 0
Pull request authors: 0
Average comments per issue: 0
Average comments per pull request: 0
Merged pull requests: 0
Bot issues: 0
Bot pull requests: 0

View more stats

Top Authors

Issue Authors

oxinabox (4)
PallHaraldsson (4)
montyvesselinov (3)
askrix (3)
daxinnan (2)
stevengj (2)
abianco88 (2)
Keno (2)
staticfloat (2)
industriousrebel (1)
MoonCoral (1)
chelseas (1)
johnnychen94 (1)
mbeach42 (1)
holmesconan (1)

Pull Request Authors

oxinabox (15)
malmaud (7)
PallHaraldsson (5)
kailaix (4)
byronbest (3)
Keno (2)
aterenin (1)
MoonCoral (1)
gustafsson (1)
askrix (1)
Alexander-Barth (1)
metanoid (1)
blair0011 (1)
johnnychen94 (1)
staticfloat (1)

Top Labels

Issue Labels

installation/libtensorflow (2) bug (1) documentation (1)

Pull Request Labels

awaiting review (1)

Packages

Total packages: 1
Total downloads: unknown

Total dependent packages: 0
Total dependent repositories: 1
Total versions: 3

juliahub.com: TensorFlow

A Julia wrapper for TensorFlow

Documentation: https://docs.juliahub.com/General/TensorFlow/stable/
License: MIT
Latest release: 0.11.0
published almost 7 years ago

Versions: 3
Dependent Packages: 0
Dependent Repositories: 1

Rankings

Stargazers count: 0.4%

Forks count: 0.8%

Dependent repos count: 7.7%

Average: 12.3%

Dependent packages count: 40.3%

Last synced: 4 months ago

Dependencies

REQUIRE julia

AutoHashEquals 0.1.0
Compat 0.18
Conda 0.6.0
Distributions 0.10.2
FileIO 0.1.2
JLD2 0.0.6
Juno 0.2.3
MLDatasets 0.3.0
MacroTools 0.3.6
ProtoBuf 0.6.1
PyCall 1.7.1
SpecialFunctions v0.7.0
SpecialFunctions 0.7.0
StatsFuns 0.3.0
julia 0.7

Docker/cpu/Dockerfile docker

tensorflow/tensorflow latest build

Docker/gpu/Dockerfile docker

tensorflow/tensorflow latest-gpu build

deps/build_libtensorflow/Dockerfile docker

docker latest build

deps/build_libtensorflow/cpu/Dockerfile docker

tensorflow/tensorflow 1.1.0-rc0-devel build

deps/build_libtensorflow/gpu/Dockerfile docker

tensorflow/tensorflow 1.1.0-rc0-devel-gpu build

test/REQUIRE julia

TensorFlow.jl

Science Score: 95.0%

Keywords

Keywords from Contributors

Scientific Fields

Repository

Basic Info

Statistics

Topics

Metadata Files

README.md

TensorFlow

Documentation

Why use TensorFlow.jl?

What's changed recently?

Basic usage

Installation

Installation via Docker

Logistic regression example

Generate some synthetic data

Build the model

Run training

Troubleshooting

Citing

Optional: Using a custom TensorFlow binary

Owner

JOSS Publication

TensorFlow.jl: An Idiomatic Julia Front End for TensorFlow

Authors

Editor

Tags

GitHub Events

Total

Last Year

Committers

All Time

Past Year

Top Committers

Committer Domains (Top 20 + Academic)

Issues and Pull Requests

All Time

Past Year

Top Authors

Issue Authors

Pull Request Authors

Top Labels

Issue Labels

Pull Request Labels

Packages

juliahub.com: TensorFlow

Rankings

Dependencies