OpenCL library to train deep convolutional networks
- C++
- OpenCL
- Deep convolutional
- Python wrappers
- Lua wrappers
- Q-learning
APIs:
Layer types:
- convolutional
- max-pooling
- normalization
- activation
- dropout
- random translations
- random patches
- loss
Loss layer types:
- softmax
- cross-entropy (synonymous with multinomial logistic, etc)
- square loss
Trainers:
- SGD
- Anneal
- Nesterov
- Adagrad
- Rmsprop
- Adadelta
Activations:
- tanh
- scaled tanh (1.7519 * tanh(2/3x) )
- linear
- sigmoid
- relu
- elu (new!)
- jpegs
- mnist
- kgsv2
- norb
Weight initializers:
- original
- uniform
- more possible...
Multicolumn net also possible, as in McDnn
- obtained 37.2% test accuracy, on next move prediction task, using 33.6 million training examples from kgsgo v2 dataset
- commandline used
./deepclrun dataset=kgsgoall netdef=12*(32c5z-relu)-500n-tanh-361n numepochs=15 learningrate=0.0001
- 2 epochs, 2 days per epoch, on an Amazon GPU instance, comprising half an NVidia GRID K520 GPU (about half as powerful as a GTX780)
- commandline used
- obtained 99.5% test accuracy on MNIST, using
netdef=rt2-8c5z-relu-mp2-16c5z-relu-mp3-150n-tanh-10n numepochs=20 multinet=6 learningrate=0.002
- epoch time 99.8 seconds, using an Amazon GPU instance, ie half an NVidia GRID K520 GPU (since we are learning 6 nets in parallel, so 16.6seconds per epoch per net)
This section installs the native libraries, and the command-line tools. You always need to do this part, even if you will use the Python wrappers.
- OpenCL-enabled GPU or APU, along with appropriate OpenCL driver installed
- Tested using Windows 2012 RC2, and (New!) Visual Studio 2015, this is how the CI builds run
- Download latest binary zip file from https://deepcl.hughperkins.com/Downloads/ (eg from v8.0.0rc8)
- unzip it, which creates the
dist
folder - To test it:
- open a cmd
- run
call dist\bin\activate.bat
(adjusting the path appropriately for wherever you downloaded deepcl binaries to) - now, eg try
deepcl_unittests
- (New!), you can choose which gpu to run tests on now, eg:
deepcl_unittests gpuindex=1
Note that you need to "activate" the installation each time you open a new cmd prompt (or you could add appropriate environment variables permanently, using Control Panel | System | Advanced System Settings | Environment Variables)
- OpenCL-enabled GPU or APU, along with appropriate OpenCL driver installed (can check by running
clinfo
, which should show your desired GPU device) - Tested using Ubuntu 14.04 32-bit/64-bit
- Download latest tar file from https://deepcl.hughperkins.com/Downloads/ (eg from v8.0.0rc8)
- untar it, which creates the
dist
sub-folder - in a bash prompt, run
source dist/bin/activate.sh
(adjust the path appropriate for wherever you untarred the binaries tar file to) - test by doing, from the same bash prompt, eg
deepcl_unittests
- (New!), you can choose which gpu to run tests on now, eg:
deepcl_unittests gpuindex=1
- (New!), you can choose which gpu to run tests on now, eg:
Note that you need to "activate" the installation each time you open a new bash prompt (or you can call activate.sh from your .bashrc
file)
- make sure you already installed the native library, and "activate"d it, by doing
call dist\bin\activate.bat
, orsource dist/bin/activate.sh
- run
pip install --pre DeepCL
- test by doing
python -c "import PyDeepCL; cl = PyDeepCL.DeepCL()"
Building from source is only needed if installing from binaries doesn't work for your configuration, or if you want to modify DeepCL.
See Build.md
- Check if you have an OpenCL-enabled device on your system
- ideally a GPU, or accelerator, since there is no attempt to optimize DeepCL for CPUs (at least, not currently, could change, feel free to submit a pull request :-) )
- Try running
gpuinfo
(from EasyCL, but built as part of this project too, for ease of use )- it should output at least one OpenCL-enabled device
- if it doesn't, then you need to make sure you have an OpenCL-enabled device, and that appropriate drivers are installed, and that the ICD is configured appropriately (registry in Windows, and
/etc/OpenCL/vendors
in linux)
Please raise an issue, let me know you're interested.
- If it's on my list of things I was going to do sooner or later anyway (see below), I might do it sooner rather than later.
- If it's to do with usability, I will try to make that a priority
- please feel free to fork this repository, tweak things, send a pull request. Or get in contact. Or both :-)
- If you're using Clover, you might want to look at:
- this thread hughperkins#35
- this branch https://github.com/hughperkins/DeepCL/tree/clover-compatibility
- kgsgo-dataset-preprocessor Dataset based on kgsgo games; 33 million data points
- cltorch
- clnn
- 7th August 2016:
- "standard" version of windows compiler changed from msvc2010 to msvc2015 update 3 (no change to linux/mac)
- "standard" version of python 3.x on windows changed from 3.4 to 3.5 (no change to linux/mac)
- (note: python2.7 continues to work as before on all of Windows 32/64, linux, Mac)
- standard c++ version on linux/mac changed from c++0x to c++11
- 29th July 2016:
- python fixes:
- CHANGE: must use numpy tensors now,
array.array
no longer accepted - New feature: can provide numpy tensors as 4d tensors now, no longer have to be 1d tensors
- Bug fix: q-learning working again now (hopefully)
- CHANGE: must use numpy tensors now,
- python fixes:
- 26th July 2016:
- fixed some bugs in manifest loader
- no longer need to specify the number of images in the first line of the manifest file
- added
gpuindex=
option todeepcl_unittests
(quite beta for now...)
- 4th January 2016:
- fixed a number of build warnings on Mac, both in OpenCL build, and C++ build
- 3rd January 2016:
- create Mac OS X build on Travis, and fix the build, https://travis-ci.org/hughperkins/DeepCL
- 27th November:
- added ELU
- Week of 26th October:
- created branch
clblas-2.8.0
, which works with Visual Studio 2015. It uses the latest 2.8.x release of clBLAS. Thank you to jakakonda for helping to test this and get it working.
- created branch
- Aug 28th:
- merged 8.x branch to master, will release first version of 8.x shortly
- installation of 8.x from binaries on Windows works now, by doing, eg on 32-bit Windows 7, and assuming you already activated an appropriate python environment (assumes 7-zip is installed, in default location, otherwise do the unzip by hand):
powershell Set-ExecutionPolicy unrestricted
rem following command is like `wget` in linux:
powershell.exe -Command (new-object System.Net.WebClient).DownloadFile('https://deepcl.hughperkins.com/Downloads/deepcl-win32-v8.0.0rc8.zip', 'deepcl-win32-v8.0.0rc8.zip')
rem following command is like `tar -xf` in linux:
"c:\program files\7-Zip\7z.exe" x deepcl-win32-v8.0.0rc8.zip
call dist\bin\activate.bat
pip install --pre DeepCL
python -c "import PyDeepCL; cl = PyDeepCL.DeepCL()"
# (last line is just to check works ok)
- Aug 26th: installation of 8.x from binaries on linux works now, by doing, eg on 64-bit Ubuntu 14.04:
mkdir 8.0.0rc4
cd 8.0.0rc4
wget https://deepcl.hughperkins.com/Downloads/deepcl-linux64-v8.0.0rc4.tar.bz2
tar -xf deepcl-linux64-v8.0.0rc4.tar.bz2
virtualenv env
source env/bin/activate
source dist/bin/activate.sh
pip install --pre DeepCL
python -c "import PyDeepCL; cl = PyDeepCL.DeepCL()"
(last line is just to check works ok)
- Aug 21st-24th:
- 8.x finally builds again on all CI tested configurations!
- ubuntu 14.04 32-bit Python 2.7
- ubuntu 14.04 32-bit Python 3.4
- ubuntu 14.04 64-bit Python 2.7
- ubuntu 14.04 64-bit Python 3.4
- visual studio 2010 32-bit python 2.7
- visual studio 2010 32-bit python 3.4
- visual studio 2010 64-bit python 2.7
- visual studio 2010 64-bit python 3.4
- 8.x finally builds again on all CI tested configurations!
- Aug 19th-20th:
- Python wrappers now built using a very thin setup.py layer, on top of the standard native DeepCL build
- Aug 18th:
- added BackwardIm2Col layer, which uses im2col for backward propagation
- added BackpropWeightsIm2Col layer, which uses im2col for weight update
- added BackwardAuto layer, which automatically selects fastest Backward layer
- added BackpropWeightsAuto layer, which automatically selects faster weight update layer
- under the covers:
- created ClBlasHelper, to handle Gemm and Gemv
- factorized im2col into Im2Col class
- week up to Aug 17th:
- added forward and backward im2col layer
- forward im2col automatically used during forward propagation, where appropriate
- backwards has yet to be integrated
- under the covers:
- added clBLAS
- migrated the Python build process to use cmake, rather than setup.py (whether this turns out to be good or bad is a bit up in the air for now)
- June 22nd:
Just create an issues, in github, in the top right of this page. Don't worry about whether you think the issue sounds silly or anything. The more feedback the better!
Note that I'm currently focused 100.000% on cuda-on-cl, so please be patient during this period.