Apache Logo
The Apache Way Contribute ASF Sponsors

This was extracted (@ 2017-10-16 20:10) from a list of minutes which have been approved by the Board.
Please Note The Board typically approves the minutes of the previous meeting at the beginning of every Board meeting; therefore, the list below does not normally contain details from the minutes of the most recent Board meeting.

2017 | 2016 | 2015 | 2014 | 2013 | 2012 | 2011 | 2010 | 2009 | 2008 | 2007 | 2006 | 2005 | 2004 | 2003 | 2002 | 2001 | 2000 | 1999 | Pre-organization meetings

SINGA

20 Sep 2017

Report was filed, but display is awaiting the approval of the Board minutes.

21 Dec 2016

Singa is a distributed deep learning platform.

Singa has been incubating since 2015-03-17.

Three most important issues to address in the move towards graduation:

 1. Add more documentations and tutorials, to make the installation and
    use of SINGA easier. We have done some work as described below in
    the answer of 'how has the project developed since the last
    report'.
 2. Grow the community. We added one more committer after the last
    report and are getting more user feedbacks. We want to further grow
    the community and to graduate in early 2017.

Any issues that the Incubator PMC (IPMC) or ASF Board wish/need to be
aware of?

 No

How has the community developed since the last report?

 There were 68, 125 and 61 emails from dev@ list in September 2016,
 October 2016 and November 2016 respectively.

 There are 100 more commits since the last report.

 One new committer was added into SINGA committer list.

How has the project developed since the last report?

 We released the v1.0 shortly after the last report, which include major
 changes in terms of the programming model and other aspects:

   + Tensor abstraction for supporting more machine learning models.
   + Device abstraction for running on different hardware devices,
     including CPU, (Nvidia/AMD) GPU and FPGA (to be tested in later
     versions).
   + Replace GNU autotool with cmake for compilation.
   + Support Mac OS.
   + Improve Python binding, including installation and programming
   + More deep learning models, including VGG and ResNet
   + Replace website building tool from maven to Sphinx and updated the
     documentations.
   + Integrate Travis-CI

 We also added python notebooks and windows support, and are working on
 the model zoo, python whl files and Docker images to make it easier
 for users to get started with SINGA.

 We are focusing on our niches, including distributed training at the
 system level and health-care models at the application level.

Date of last release:

 2016-09-09

When were the last committers or PMC members elected?

 2016-10-21

Signed-off-by:

 [ ](singa) Daniel Dai
 [X](singa) Alan Gates
 [X](singa) Ted Dunning
 [ ](singa) Thejas Nair

21 Sep 2016

Singa is a distributed deep learning platform.

Singa has been incubating since 2015-03-17.

Three most important issues to address in the move towards graduation:

 1. Grow the community. One committer was added in July, more would be
    nominated soon after releasing V1.0.  We are also expanding the user
    base by participating in meetups of local organizations, including
    PyDataSG.
 2. Optimize V1 release in terms of memory and efficiency for both
    stand-alone training and distributed training.
 3. Add more examples and tutorials to attract users, including deployment
    on Android phones.

Any issues that the Incubator PMC (IPMC) or ASF Board wish/need to be
aware of?

 No

How has the community developed since the last report?

 There were 187, 100 and 69 emails from dev@ list in June 2016, July 2016
 and August 2016 respectively.  The number of commits has increased 50%
 (about 200 commits) since last report.  One new committer was added.

How has the project developed since the last report?

 We have been working on V1.0 since the last release. The vote for releasing
 V1.0 has been sent to the general mailing list.

 The new version has major improvements in term of programming abstraction
 and usability. Some features are listed here,

 * Support heterogeneous hardware: CPP for CPU, CUDA for Nvidia GPU, OpenCL
   for other devices (including AMD GPUs)
 * New programming abstractions supporting complex deep learning models
 * Re-write the Python binding and provide running examples
 * Cross-platform (Linux and Mac OSX)
 * Upgrade the Cudnn library version to V5 for running RNN models
 * New website for documentation

Date of last release:

 2016-04-20

When were the last committers or PMC members elected?

 2016-07-20

Signed-off-by:

 [ ](singa) Daniel Dai
 [X](singa) Alan Gates
 [ ](singa) Ted Dunning
 [ ](singa) Thejas Nair

15 Jun 2016

Singa is a distributed deep learning platform.

Singa has been incubating since 2015-03-17.

Three most important issues to address in the move towards graduation:

 1. Grow the community. We anticipate stronger growth of the community when
    the upcoming, major release (v1.0) is available. SINGA v1.0 will have
    new programming abstractions and be able to run on heterogeneous
    hardware (e.g., using Cuda and OpenCL). It will also have improved
    usability in terms of installation and training new models, etc (see
    details below).
 2. Clean and improve existing documentations. We plan to replace the
    website template with a better one.
 3. Focus on Rafiki --- SINGA-based deep learning as a service.

Any issues that the Incubator PMC (IPMC) or ASF Board wish/need to be
aware of?

 No

How has the community developed since the last report?

 The Github repository has scored 602 stars and 205 forks.

 There were 107, 73 and 37 emails from dev@ list in March 2016, April 2016
 and May 2016 respectively.

 There were 9, 35, and 30 commits in March 2016, April 2016 and May 2016.

How has the project developed since the last report?

 In April 2016, we released our third version, i.e. v0.3.0, with the
 following major features:

 * Training on GPU cluster
 * Python wrapper improvements
 * New SGD updaters
 * Remove several dependencies for single node training.
 * Heterogeneous training with CPU and GPU.
 * Support cuDNNv4.

 In May, we started working on the first major release (v1.0) with the
 following notable features(http://singa.apache.org/develop/schedule.html):

 * Cross-platform
 * Support heterogeneous hardware: CUDA, OpenCL
 * Enhance programming abstraction, focusing on Tensor as core abstraction.
 * Python binding, cloud integration (Spark, Hadoop, etc.)
 * Runtime optimization.

 We are working on a system called Rafiki which uses SINGA to provide deep
 learning as a service and has the following features:

 * Sharing of pre-training models
 * Visualizing training progress
 * Configuring model via GUI (web browsers)

Date of last release:

 2016-04-20

When were the last committers or PMC members elected?

 N/A


Signed-off-by:

 [ ](singa) Daniel Dai
 [X](singa) Alan Gates
 [X](singa) Ted Dunning
 [ ](singa) Thejas Nair

Shepherd/Mentor notes:

 Alan Gates:
   I sent an email to the community sharing my concerns about the slow
   progress of adding new committers and asking if they had any contributors
   that looked like they might be ready for committership.

 Ted Dunning:
   The trend in dev mailing list traffic is worrisome as well. The opposite
   trend in JIRA volume is curious.

16 Mar 2016

Singa is a distributed deep learning platform.

Singa has been incubating since 2015-03-17.

Three most important issues to address in the move towards graduation:

 1. Improve the system, including performance and features like distributed
    GPU training over GPU clusters, R binding, etc.
 2. Add more examples and documentation to attach users
 3. Grow the community. We have been trying to reach out to a more diverse
    community of developers and users by presenting SINGA for various
    audiences. Completing items 1 and 2 above would attract more
    contributors and users.

Any issues that the Incubator PMC (IPMC) or ASF Board wish/need to be
aware of?

 No

How has the community developed since the last report?

 The Github repository has scored 540 stars and 190 forks.

 There were 60, 180 and 7 commits in Dec 2015, Jan 2016 and Feb 2016
 respectively.

 There were 77, 105, 13 emails from dev@ list in Dec 2015, Jan 2016 and Feb
 2016 respectively.

How has the project developed since the last report?

 In Jan 2016, we released our second version, i.e., incubating-v0.2.0, with
 the following major features:

 * GPU support, which enables training of complex models on a single node
   with multiple GPU cards.
 * Hybrid neural net partitioning, which supports data and model
   parallelism at the same time.
 * Python wrapper, which makes it easy to configure the job, including
   neural net and SGD algorithm.
 * RNN model and BPTT algorithms, to support applications based on RNN
   models, e.g., GRU.
 * Cloud software integration includes Mesos, Docker and HDFS.

Date of last release:

 2016-01-14

When were the last committers or PMC members elected?

 N/A

Signed-off-by:

 [ ](singa) Daniel Dai
 [X](singa) Alan Gates
 [X](singa) Ted Dunning
 [ ](singa) Thejas Nair

16 Dec 2015

Singa is a distributed deep learning platform.

Singa has been incubating since 2015-03-17.

Three most important issues to address in the move towards graduation:

 1. Expand the community, both in its size and its diversity.
 2. Improve training efficiency by supporting GPUs, and improve scalability
    by reducing communication cost.
 2. Improve the usability by providing more built-in models and python
    binding.

Any issues that the Incubator PMC (IPMC) or ASF Board wish/need to be
aware of?

 The Github repository has scored 400+ stars and 150 forks. The developer
 community is growing slowly: the total number now is 14 developers. After
 the second release (with GPU) in December, it is expected that more
 developers will come.

How has the community developed since the last report?

 The community has been active with the development for the next release.
 Since September, there've been 296 and 369 new emails in the dev@ and
 commit@ mailing list respectively.

How has the project developed since the last report?

 The first version was released shortly after the last report. We are
 implementing features using the following schedule:
 http://singa.apache.org/develop/schedule.html.

 The planned features shall be finished soon. The second version is set to
 be released in December.

Date of last release:

 2015-10-08

When were the last committers or PMC members elected?



Signed-off-by:

 [ ](singa) Daniel Dai
 [X](singa) Alan Gates
 [x](singa) Ted Dunning
 [ ](singa) Thejas Nair

Shepherd/Mentor notes:

16 Sep 2015

Singa is a distributed deep learning platform.

Singa has been incubating since 2015-03-17.

Three most important issues to address in the move towards graduation:

 1. Complete unit testing
 2. Follow the release guide to complete the first release
 3. Plan the features for the second release

Any issues that the Incubator PMC (IPMC) or ASF Board wish/need to be
aware of?

 N/A

How has the community developed since the last report?

 1. Researchers from a Stanford group contacted Singa committers regarding
    adding new features to Singa.

 2. Some Singa team members from NUS attended workshops associated with
    SIGMOD15 (Melbourne), VLDB15 (Hawaii), and a local workshop on deep
    learning in Singapore, to extend the reach of the project to developers
    from other institutions and countries.

 3. There are new developers contributing to the code base.

How has the project developed since the last report?

 1. Implementation of the features included in the first release was
    completed. Particularly, we added contrastive divergence algorithm for
    training RBM models. Checkpointing and restoring are also implemented.

 2. Code review is completed for 3/4 of the code. We also tuned and
    optimised the code to improve efficiency and scalability.

 3. User documents are added at
    http://www.comp.nus.edu.sg/~dbsystem/singa//docs.html. We are moving
    them to the Apache website.

Date of last release:

 N/A

When were the last committers or PMC members elected?

 N/A

Signed-off-by:

 [ ](Singa) Daniel Dai
 [X](Singa) Alan Gates
 [ ](Singa) Ted Dunning
 [X](Singa) Thejas Nair

Shepherd/Mentor notes:

 Thejas Nair (thejas):

   The documentation url mentioned above is a staging url. Content will be
   moved/copied to apache website after initial version is ready.

17 Jun 2015

Singa is a distributed platform for deep learning that is easy to use,
extensible and scalable.

Singa has been incubating since 2015-03-17.

Three most important issues to address in the move towards graduation:

 1. Finish the features planed for first release.
 2. Grow the community of developers and users.


Any issues that the Incubator PMC (IPMC) or ASF Board wish/need to be
aware of?

 N/A.

The project Incubation Status file is NOT yet up to date

 Updated.

How has the community developed since the last report?

The community is growing slowly with some individual users and developers
joining.

We are trying to get the first release ready as soon as possible to
interest more developers and users.

What has been done since the last report?

 1. We have set up the incubator website at
 singa.incubator.apache.org. The next step is to update and add
 documentations.

 2. We have finished 1/3 of the features planned for first release
 http://singa.incubator.apache.org/develop/schedule.html. The rest
 should be done in this month. The JIRA tickets and pull requests are
 increasing.

 3. We gave a talk in Danac (http://danac.org/) at Sigmod 2015 to
 publicize Singa .

Date of last release:

 N/A.

Signed-off-by:

 [ ](Singa ) Daniel Dai
 [ ](Singa ) Alan Gates
 [x](Singa ) Ted Dunning
 [ ](Singa ) Thejas Nair

20 May 2015

Singa is a distributed deep learning platform.

Singa has been incubating since 2015-03-17.

Three most important issues to address in the move towards graduation:

 1. Grow the community of developers and users.
 2. Build the homepage and add more documentations.
 2. Prepare for the first release.

Any issues that the Incubator PMC (IPMC) or ASF Board wish/need to be
aware of?

 N/A.

How has the community developed since the last report?

 Three software engineers from NetEase have joined the dev@ list. They have
 been working on deep learning using GPUs for speech processing.

How has the project developed since the last report?

 1. We have updated the Incubation Status file.
 2. We have finished migrating the code to Apache git repository.
 3. We have discussed on the features and development schedule for first
    release.
 4. A new communication framework has been implemented to unify the
    training architectures used by existing systems.
 5. We have added GNU autotools for installing Singa.

Date of last release:

 N/A.

When were the last committers or PMC members elected?

 N/A.

Signed-off-by:

 [ ](singa) Daniel Dai
 [ ](singa) Alan Gates
 [x](singa) Ted Dunning
 [ ](singa) Thejas Nair

22 Apr 2015

SINGA is a distributed platform for deep learning that is easy to use,
exetendible and scalable.

SINGA has been incubating since 2015-03-17.

Three most important issues to address in the move towards graduation:

 1. Grow the community of developers and users.
 2. Finish moving the project to Apache infrastructure.
 3. Add examples, documents, tests and benchmarks for the first release.

Any issues that the Incubator PMC (IPMC) or ASF Board wish/need to be
aware of?

 N/A.

The project Incubation Status file is NOT yet up to date

 The current project URL is: www.comp.nus.edu.sg/~dbsystem/singa

What has been done since the last report?

 N/A. This is the first report.

Other project activities:

 1. The Github repository now has 12 forks as developers are contributing
    features and doing testing.
 2. SINGA is being used for several scientific papers being under
    submission.

What are the plans and expectations for the next period?

 1. Finish moving the project to the Apache's ecosystem.
 2. Focus on the scalability feature, documentation and testing.

Date of last release:

 N/A.

Signed-off-by:

 [ ](SINGA) Daniel Dai
 [ ](SINGA) Alan Gates
 [X](SINGA) Ted Dunning
 [X](SINGA) Thejas Nair