Re: [DISCUSS] Improve code quality

[Wang Wei]

Hi Moaz,

I agree with you.
We did take some efforts to improve the code quality.
[1] introduces some tools for enforcing the coding style.
[2] introduces some tools for adding documentations.

The current issue is that our contributors may have applied different
coding styles and tools using their different editors.
I suggest to do some tests during the CI process, e.g., running the cpplint
and pylint.
If all tests pass, then we merge the PR.

Thanks!

Best,
Wei

[1]. http://singa.apache.org/en/develop/contribute-code.html
[2]. http://singa.apache.org/en/develop/contribute-docs.html

On Sun, Sep 15, 2019 at 2:36 AM Moaz Reyad  wrote:

> Dear team,
>
> Since SINGA is going to graduate soon from the incubator, I propose to use
> some tools to ensure high code quality. These tools check for known
> problems in the code and provide a detailed report for fixing them. May be
> some code came from scientific experimental projects. We need to improve
> this code according to industry standards, so it can be used with more real
> life projects.
>
> 1. I propose to add the code quality tools (cpplint[1], pylint[2] and
> lgtm[3]) to SINGA contribution guideline[4], so that each developer is
> encouraged to install and run code quality checks in his local repo and fix
> any problems before creating a pull request.
>
> 1.A CPP Lint: running cpplint in the src directory shows 822 errors, while
> running in the include directory shows 708 errors. The guidelines [4] has
> an outdated information that instructs developers to use an old
> non-existing file tool/cpplint.py.
>
> 1.B Python Lint: running pylint in python/singa shows 5.00/10 rating, while
> running in python/rafiki shows 0.00/10 rating.
>
> 1.C LGTM :There is a Jira ticket for adding LGTM badges to the README[5],
> so the quality of the code becomes more clear to everyone. LGTM pull
> request automation can't be enabled in Apache repo due to infra
> restrictions[6], but it works on personal forks of the project. Currently
> LGTM rates both C++ and Python code in SINGA as grade D.
>
> 2. I propose also to give the code quality higher priority in the next
> release since it is probably going to be the first release after
> graduation. The team is invited to fix as much as possible from the current
> code issues and to use tools that check their new code before pushing it to
> SINGA. Let's try to make the lgtm grade and lint rating as high as
> possible.
>
> Improving code quality is required to attract new users and developers.
> Users will trust more the project with better code and developers will be
> happy to contribute to it. It will also make the code review process easier
> and more productive instead of wasting time in finding and fixing known
> code problems.
>
> New developers (or old developers who did not contribute for a while and
> would like to warm up) can start working on fixing lgtm and lint issues,
> since they are usually easy and there is a clear explanation of the problem
> and how to solve it.
>
> What do you think?
>
> p.s. This discussion is the first topic in a series of proposals to improve
> SINGA as it will be an Apache top level project soon. The next proposal
> will discuss improving the build and test pipeline in a separate thread to
> avoid discussing too many things in one thread.
>
> best regards,
> Moaz
>
> [1] https://pypi.org/project/cpplint/
> [2] https://www.pylint.org/
> [3] https://lgtm.com/
> [4] http://singa.apache.org/develop/contribute-code.html
> [5] https://issues.apache.org/jira/projects/SINGA/issues/SINGA-484
> [6] https://issues.apache.org/jira/browse/INFRA-17954
>

Re: [VOTE] Graduate Apache SINGA as TLP

[Wang Wei]

+1
Thanks, Moaz!

Regards,
Wei

On Wed, Aug 21, 2019 at 8:53 PM Moaz Reyad  wrote:

> Dear All,
>
> Apache SINGA entered the incubator on March 2015. Since then, the community
> has grown and several releases were published.
>
> The last incubator report listed SINGA as "Nearing graduation":
> https://cwiki.apache.org/confluence/display/INCUBATOR/June2019
>
> The maturity assessment of SINGA can be found here:
> https://cwiki.apache.org/confluence/display/SINGA/Maturity+model+assessment
>
> The graduation resolution can be found here:
> https://cwiki.apache.org/confluence/display/SINGA/Graduation+Resolution
>
> Please take a minute to vote on whether or not Apache SINGA should graduate
> to a Top Level Project by responding with one of the following:
>
>  [ ] +1 Apache SINGA should graduate.
>  [ ] +0 No opinion
>  [ ] -1 Apache SINGA should not graduate (please provide the reason)
>
> The VOTE is open for a minimum of 72 hours.
>
> Thank you,
> Moaz
>

Re: New services for SINGA (Mentors help needed)

[Wang Wei]

Hi Moaz,

Thanks.
Pls check my comments below.

Regards,
Wei

On Sat, Aug 3, 2019 at 10:51 PM Moaz Reyad  wrote:

> Hi Wei,
>
> Yes, the repo is : https://github.com/apache/incubator-singa-site
>
> It can be updated using the standard git commands:
>
> git clone https://github.com/apache/incubator-singa-site.git
> git commit
> and so on...
>
> However, we still did not ask INFRA to update the pubsub to use git instead
> of svn. So now the old SVN repository is still active and the new git
> repository is not yet connected to the website (singa.apache.org).
>
> We should create a ticket similar to this one:
> https://issues.apache.org/jira/browse/INFRA-16990


 Ok. I will do it.

>
>
> and specify a git branch and a folder in the repo (
> https://github.com/apache/incubator-singa-site) to be used as the main
> folder of the website.
>
> For instance, may be we use the master branch to store the "source code" of
> the website, and use another branch such as "asf-site" to contain the
> actual built website after running Sphinx site build script. Currently, the
> source code of the website is at this folder in the original git repo:
> https://github.com/apache/incubator-singa/tree/master/doc, and the master
> branch of https://github.com/apache/incubator-singa-site contains the
> built
> website.
>
Since Sphinx needs the source Python code to generate the API, I think it
would be good to put the documentation files in the
https://github.com/apache/incubator-singa
<https://github.com/apache/incubator-singa/tree/master/doc>.

>
> We can complicate it further by using other git branches in
> incubator-singa-site for different versions (1.0.0, 2.0.0, ..) and
> different languages (Chinese, Korean, ..). Then the site build script
> (build.sh) can be updated to perform "git clone" on them and build all of
> them, and finally push them to the asf-site branch.
>
I agree that it would be good to have a branch for storing the html files
of each version .

>
> What do you think?
> Moaz
> On Sat, Aug 3, 2019 at 4:34 PM Wang Wei  wrote:
>
> > Hi Moaz,
> >
> > If I want to update incubator-singa-site, what's the upstream to push to?
> > Is there a repo at gitbox.apache.org?
> > Thanks.
> >
> > Regards,
> > Wei
> >
> > On Thu, Aug 1, 2019 at 8:33 PM Moaz Reyad  wrote:
> >
> > > The notebooks were not in the SVN. They are in singa github:
> > >
> > >
> >
> https://github.com/apache/incubator-singa/tree/master/doc/en/docs/notebook
> > >
> > > Sphinx generates the website pages which contain links to the URLs in
> the
> > > above folder.
> > >
> > > Regards,
> > > Moaz
> > >
> > > On Thu, Jul 25, 2019 at 1:00 PM Wang Wei 
> > wrote:
> > >
> > > > Thanks, Moaz!
> > > > Github can display notebooks directly. But I cannot find the
> > notebooks..
> > > > Did Sphinx ignore them?
> > > >
> > > > Regards,
> > > > Wei
> > > >
> > > > On Thu, Jul 25, 2019 at 6:12 PM Moaz Reyad  wrote:
> > > >
> > > > > I pushed the current site from SVN to the new repo:
> > > > >
> > > > > https://github.com/apache/incubator-singa-site
> > > > >
> > > > > Please check and if everything is fine, we can ask INFRA to perform
> > the
> > > > > switch to this new repository.
> > > > >
> > > > > Best regards,
> > > > > Moaz
> > > > >
> > > > > On Thu, Jul 11, 2019 at 3:48 AM Wang Wei 
> > > > wrote:
> > > > >
> > > > > > Thanks, Alan. We will start using the new repo for the website.
> > > > > >
> > > > > > Regards,
> > > > > > Wei
> > > > > >
> > > > > > On Thu, Jul 11, 2019 at 4:14 AM Alan Gates  >
> > > > wrote:
> > > > > >
> > > > > > > I've created incubator-singa-site.git.  You should now be able
> to
> > > > port
> > > > > > the
> > > > > > > contents over from the SVN site repo and then infra can switch
> > your
> > > > > > pubsub
> > > > > > > from there.
> > > > > > >
> > > > > > > Alan.
> > > > > > >
> > > > > > > On Sun, Jun 30, 2019 at 6:41 AM Alan Gates <
> alanfga...@gmail.com
> > >
> > > > > wrote:
&g

Re: New services for SINGA (Mentors help needed)

[Wang Wei]

Hi Moaz,

If I want to update incubator-singa-site, what's the upstream to push to?
Is there a repo at gitbox.apache.org?
Thanks.

Regards,
Wei

On Thu, Aug 1, 2019 at 8:33 PM Moaz Reyad  wrote:

> The notebooks were not in the SVN. They are in singa github:
>
> https://github.com/apache/incubator-singa/tree/master/doc/en/docs/notebook
>
> Sphinx generates the website pages which contain links to the URLs in the
> above folder.
>
> Regards,
> Moaz
>
> On Thu, Jul 25, 2019 at 1:00 PM Wang Wei  wrote:
>
> > Thanks, Moaz!
> > Github can display notebooks directly. But I cannot find the notebooks..
> > Did Sphinx ignore them?
> >
> > Regards,
> > Wei
> >
> > On Thu, Jul 25, 2019 at 6:12 PM Moaz Reyad  wrote:
> >
> > > I pushed the current site from SVN to the new repo:
> > >
> > > https://github.com/apache/incubator-singa-site
> > >
> > > Please check and if everything is fine, we can ask INFRA to perform the
> > > switch to this new repository.
> > >
> > > Best regards,
> > > Moaz
> > >
> > > On Thu, Jul 11, 2019 at 3:48 AM Wang Wei 
> > wrote:
> > >
> > > > Thanks, Alan. We will start using the new repo for the website.
> > > >
> > > > Regards,
> > > > Wei
> > > >
> > > > On Thu, Jul 11, 2019 at 4:14 AM Alan Gates 
> > wrote:
> > > >
> > > > > I've created incubator-singa-site.git.  You should now be able to
> > port
> > > > the
> > > > > contents over from the SVN site repo and then infra can switch your
> > > > pubsub
> > > > > from there.
> > > > >
> > > > > Alan.
> > > > >
> > > > > On Sun, Jun 30, 2019 at 6:41 AM Alan Gates 
> > > wrote:
> > > > >
> > > > > > I’m out on vacation and won’t be back until July 9th. I’ll be
> happy
> > > to
> > > > do
> > > > > > it then if one of the other mentors doesn’t get to it first.
> > > > > >
> > > > > > Alan.
> > > > > >
> > > > > > Sent from my iPhone
> > > > > >
> > > > > > > On Jun 30, 2019, at 04:47, Moaz Reyad  wrote:
> > > > > > >
> > > > > > > Dear Alan,
> > > > > > >
> > > > > > > Can we create the new git repository incubator-singa-site ?
> > > > > > >
> > > > > > > Thank you,
> > > > > > > Moaz
> > > > > > >
> > > > > > >> On 2019/04/04 16:23:32, Moaz Reyad  wrote:
> > > > > > >> Dear Alan,
> > > > > > >>
> > > > > > >> Thank you for the help.
> > > > > > >>
> > > > > > >> About Gitbox, the INFRA-17328 issue created a git repository
> for
> > > the
> > > > > > source
> > > > > > >> code but not for the site.
> > > > > > >>
> > > > > > >> We would like to create a new git repository
> > > (incubator-singa-site)
> > > > > > instead
> > > > > > >> of the current SVN repository (
> > > > > > >> https://svn.apache.org/repos/asf/incubator/singa/site/)
> > > > > > >>
> > > > > > >> Thank you,
> > > > > > >> Moaz
> > > > > > >>
> > > > > > >>> On Wed, Apr 3, 2019 at 7:49 PM Alan Gates 
> > > > wrote:
> > > > > > >>>
> > > > > > >>> https://cwiki.apache.org/confluence/display/SINGA has been
> > > > created.
> > > > > > >>> Moaz, I've made you the admin, you should be able to add
> > others.
> > > > > > >>>
> > > > > > >>> Regarding gitbox, it appears that is finished based on the
> > JIRA.
> > > > Is
> > > > > > there
> > > > > > >>> something more to do?
> > > > > > >>>
> > > > > > >>> I have submitted a request to create the security email list
> > and
> > > > made
> > > > > > moaz
> > > > > > >>> and wangwei the admins.  This is a private list, so you
> should
> > > only
> > > > > 

Re: New services for SINGA (Mentors help needed)

[Wang Wei]

Thanks, Moaz!
Github can display notebooks directly. But I cannot find the notebooks..
Did Sphinx ignore them?

Regards,
Wei

On Thu, Jul 25, 2019 at 6:12 PM Moaz Reyad  wrote:

> I pushed the current site from SVN to the new repo:
>
> https://github.com/apache/incubator-singa-site
>
> Please check and if everything is fine, we can ask INFRA to perform the
> switch to this new repository.
>
> Best regards,
> Moaz
>
> On Thu, Jul 11, 2019 at 3:48 AM Wang Wei  wrote:
>
> > Thanks, Alan. We will start using the new repo for the website.
> >
> > Regards,
> > Wei
> >
> > On Thu, Jul 11, 2019 at 4:14 AM Alan Gates  wrote:
> >
> > > I've created incubator-singa-site.git.  You should now be able to port
> > the
> > > contents over from the SVN site repo and then infra can switch your
> > pubsub
> > > from there.
> > >
> > > Alan.
> > >
> > > On Sun, Jun 30, 2019 at 6:41 AM Alan Gates 
> wrote:
> > >
> > > > I’m out on vacation and won’t be back until July 9th. I’ll be happy
> to
> > do
> > > > it then if one of the other mentors doesn’t get to it first.
> > > >
> > > > Alan.
> > > >
> > > > Sent from my iPhone
> > > >
> > > > > On Jun 30, 2019, at 04:47, Moaz Reyad  wrote:
> > > > >
> > > > > Dear Alan,
> > > > >
> > > > > Can we create the new git repository incubator-singa-site ?
> > > > >
> > > > > Thank you,
> > > > > Moaz
> > > > >
> > > > >> On 2019/04/04 16:23:32, Moaz Reyad  wrote:
> > > > >> Dear Alan,
> > > > >>
> > > > >> Thank you for the help.
> > > > >>
> > > > >> About Gitbox, the INFRA-17328 issue created a git repository for
> the
> > > > source
> > > > >> code but not for the site.
> > > > >>
> > > > >> We would like to create a new git repository
> (incubator-singa-site)
> > > > instead
> > > > >> of the current SVN repository (
> > > > >> https://svn.apache.org/repos/asf/incubator/singa/site/)
> > > > >>
> > > > >> Thank you,
> > > > >> Moaz
> > > > >>
> > > > >>> On Wed, Apr 3, 2019 at 7:49 PM Alan Gates 
> > wrote:
> > > > >>>
> > > > >>> https://cwiki.apache.org/confluence/display/SINGA has been
> > created.
> > > > >>> Moaz, I've made you the admin, you should be able to add others.
> > > > >>>
> > > > >>> Regarding gitbox, it appears that is finished based on the JIRA.
> > Is
> > > > there
> > > > >>> something more to do?
> > > > >>>
> > > > >>> I have submitted a request to create the security email list and
> > made
> > > > moaz
> > > > >>> and wangwei the admins.  This is a private list, so you should
> only
> > > > accept
> > > > >>> people on the PPMC or Apache members onto the list.
> > > > >>>
> > > > >>> Alan.
> > > > >>>
> > > > >>>> On Mon, Apr 1, 2019 at 1:08 AM Moaz Reyad 
> > wrote:
> > > > >>>>
> > > > >>>> Dear SINGA mentors,
> > > > >>>>
> > > > >>>> What is your advice about this request to add new services for
> > > SINGA?
> > > > >>>>
> > > > >>>> Best regards
> > > > >>>> Moaz
> > > > >>>>
> > > > >>>> -- Forwarded message -
> > > > >>>> From: Moaz Reyad 
> > > > >>>> Date: Sat, Mar 9, 2019 at 11:39 AM
> > > > >>>> Subject: New services for SINGA (Mentors help needed)
> > > > >>>> To: 
> > > > >>>>
> > > > >>>>
> > > > >>>> Hi All,
> > > > >>>>
> > > > >>>> I would like to propose three new services for SINGA. These
> > services
> > > > can
> > > > >>>> be requested from (https://selfserve.apache.org/) but I do not
> > have
> > > > >>>> permission to create them. I ope

Re: New services for SINGA (Mentors help needed)

[Wang Wei]

Thanks, Alan. We will start using the new repo for the website.

Regards,
Wei

On Thu, Jul 11, 2019 at 4:14 AM Alan Gates  wrote:

> I've created incubator-singa-site.git.  You should now be able to port the
> contents over from the SVN site repo and then infra can switch your pubsub
> from there.
>
> Alan.
>
> On Sun, Jun 30, 2019 at 6:41 AM Alan Gates  wrote:
>
> > I’m out on vacation and won’t be back until July 9th. I’ll be happy to do
> > it then if one of the other mentors doesn’t get to it first.
> >
> > Alan.
> >
> > Sent from my iPhone
> >
> > > On Jun 30, 2019, at 04:47, Moaz Reyad  wrote:
> > >
> > > Dear Alan,
> > >
> > > Can we create the new git repository incubator-singa-site ?
> > >
> > > Thank you,
> > > Moaz
> > >
> > >> On 2019/04/04 16:23:32, Moaz Reyad  wrote:
> > >> Dear Alan,
> > >>
> > >> Thank you for the help.
> > >>
> > >> About Gitbox, the INFRA-17328 issue created a git repository for the
> > source
> > >> code but not for the site.
> > >>
> > >> We would like to create a new git repository (incubator-singa-site)
> > instead
> > >> of the current SVN repository (
> > >> https://svn.apache.org/repos/asf/incubator/singa/site/)
> > >>
> > >> Thank you,
> > >> Moaz
> > >>
> > >>> On Wed, Apr 3, 2019 at 7:49 PM Alan Gates  wrote:
> > >>>
> > >>> https://cwiki.apache.org/confluence/display/SINGA has been created.
> > >>> Moaz, I've made you the admin, you should be able to add others.
> > >>>
> > >>> Regarding gitbox, it appears that is finished based on the JIRA.  Is
> > there
> > >>> something more to do?
> > >>>
> > >>> I have submitted a request to create the security email list and made
> > moaz
> > >>> and wangwei the admins.  This is a private list, so you should only
> > accept
> > >>> people on the PPMC or Apache members onto the list.
> > >>>
> > >>> Alan.
> > >>>
> >  On Mon, Apr 1, 2019 at 1:08 AM Moaz Reyad  wrote:
> > 
> >  Dear SINGA mentors,
> > 
> >  What is your advice about this request to add new services for
> SINGA?
> > 
> >  Best regards
> >  Moaz
> > 
> >  -- Forwarded message -
> >  From: Moaz Reyad 
> >  Date: Sat, Mar 9, 2019 at 11:39 AM
> >  Subject: New services for SINGA (Mentors help needed)
> >  To: 
> > 
> > 
> >  Hi All,
> > 
> >  I would like to propose three new services for SINGA. These services
> > can
> >  be requested from (https://selfserve.apache.org/) but I do not have
> >  permission to create them. I opened an INFRA ticket (
> >  https://issues.apache.org/jira/browse/INFRA-17981), but the INFRA
> > team
> >  suggested that we ask assistance from mentors.
> > 
> >  1. Confluence wiki:
> > 
> >  We need to create a space for SINGA at the Confluence wiki. The wiki
> > is
> >  required to keep the development pages which are not directly useful
> > to
> >  SINGA users. For example: the development schedule, how to prepare a
> >  release, maturity assessment, etc.
> > 
> >  This issue was open before (
> >  https://issues.apache.org/jira/browse/INFRA-9473) but it was not
> >  completed.
> > 
> >  Please add my account:
> > 
> >  Wiki: https://cwiki.apache.org/confluence/display/~moazreyad
> >  Apache ID: moaz
> > 
> >  SINGA team members who would like to contribute to the wiki may
> share
> >  their wiki account, so they can have the write permission on the
> wiki
> > pages.
> > 
> >  We can also try to enable the GitHub wiki, but I think Apache does
> not
> >  allow this any more and the ASF encourages the use of Confluence
> > instead.
> > 
> >  2. GitBox repository for incubator-singa-site:
> > 
> >  This is required to move the singa site from SVN to Git. For more
> >  information see (https://issues.apache.org/jira/browse/INFRA-17328)
> > 
> >  3. Security mail list:
> > 
> >  We would like to create the private address:
> > secur...@singa.apache.org
> >  for security issues. There are some vulnerabilities already reported
> > to
> >  SINGA and we need to resolve them. The security mail list will help
> in
> >  handling current and future security issues.
> > 
> >  Please add wang...@apache.org to this private mail list. For more
> >  information see (https://issues.apache.org/jira/browse/SINGA-417).
> > SINGA
> >  team members who would like to join the security team may also share
> > their
> >  email.
> > 
> >  Best regards,
> >  Moaz
> > 
> > >>>
> > >>
> >
>

Re: [jira] [Commented] (SINGA-449) Preparing Podling TLP resolution

[Wang Wei]

 Hi Moaz,

Any specific requirements or eligibility criteria for the PMC chair?

Regards,
Wei

On Thu, Jun 27, 2019 at 10:14 AM Moaz Reyad (JIRA)  wrote:

>
> [
> https://issues.apache.org/jira/browse/SINGA-449?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16873757#comment-16873757
> ]
>
> Moaz Reyad commented on SINGA-449:
> --
>
> This issue is still waiting for discussion. We need to choose the project
> chair (in the private list) and the initial project members to complete the
> Podling TLP resolution.
>
> Should we copy all the current ppmc members to the list of initial project
> members without asking their permission (assuming everyone agrees) or we
> must ask every member to explicitly agree to be added to the initial
> project members?
>
> > Preparing Podling TLP resolution
> > 
> >
> > Key: SINGA-449
> > URL: https://issues.apache.org/jira/browse/SINGA-449
> > Project: Singa
> >  Issue Type: Task
> >Reporter: Moaz Reyad
> >Priority: Major
> >
> > This issue is for preparing the graduation resolution for SINGA given
> the [template|
> https://svn.apache.org/repos/private/committers/board/templates/podling-tlp-resolution.txt
> ].
> > The resolution draft for SINGA is [here|
> https://cwiki.apache.org/confluence/display/SINGA/Graduation+Resolution].
> > Please check it and give your feedback.
> > SINGA PPMC need to discuss and choose the project Chair (see private
> mail list).
> > We also need to specify initial project members. We can copy all the
> current members without asking their permission (assuming everyone agrees)
> or we may ask every member to explicitly agree to be added to the initial
> project members.
>
>
>
> --
> This message was sent by Atlassian JIRA
> (v7.6.3#76005)
>

Re: [jira] [Created] (SINGA-459) Remove support for Python 2.7

[Wang Wei]

+1.

On Thu, May 23, 2019 at 8:32 PM Moaz Reyad (JIRA)  wrote:

> Moaz Reyad created SINGA-459:
> 
>
>  Summary: Remove support for Python 2.7
>  Key: SINGA-459
>  URL: https://issues.apache.org/jira/browse/SINGA-459
>  Project: Singa
>   Issue Type: Improvement
> Reporter: Moaz Reyad
>
>
> The [support of Python 2 will end soon|https://pythonclock.org/] and some
> projects already stopped supporting it. (For example,[scikit-learn|
> https://github.com/scikit-learn/scikit-learn#installation] 0.20 was the
> last version to support Python2.7).
>
> If everyone agrees, we can plan to remove the Python 2 support from SINGA
> starting from the next release.
>
> This will help in improving the python code and simplifying the building
> process.
>
> For the code improvement: currently the code uses 'future' library to
> support both python 2 and 3. This will not be needed if only python 3 code
> is supported. There is also string encoding code that is complicated by the
> need to be backward compatible with python 2.
>
> For the building process: the build options will contain only python 3 by
> default. The developers will not confuse python 2 and python 3 options and
> dependencies. Testing will be easier because only python 3 scenarios will
> be tested.
>
>
>
> --
> This message was sent by Atlassian JIRA
> (v7.6.3#76005)
>

Re: [jira] [Created] (SINGA-458) ONNX backend test failed

[Wang Wei]

Do we need to pass all 1116 tests?



On Tue, May 21, 2019 at 9:14 PM Moaz Reyad (JIRA)  wrote:

> Moaz Reyad created SINGA-458:
> 
>
>  Summary: ONNX backend test failed
>  Key: SINGA-458
>  URL: https://issues.apache.org/jira/browse/SINGA-458
>  Project: Singa
>   Issue Type: Bug
> Reporter: Moaz Reyad
>
>
> The test of [SINGA ONNX backend|
> https://github.com/apache/incubator-singa/blob/master/python/singa/sonnx.py#L224]
> using [onnx.backend.test.BackendTest|
> https://github.com/apache/incubator-singa/pull/462/commits/6dec4b24df8e469255c7399600eef27773f92052]
> failed.
>
> There are 1116 unit tests and all of them fail with the error:
>
>
> {code:sh}
> Traceback (most recent call last):
>   File
> "/home/moaz/singa/venv/local/lib/python2.7/site-packages/onnx/backend/test/runner/__init__.py",
> line 248, in device_test_func
> return test_func(*args, device=device, **kwargs)
>   File
> "/home/moaz/singa/venv/local/lib/python2.7/site-packages/onnx/backend/test/runner/__init__.py",
> line 310, in run
> outputs = list(prepared_model.run(inputs))
>   File
> "/home/moaz/singa/venv/local/lib/python2.7/site-packages/singa/sonnx.py",
> line 198, in run
> tensors[x.name] = x
> AttributeError: 'numpy.ndarray' object has no attribute 'name'
>
> --
>
> {code}
>
>
>
>
>
> --
> This message was sent by Atlassian JIRA
> (v7.6.3#76005)
>

Re: [EU-FOSSA 2 2019 Hackathon] Integration between SINGA and PLC4X

[Wang Wei]

Hi all,

+1. Looking forward to the collaborations!

Regards,
Wei (from SINGA team)

On Sun, May 5, 2019 at 10:33 PM Julian Feinauer <
j.feina...@pragmaticminds.de> wrote:

> Hi Moaz,
>
> Big +1 from me... It really was a pleasure with you! And I guess toddy and
> your lion will make a pretty lovely pair :)
>
> Julian
>
> Von meinem Mobiltelefon gesendet
>
>
>  Ursprüngliche Nachricht 
> Betreff: [EU-FOSSA 2 2019 Hackathon] Integration between SINGA and PLC4X
> Von: Moaz Reyad
> An: d...@plc4x.apache.org,dev@singa.incubator.apache.org
> Cc: "Saranjit-Singh.ARORA" ,Suwon Ham ,Sally Khudairi ,Christofer Dutz
>
> Dear SINGA and PLC4X dev,
>
> I met with Christofer in the EU-FOSSA 2 2019 Hackathon and we discussed a
> possible collaboration between the two projects.
>
> More specifically, we could check the integration between SINGA I/O modules
> and PLC4X which will probably allow better communication between industrial
> IoT and machine learning.
>
> PLC4X already implemented (or planned to implemented) integrations with
> several projects and a deep learning integration module with SINGA can be
> useful for both projects.
>
> Both dev teams are welcome to share their ideas and discuss this further.
>
> Best regards,
> Moaz
>

Re: [EU-FOSSA 2 2019 Hackathon] SINGA and SpamAssassin collaboration in GSoC 2019

[Wang Wei]

+1.

I have registered as a GSoC mentor.

Regards,
Wei

On Sun, May 5, 2019 at 10:26 PM Moaz Reyad  wrote:

> Dear SINGA and SpamAssassin dev,
>
> I met with Giovanni Bechis in the EU-FOSSA 2 2019 Hackathon and we
> discussed a possible collaboration between the two projects.
>
> SpamAssassin has a proposal in Google Summer of Code 2019 about Statistical
> Classifier Plugin for spam email detection. If both teams and the student
> agree, we may consider building this plugin with SINGA.
>
> The GSoC guide says "Some organizations choose to assign more than one
> mentor to each of their students.". This means Apache can assign mentors
> from SINGA and SpamAssassin to help in developing this plugin. We have two
> mentors from SINGA in GSoC and one or both of them can help in co-mentoring
> this project.
>
> Best regards,
> Moaz
>

Re: [jira] [Created] (SINGA-452) Adding the ConvLSTM layer implementation

[Wang Wei]

+1.

It would be better to use implement it as a new layer that combines the
conv and lstm operations here
https://github.com/apache/incubator-singa/blob/master/python/singa/autograd.py

On Sun, May 5, 2019 at 3:29 AM Faouzi Amrouche (JIRA) 
wrote:

> Faouzi Amrouche created SINGA-452:
> -
>
>  Summary: Adding the ConvLSTM layer implementation
>  Key: SINGA-452
>  URL: https://issues.apache.org/jira/browse/SINGA-452
>  Project: Singa
>   Issue Type: Improvement
> Reporter: Faouzi Amrouche
>
>
> ConvLSTM is an improved version of the LSTM layer. It replaces the
> traditional matrix multiplications by convolutions. In this new
> implementation, the input transformations and recurrent transformations are
> both convolutional.
>
> This layer has become really important recently and is already supported
> in many existing frameworks (Keras, Tensorflow...etc).
>
>
>
> --
> This message was sent by Atlassian JIRA
> (v7.6.3#76005)
>

Re: [jira] [Comment Edited] (SINGA-450) Docker Hub tag 1.2.0-cpu-devel-ubuntu18.04 is missing Singa

[Wang Wei]

Currently, all devel images put the code at $HOME/incubator-singa, e.g.,
https://github.com/apache/incubator-singa/blob/master/tool/docker/devel/ubuntu/cuda10/Dockerfile#L77
If we want to move it to /incubator-singa, we have to update all images.
But we need a reason for this change.

On Sun, May 5, 2019 at 1:00 AM Alexander Davis (JIRA) 
wrote:

>
> [
> https://issues.apache.org/jira/browse/SINGA-450?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16833109#comment-16833109
> ]
>
> Alexander Davis edited comment on SINGA-450 at 5/4/19 4:59 PM:
> ---
>
> I have fixed the location issue of the git clone and build [here|[
> https://github.com/apache/incubator-singa/pull/459]]
>
>
>
> The git clone clones to /root/incubator-singa when it should be cloned to
> /incubator-singa which is where the user is sent to
>
>
> was (Author: alexander7161):
> I have fixed the location issue of the git clone and build [here|[
> https://github.com/apache/incubator-singa/pull/459]]
>
> > Docker Hub tag 1.2.0-cpu-devel-ubuntu18.04 is missing Singa
> > ---
> >
> > Key: SINGA-450
> > URL: https://issues.apache.org/jira/browse/SINGA-450
> > Project: Singa
> >  Issue Type: Bug
> >Reporter: Alexander Davis
> >Priority: Major
> > Attachments: image-2019-05-04-15-15-44-913.png
> >
> >  Time Spent: 10m
> >  Remaining Estimate: 0h
> >
> > See [https://hub.docker.com/r/apache/singa/tags], 
> > 1.2.0-cpu-devel-ubuntu18.04
> has a much smaller size than other releases. I have tested it and
> the incubator-singa folder appears to be empty.
> !image-2019-05-04-15-15-44-913.png!
>
>
>
> --
> This message was sent by Atlassian JIRA
> (v7.6.3#76005)
>

Re: SINGA team at EU FOSSA Apache Hackathon

[Wang Wei]

Hi all,

Welcome and looking forward to your contributions!

Regards,
Wei

On Sat, May 4, 2019 at 9:50 PM Moaz Reyad  wrote:

> Dear SINGA dev,
>
> I would like to introduce the team who is working on SINGA in the Apache
> Hackathon.
>
> May be each one of the hackathon team can reply and introduce himself and
> what he plans to do, so the dev team can help you with your plan. Some of
> you already created Jira tickets and pull requests.
>
> Thank you,
> Moaz
>

Re: [VOTE] Release apache-singa-incubating-2.0.0 (RC1)

[Wang Wei]

+1 (binding)

I have run apache-rat for the license header check.

Regards,
Wei

On Sun, Apr 7, 2019 at 9:20 PM Beng Chin OOI  wrote:

>
> +1
>
> regards
> beng chin
>
> On 2019-04-07 15:41, Chonho Lee wrote:
> > Hi all,
> >
> > We have created a build for Apache SINGA 2.0.0-incubating, release
> > candidate RC1.
> >
> > The artifacts to be voted on are located here:
> > https://dist.apache.org/repos/dist/dev/incubator/singa/2.0.0/
> >
> > The hashes of the artifacts are as follows:
> > SHA512: D422ED7C 1AD370CA 66D5591B E581DC3C 2513396E 465CF983 80785929
> > A70189AA 8944452A 1B10AF23 03F781BF 5D4066CE D994FE97 123280A1 EA34A681
> > 3402BA12
> >
> > Release artifacts are signed with the following key:
> > https://people.apache.org/keys/committer/dinhtta.asc
> >
> > and the signature file is:
> >
> https://dist.apache.org/repos/dist/dev/incubator/singa/2.0.0/apache-singa-incubating-2.0.0-RC1.tar.gz.asc
> >
> >
> > The Github tag is at:
> > https://github.com/apache/incubator-singa/releases/tag/2.0.0rc1
> > commit ID is: 5f14f3dbc005a64fdbc508bff63db92e6d3e05b6
> >
> > To check the license, you can use the Apache Rat tool as follows:
> > 1. download & decompress apache rat from
> > http://creadur.apache.org/rat/download_rat.cgi
> > 2. run the following command under singa folder:
> > java -jar /PATH/TO/RAT/apache-rat-0.11.jar -E rat-excludes -d . >
> > rat_check
> > 3. check the results in file named "rat_check"
> >
> > Please check and vote on releasing this package. The vote is open for
> > at
> > least 72 hours and passes if a majority of at least three +1 votes are
> > cast.
> >
> > [ ] +1 Release this package as Apache SINGA 2.0.0-incubating
> > [ ]  0 I don't feel strongly about it, but I'm okay with the release
> > [ ] -1 Do not release this package because...
> >
> > Here is my vote:
> > +1
>

Re: [jira] [Commented] (PODLINGNAMESEARCH-159) Establish whether "Apache SINGA" is a suitable name

[Wang Wei]

Thanks, Moaz!

Cheers!

On Mon, Mar 11, 2019 at 8:24 PM Moaz Reyad  wrote:

> Hi All,
>
> The Apache Brand Management has approved the name "Apache SINGA".
>
> Best regards,
> Moaz
>
> -- Forwarded message -
> From: Mark Thomas (JIRA) 
> Date: Sun, Mar 10, 2019 at 11:24 PM
> Subject: [jira] [Commented] (PODLINGNAMESEARCH-159) Establish whether
> "Apache SINGA" is a suitable name
>
> [
>
> https://issues.apache.org/jira/browse/PODLINGNAMESEARCH-159?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16789046#comment-16789046
> ]
>
> Mark Thomas commented on PODLINGNAMESEARCH-159:
> ---
>
> Approved.
>
> Mark
> VP, Brand Management
>
> > Establish whether "Apache SINGA" is a suitable name
> > ---
> >
> > Key: PODLINGNAMESEARCH-159
> > URL:
> https://issues.apache.org/jira/browse/PODLINGNAMESEARCH-159
> > Project: Podling Suitable Names Search
> >  Issue Type: Suitable Name Search
> >Reporter: Moaz Reyad
> >
> > SINGA is a distributed deep learning platform which is currently in the
> process of graduation from the incubator (
> https://issues.apache.org/jira/browse/SINGA-405).
> > This issue is open to establish whether "Apache SINGA" is a suitable
> name. A name search (https://incubator.apache.org/guides/names.html) is
> required to complete the graduation.
>
>
>
> --
> This message was sent by Atlassian JIRA
> (v7.6.3#76005)
>

March Report Draft

[Wang Wei]

Hi all,

Here is a draft.
@Anh Dinh  , could you help to upload it asap. Thanks.
-
SINGA

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. Release version 2.0 which was scheduled in Dec 2018 but has been
delayed
  2. Start the graduation process in about 2-3 month
  3.

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?

  We have a new PPMC member, Moaz, who was a committer.
  We have a new committer, Wanqi Xue.

  Number of emails from Sep to Dec.
  56 Dec 2018
  22 Jan 2019
  22 Feb 2019
  15 Mar 2019

How has the project developed since the last report?

  We have done the name search for "Apache Singa"
  We enhanced Singa's performance on Intel CPU by calling Intel's MKLDNN
library.
  We have updated the website and documentation to include the instructions
for "How to prepare a release".

  There are 96 Commits since last report.

How would you assess the podling's maturity?
Please feel free to add your own commentary.

  We will start the graduation process after releasing V2.0.

  [ ] Initial setup
  [ ] Working towards first release
  [ ] Community building
  [X] Nearing graduation
  [ ] Other:

Date of last release:

  2018-06-06

When were the last committers or PPMC members elected?

  2018-12-21

Have your mentors been helpful and responsive or are things falling
through the cracks? In the latter case, please list any open issues
that need to be addressed.

  The mentors are helpful, e.g., they have given the suggestions on name
search and release policy.

Signed-off-by:

  [ ](singa) Alan Gates
 Comments:
  [ ](singa) Ted Dunning
 Comments:
  [ ](singa) Thejas Nair
 Comments:

Re: [jira] [Commented] (SINGA-428) Move Docker images under Apache user name

[Wang Wei]

How can we push to https://hub.docker.com/r/apache/singa?

On Tue, Feb 26, 2019 at 1:40 PM Wang Wei  wrote:

> +1. Rafiki is using docker.
>
> On Tue, Feb 26, 2019 at 12:58 PM Moaz Reyad (JIRA) 
> wrote:
>
>>
>> [
>> https://issues.apache.org/jira/browse/SINGA-428?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16777577#comment-16777577
>> ]
>>
>> Moaz Reyad commented on SINGA-428:
>> --
>>
>> May be also we can put Rafiki Docker images in [SINGA DockerHub page|
>> https://hub.docker.com/r/apache/singa/]. This can be done by building
>> the Admin, Advisor, Predictor and Worker images in the Rafiki/dockerfiles
>> (from an official release of SINGA) and push them to the DockerHub. What do
>> you think?
>>
>> > Move Docker images under Apache user name
>> > -
>> >
>> > Key: SINGA-428
>> > URL: https://issues.apache.org/jira/browse/SINGA-428
>> > Project: Singa
>> >  Issue Type: Bug
>> >Reporter: Moaz Reyad
>> >Priority: Major
>> >  Time Spent: 10m
>> >  Remaining Estimate: 0h
>> >
>> > The Docker images of SINGA are now on Docker Hub [here|
>> https://hub.docker.com/r/nusdbsystem/singa/]. They are not under the
>> [Apache docker username|https://hub.docker.com/u/apache/].
>> > To solve this problem, we create a ticket at INFRA to provide the name
>> space SINGA.
>> > We also need to make sure that the docker images of SINGA are built
>> using official releases and not using latest (master) branch which is not
>> official.
>>
>>
>>
>> --
>> This message was sent by Atlassian JIRA
>> (v7.6.3#76005)
>>
>

Re: [jira] [Commented] (SINGA-428) Move Docker images under Apache user name

[Wang Wei]

+1. Rafiki is using docker.

On Tue, Feb 26, 2019 at 12:58 PM Moaz Reyad (JIRA)  wrote:

>
> [
> https://issues.apache.org/jira/browse/SINGA-428?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16777577#comment-16777577
> ]
>
> Moaz Reyad commented on SINGA-428:
> --
>
> May be also we can put Rafiki Docker images in [SINGA DockerHub page|
> https://hub.docker.com/r/apache/singa/]. This can be done by building the
> Admin, Advisor, Predictor and Worker images in the Rafiki/dockerfiles (from
> an official release of SINGA) and push them to the DockerHub. What do you
> think?
>
> > Move Docker images under Apache user name
> > -
> >
> > Key: SINGA-428
> > URL: https://issues.apache.org/jira/browse/SINGA-428
> > Project: Singa
> >  Issue Type: Bug
> >Reporter: Moaz Reyad
> >Priority: Major
> >  Time Spent: 10m
> >  Remaining Estimate: 0h
> >
> > The Docker images of SINGA are now on Docker Hub [here|
> https://hub.docker.com/r/nusdbsystem/singa/]. They are not under the
> [Apache docker username|https://hub.docker.com/u/apache/].
> > To solve this problem, we create a ticket at INFRA to provide the name
> space SINGA.
> > We also need to make sure that the docker images of SINGA are built
> using official releases and not using latest (master) branch which is not
> official.
>
>
>
> --
> This message was sent by Atlassian JIRA
> (v7.6.3#76005)
>

[DISCUSSION] Release Singa 2.0

[Wang Wei]

Hi all,

I want to start the discussion of releasing Singa 2.0.
Since V1.0, we have added many new major features, e.g.,
1. autograd
2. rafiki
3. mkl-dnn (to be merged)
4. onnx (to be merged)

I think it is time to release v2.0 (after merging the PRs for mkl-dnn and
onnx).
May I have your comments?
Thanks!

Regards,
Wei

Re: [jira] [Commented] (SINGA-415) Moving SINGA to Gitbox

[Wang Wei]

Shall we request to create a new git repo like incubator-singa-website for
the website?

Regards,
Wei

On Mon, Dec 10, 2018 at 3:37 AM Moaz Reyad  wrote:

> Thank you for the vote.
>
> I don't have access to request the website repository from
> selfserve.apache.org . It says "You MUST be on the PMC or PPMC of the
> project you wish to create a new repository for."
>
> We need someone from the PMC or PPMC to create the singa-site repository
> and populate it as described below.
>
> Best,
> Moaz
>
> On Mon, Dec 3, 2018 at 3:13 PM Wang Wei  wrote:
>
> > +1
> >
> > On Mon, Dec 3, 2018 at 6:19 PM Moaz Reyad (JIRA) 
> wrote:
> >
> > >
> > > [
> > >
> >
> https://issues.apache.org/jira/browse/SINGA-415?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16706942#comment-16706942
> > > ]
> > >
> > > Moaz Reyad commented on SINGA-415:
> > > --
> > >
> > > SINGA code repo is moved to Gitbox, but SINGA site is still in [SVN|[
> > > https://svn.apache.org/repos/asf/incubator/singa/site/]. I think it is
> > > better to move it also to git to have both repos in gitbox. To do this
> we
> > > need to request a web site repository first.
> > >
> > > From the INFRA ticket:
> > >
> > > "For the svn repo for your web site, you'll have to request a web site
> > > repository yourself via selfserve.apache.org and populate an asf-site
> > > branch with content before we can switch over to gitpubsub."
> > >
> > > > Moving SINGA to Gitbox
> > > > --
> > > >
> > > > Key: SINGA-415
> > > > URL: https://issues.apache.org/jira/browse/SINGA-415
> > > > Project: Singa
> > > >  Issue Type: Task
> > > >Reporter: Moaz Reyad
> > > >Priority: Minor
> > > >
> > > > This issue collects the internal work which is required in SINGA
> after
> > > [moving to GitBox|https://issues.apache.org/jira/browse/INFRA-17328].
> > > > Some modifications are required such as updating the [source
> > repository|
> > > http://singa.apache.org/en/community/source-repository.html] page in
> the
> > > website.
> > >
> > >
> > >
> > > --
> > > This message was sent by Atlassian JIRA
> > > (v7.6.3#76005)
> > >
> >
>

Re: [jira] [Commented] (SINGA-415) Moving SINGA to Gitbox

[Wang Wei]

+1

On Mon, Dec 3, 2018 at 6:19 PM Moaz Reyad (JIRA)  wrote:

>
> [
> https://issues.apache.org/jira/browse/SINGA-415?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16706942#comment-16706942
> ]
>
> Moaz Reyad commented on SINGA-415:
> --
>
> SINGA code repo is moved to Gitbox, but SINGA site is still in [SVN|[
> https://svn.apache.org/repos/asf/incubator/singa/site/]. I think it is
> better to move it also to git to have both repos in gitbox. To do this we
> need to request a web site repository first.
>
> From the INFRA ticket:
>
> "For the svn repo for your web site, you'll have to request a web site
> repository yourself via selfserve.apache.org and populate an asf-site
> branch with content before we can switch over to gitpubsub."
>
> > Moving SINGA to Gitbox
> > --
> >
> > Key: SINGA-415
> > URL: https://issues.apache.org/jira/browse/SINGA-415
> > Project: Singa
> >  Issue Type: Task
> >Reporter: Moaz Reyad
> >Priority: Minor
> >
> > This issue collects the internal work which is required in SINGA after
> [moving to GitBox|https://issues.apache.org/jira/browse/INFRA-17328].
> > Some modifications are required such as updating the [source repository|
> http://singa.apache.org/en/community/source-repository.html] page in the
> website.
>
>
>
> --
> This message was sent by Atlassian JIRA
> (v7.6.3#76005)
>

Re: [VOTE] Move SINGA to Gitbox

[Wang Wei]

Thanks. Sounds good!
+1

Regards,
Wei

On Thu, Nov 22, 2018 at 3:52 PM Moaz Reyad  wrote:

> Hi Wei,
>
> Currently SINGA code is using ASF git :
>
> https://git-wip-us.apache.org/repos/asf?p=incubator-singa.git
>
> There is a mirror of the github repository:
>
> https://github.com/apache/incubator-singa
>
> However, changes in SINGA master at github mirror are not automatically
> written in the ASF git master. ASF committers need to use their ASF account
> to push to
> https://git-wip-us.apache.org/repos/asf?p=incubator-singa.git.
>
> GitBox is a new service that provides Dual Master instead. The write
> feature is available to ASF committers only.
>
> ASF committers can link their ASF account with their GitHub account through
> the page(https://gitbox.apache.org/setup/). They are asked to authorize
> (login) ASF and Github accounts and link both accounts together.
>
> This will simplify the process of merging a pull request in SINGA:
> (http://singa.apache.org/en/community/source-repository.html#committers).
> Since both github and asf masters are mirrored in both directions, no need
> to clone from github and push to ASF when merging a pull request. Merging
> works from Github Merge button directly for ASF committers.
>
> SINGA can also ask for two new mail lists for commits and notifications, so
> these do not go to dev mailing list by default.
>
> Gitbox allows better integration with Jira and Travis CI. Jira can have
> "worklog" option which enables ASF GitHub Bot to write to the Work Log
> section under Activity in Jira automatically. Travis CI configuration is
> made easier with write permission on the git repo.
>
> For the SINGA website which is on SVN:
>
> https://svn.apache.org/repos/asf/incubator/singa/site/
>
> If it is moved to a git-based website with GitBox (hence to github also),
> this will allow SINGA website pages to have "Edit this page" links which
> direct to github to make pull requests for website updates.
>
> This is not a complete list of features, there are more things to discover.
> More than 100 ASF projects already moved to GitBox.
>
> Best Regards,
> Moaz
>
> On Thu, Nov 22, 2018 at 2:40 AM Wang Wei  wrote:
>
> > Hi Moaz,
> >
> > Could you explain a bit on the features of Gitbox (the advantages over
> > github)?
> > Thanks!
> >
> > Regards,
> > Wei
> >
> > On Thu, Nov 22, 2018 at 12:13 AM Moaz Reyad  wrote:
> >
> > > Gitbox (https://gitbox.apache.org/) is a new service from ASF that
> > allows
> > > committing code through GitHub or through ASF. Many ASF projects are
> > > already moved to Gitbox (See https://gitbox.apache.org/repos/asf)
> > >
> > > SINGA has two repositories that can be moved to GitBox:
> > >
> > > code:
> > > https://git-wip-us.apache.org/repos/asf?p=incubator-singa.git
> > >
> > > website:
> > > https://svn.apache.org/repos/asf/incubator/singa/site/
> > >
> > > This vote will be open until 28-Nov-2018 and passes if a majority of
> > > at least three +1 Apache SINGA PMC votes are cast.
> > >
> > > Once the vote passes, we can create a JIRA ticket on INFRA to move the
> > > repositories.
> > >
> > > Thank you,
> > > Moaz
> > >
> >
>

Re: [VOTE] Move SINGA to Gitbox

[Wang Wei]

Hi Moaz,

Could you explain a bit on the features of Gitbox (the advantages over
github)?
Thanks!

Regards,
Wei

On Thu, Nov 22, 2018 at 12:13 AM Moaz Reyad  wrote:

> Gitbox (https://gitbox.apache.org/) is a new service from ASF that allows
> committing code through GitHub or through ASF. Many ASF projects are
> already moved to Gitbox (See https://gitbox.apache.org/repos/asf)
>
> SINGA has two repositories that can be moved to GitBox:
>
> code:
> https://git-wip-us.apache.org/repos/asf?p=incubator-singa.git
>
> website:
> https://svn.apache.org/repos/asf/incubator/singa/site/
>
> This vote will be open until 28-Nov-2018 and passes if a majority of
> at least three +1 Apache SINGA PMC votes are cast.
>
> Once the vote passes, we can create a JIRA ticket on INFRA to move the
> repositories.
>
> Thank you,
> Moaz
>

Re: [VOTE] Release apache-singa-incubating-1.2.0 (RC1)

[Wang Wei]

+1,

Checked the licenses via Apache rat.

Regards,
Wei

On Mon, May 21, 2018 at 6:49 PM, Moaz Reyad  wrote:

> Hi all,
>
> We have created a build for Apache SINGA 1.2.0-incubating, release
> candidate RC1.
>
> The artifacts to be voted on are located here:
> https://dist.apache.org/repos/dist/dev/incubator/singa/1.2.0/
>
> The hashes of the artifacts are as follows:
> MD5:  2D 7E 67 84 6B AA B5 70  5D 77 52 C3 5E C6 4B 98
>
> Release artifacts are signed with the following key:
> https://people.apache.org/keys/committer/dinhtta.asc
>
> and the signature file is:
> https://dist.apache.org/repos/dist/dev/incubator/singa/1.2.
> 0/apache-singa-incubating-1.2.0-RC1.tar.gz.asc
>
> The Github tag is at:
> https://github.com/apache/incubator-singa/releases/tag/v1.2.0-rc1
> commit ID is: c343ff95019e05512c95f05588ae374684d9c5df
>
> To check the license, you can use the Apache Rat tool as follows:
> 1. download & decompress apache rat from
> http://creadur.apache.org/rat/download_rat.cgi
> 2. run the following command under singa folder:
> java -jar /PATH/TO/RAT/apache-rat-0.11.jar -E rat-excludes -d . >
> rat_check
> 3. check the results in file named "rat_check"
>
> Please check and vote on releasing this package. The vote is open for at
> least 72 hours and passes if a majority of at least three +1 votes are
> cast.
>
> [ ] +1 Release this package as Apache SINGA 1.2.0-incubating
> [ ]  0 I don't feel strongly about it, but I'm okay with the release
> [ ] -1 Do not release this package because...
>
> Here is my vote:
> +1
>
> Regards,
> Moaz
>

Re: [jira] [Commented] (SINGA-350) Error from python3 test

[Wang Wei]

It was tested with the master branch code from github.
Ubuntu 16.04 + CUDA9.0.

Can you pass the unit tests?

On Mon, Apr 30, 2018 at 1:43 PM, Moaz Reyad (JIRA)  wrote:

>
> [ https://issues.apache.org/jira/browse/SINGA-350?page=
> com.atlassian.jira.plugin.system.issuetabpanels:comment-
> tabpanel=16458326#comment-16458326 ]
>
> Moaz Reyad commented on SINGA-350:
> --
>
> After several tries, I couldn't reproduce this bug because the description
> is not complete.
>
> Please remember to include the precise steps to reproduce the bug and not
> only an error message. The steps include which version of the SINGA did you
> use, what are building options and what is the target platform.
>
> The Pull Request for SINGA-290 doesn't generate this bug as far as I know.
> It is probably related to changes made after SINGA-290 was done.
>
> > Error from python3 test
> > 
> >
> > Key: SINGA-350
> > URL: https://issues.apache.org/jira/browse/SINGA-350
> > Project: Singa
> >  Issue Type: Bug
> >Reporter: wangwei
> >Priority: Major
> >
> > If we compile Singa with Python 3, the unit tests report the following
> errors:
> >
> > {code:java}
> > + cd ../test/python
> > + '[' PYTHON3 = PYTHON3 ']'
> > + PYTHONPATH=../../build/python/
> > + python3 run.py
> > Running tests...
> > --
> > .E.EE.
> > ==
> > ERROR [0.000s]: test_activation (test_layer.TestPythonLayer)
> > --
> > Traceback (most recent call last):
> >   File "/var/jenkins/workspace/gpu-test-py3/label/cuda9.0-cudnn7.
> 1.2-py3/test/python/test_layer.py", line 187, in test_activation
> > act = layer.Activation('act', input_sample_shape=input_sample_shape)
> >   File "/var/jenkins/workspace/gpu-test-py3/label/cuda9.0-cudnn7.
> 1.2-py3/build/python/singa/layer.py", line 760, in __init__
> > self.layer = _create_layer(engine, mode)
> >   File "/var/jenkins/workspace/gpu-test-py3/label/cuda9.0-cudnn7.
> 1.2-py3/build/python/singa/layer.py", line 1188, in _create_layer
> > return singa_wrap.CreateLayer(layer_type.lower().encode())
> > TypeError: in method 'CreateLayer', argument 1 of type 'std::string
> const &'
> > ==
> > ERROR [0.000s]: test_avg_pooling1D (test_layer.TestPythonLayer)
> > --
> > Traceback (most recent call last):
> >   File "/var/jenkins/workspace/gpu-test-py3/label/cuda9.0-cudnn7.
> 1.2-py3/test/python/test_layer.py", line 158, in test_avg_pooling1D
> > input_sample_shape=in_sample_shape)
> >   File "/var/jenkins/workspace/gpu-test-py3/label/cuda9.0-cudnn7.
> 1.2-py3/build/python/singa/layer.py", line 532, in __init__
> > data_format, input_sample_shape)
> >   File "/var/jenkins/workspace/gpu-test-py3/label/cuda9.0-cudnn7.
> 1.2-py3/build/python/singa/layer.py", line 485, in __init__
> > pad, data_format, input_sample_shape)
> >   File "/var/jenkins/workspace/gpu-test-py3/label/cuda9.0-cudnn7.
> 1.2-py3/build/python/singa/layer.py", line 451, in __init__
> > self.layer = _create_layer(engine, 'Pooling')
> >   File "/var/jenkins/workspace/gpu-test-py3/label/cuda9.0-cudnn7.
> 1.2-py3/build/python/singa/layer.py", line 1188, in _create_layer
> > return singa_wrap.CreateLayer(layer_type.lower().encode())
> > TypeError: in method 'CreateLayer', argument 1 of type 'std::string
> const &'
> > {code}
> > It could be related with swig SINGA-290
>
>
>
> --
> This message was sent by Atlassian JIRA
> (v7.6.3#76005)
>

Re: [jira] [Commented] (SINGA-333) Add support for Open Neural Network Exchange (ONNX) format

[Wang Wei]

The config proto objects can be created by reading the config file like
Caffe or be created using the protobuf APIs (not widely used,
https://github.com/apache/incubator-singa/blob/master/
examples/cifar10/alexnet.cc#L37)

On Mon, Oct 23, 2017 at 1:08 PM, Wang Wei <wang...@comp.nus.edu.sg> wrote:

> The declarative config is mainly used for using SINGA C++ APIs.
> For Python APIs, it is more convenient to use the python layer APIs.
>
>
> On Sat, Oct 21, 2017 at 5:41 AM, Cliff Berg (JIRA) <j...@apache.org>
> wrote:
>
>>
>> [ https://issues.apache.org/jira/browse/SINGA-333?page=com.atl
>> assian.jira.plugin.system.issuetabpanels:comment-tabpanel&
>> focusedCommentId=16213282#comment-16213282 ]
>>
>> Cliff Berg commented on SINGA-333:
>> --
>>
>> Related to that, it appears that the declarative config that is used in
>> the Singa documentation (e.g., the layer definitions) do not appear in any
>> of the examples. Is this because the declarative approach has been shelved,
>> in favor of an API approach?
>>
>>
>> > Add support for Open Neural Network Exchange (ONNX) format
>> > --
>> >
>> > Key: SINGA-333
>> > URL: https://issues.apache.org/jira/browse/SINGA-333
>> > Project: Singa
>> >  Issue Type: New Feature
>> >Reporter: Moaz Reyad
>> >
>> > The [Open Neural Network Exchange (ONNX)|https://onnx.ai/] is a format
>> for interchanging neural network models between AI systems.
>> > The ONNX code is available here:
>> > https://github.com/onnx/onnx
>> > It can be a useful feature for SINGA to support this format. However,
>> ONNX is graph based and some extra work will be required to convert SINGA
>> models to this format. May be some core parts of SINGA must to be upgraded
>> first.
>> > This issue is open for studying and discussing the proposed feature of
>> supporting ONNX.
>>
>>
>>
>> --
>> This message was sent by Atlassian JIRA
>> (v6.4.14#64029)
>>
>
>

Re: [jira] [Commented] (SINGA-333) Add support for Open Neural Network Exchange (ONNX) format

[Wang Wei]

The declarative config is mainly used for using SINGA C++ APIs.
For Python APIs, it is more convenient to use the python layer APIs.


On Sat, Oct 21, 2017 at 5:41 AM, Cliff Berg (JIRA)  wrote:

>
> [ https://issues.apache.org/jira/browse/SINGA-333?page=
> com.atlassian.jira.plugin.system.issuetabpanels:comment-
> tabpanel=16213282#comment-16213282 ]
>
> Cliff Berg commented on SINGA-333:
> --
>
> Related to that, it appears that the declarative config that is used in
> the Singa documentation (e.g., the layer definitions) do not appear in any
> of the examples. Is this because the declarative approach has been shelved,
> in favor of an API approach?
>
>
> > Add support for Open Neural Network Exchange (ONNX) format
> > --
> >
> > Key: SINGA-333
> > URL: https://issues.apache.org/jira/browse/SINGA-333
> > Project: Singa
> >  Issue Type: New Feature
> >Reporter: Moaz Reyad
> >
> > The [Open Neural Network Exchange (ONNX)|https://onnx.ai/] is a format
> for interchanging neural network models between AI systems.
> > The ONNX code is available here:
> > https://github.com/onnx/onnx
> > It can be a useful feature for SINGA to support this format. However,
> ONNX is graph based and some extra work will be required to convert SINGA
> models to this format. May be some core parts of SINGA must to be upgraded
> first.
> > This issue is open for studying and discussing the proposed feature of
> supporting ONNX.
>
>
>
> --
> This message was sent by Atlassian JIRA
> (v6.4.14#64029)
>

Re: usage of SINGA

[Wang Wei]

Hi Zhongle,

Did you ever encountered such error?
How did you resolve it?
Thanks.

regards,
Wei

On Tue, Mar 7, 2017 at 6:40 PM, Lee Hwee Kuan <le...@bii.a-star.edu.sg>
wrote:

> hi
>
> thanks. i have a chance to try your suggestion.
>
> glog most probably is being used by other of my programs. i also may have
> multiple glog around, in /usr/local/lib as well as in /usr/local/Cellar
> (from Brew) and perhaps others.
>
> so i opt for option 2, link -lglog manually in my CMakeList.txt file.
> delete build directory, reinvoke "cmake ../“ and then make. howevery i got
> the same errors as before
> —
> 175 [100%] Linking CXX executable mytest
> 176 Undefined symbols for architecture x86_64:
> 177   "singa::logging::LogMessageFatal::LogMessageFatal(char const*,
> int)", referenced   from:
> 178   singa::Block::data() const in mytest.cc.o
> 179   "singa::logging::LogMessageFatal::~LogMessageFatal()", referenced
> from:
> 180   singa::Block::data() const in mytest.cc.o
> 181 ld: symbol(s) not found for architecture x86_64
> 182 clang: error: linker command failed with exit code 1 (use -v to see
> invocation)
> 183 make[2]: *** [mytest] Error 1
> 184 make[1]: *** [CMakeFiles/mytest.dir/all] Error 2
> 185 make: *** [all] Error 2
>
>
> hk
>
> On 7 Mar 2017, at 5:13 PM, Wang Wei <wang...@comp.nus.edu.sg> wrote:
>
> It should be caused by glog. If you have gloged instlled, libsinga.so
> would link to libglog.
> There are 2 solutions:
> 1. uninstall glog and re-compile libsinga.so
> 2. add -lglog into your link flags when generating mytest.
>
> regards,
> wang wei
>
> On Tue, Mar 7, 2017 at 5:10 PM, Lee Hwee Kuan <le...@bii.a-star.edu.sg>
> wrote:
>
>> Dear Wang Wei,
>>
>> i managed to assign values to Tensor object and compile. when i wish to
>> check if the values are correctly assigned, i found that i cannot link to
>> implementation of LogMessageFatal class. i try to dig at where is the
>> implementation and also change my CMakeLists.txt to manually link libraries
>> but unable to get it to work. line 15 of the mytest.cc leads to error,
>> try to call data() function and data() function calls block->data() which
>> leads to linking errors. may i know in which library you put
>> LogMessageFatal implementation at? I linked everything in
>> incubator-singa/build/lib directory. thanks again.
>>
>> below is cut-and-paste code segments of LogMessageFatal, compile errors,
>> mytest.cc and CMakeList.txt
>>
>> hk
>>
>>
>> /Users/hweekuan/incubator-singa/src/utils/logging.cc
>>
>> 112 LogMessage::~LogMessage() { GenerateLogMessage(); }
>> 113
>> 114 LogMessageFatal::LogMessageFatal(const char* file, int line)
>> 115   : LogMessage(file, line, FATAL) {}
>> 116 LogMessageFatal::~LogMessageFatal() {
>> 117   // abort() ensures we don't return
>> 118   GenerateLogMessage();
>> 119   abort();
>> 120 }
>>
>> 
>> hweekuans-MacBook-Pro:buid hweekuan$ make
>> [ 50%] *Linking CXX executable mytest*
>> Undefined symbols for architecture x86_64:
>>   "singa::logging::LogMessageFatal::LogMessageFatal(char const*, int)",
>> referenced from:
>>   singa::Block::data() const in mytest.cc.o
>>   "singa::logging::LogMessageFatal::~LogMessageFatal()", referenced from:
>>   singa::Block::data() const in mytest.cc.o
>> ld: symbol(s) not found for architecture x86_64
>> clang: error: linker command failed with exit code 1 (use -v to see
>> invocation)
>> make[2]: *** [mytest] Error 1
>> make[1]: *** [CMakeFiles/mytest.dir/all] Error 2
>> make: *** [all] Error 2
>>
>> 
>>   1 #include "singa/model/metric.h"
>>   2
>>   3 namespace singa {
>>   4
>>   5 void Train() {
>>   6
>>   7   const size_t train_size = 10;
>>   8   Tensor train_x(Shape{train_size,1});
>>   9   int tx[train_size];
>>  10
>>  11   for(unsigned int i=0;i<train_size;++i) { tx[i] = 1; }
>>  12   train_x.CopyDataFromHostPtr(tx,train_size);
>>  13
>>  14   // read back and see if this is correct
>>  15   const int * rx = train_x.data(); // ## error generate from
>> this line
>>  16 //  train_x.GetValue(rx,train_size);
>>  17 }
>>  18 } // namespace singa
>>  19
>>  20 int main(int argc, char **argv) { singa::Train(); }
>>
>> =
>>  19 CMAKE_MINIMUM_REQUIRED(VERSION 2.8)
>>  20
>>  21 SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLA

Re: gpu ami

[Wang Wei]

One more thing, even using docker, we still need to install the nvidia
driver on the host OS. nvidia-docker images only have cuda and cudnn.

btw, I think the AMI is similar to docker images to end users.
If users want a new clean environment, he can start a new EC2 instance
using the AMI directly..

regards,
wang wei

On Thu, Feb 16, 2017 at 1:11 PM, Moaz Reyad <m...@comp.nus.edu.sg> wrote:

> docker comes automatically in the AMI :
>
> http://docs.aws.amazon.com/AmazonECS/latest/developerguide/
> ecs-optimized_AMI.html
>
> There is a trick to make nvidia docker works:
>
> https://blog.cloudsight.ai/deep-learning-image-recognition-
> using-gpus-in-amazon-ecs-docker-containers-5bdb1956f30e#.2l35ojv1p
>
> I assume Amazon EC2 will hide the docker stuff form the user. users may be
> able to see the docker container as any normal Amazon EC2 machine.
> Moaz
>
>
> On 16/02/2017 13:05, Wang Wei wrote:
>
> docker is very useful. but in this way, we need to install docker and
> nvida-docker, and assume users know how to use docker.
>
> On Thu, Feb 16, 2017 at 1:04 PM, Moaz Reyad <m...@comp.nus.edu.sg> wrote:
>
>> Ok. Let me think about the best way to do it.
>>
>> I was thinking to use the nvidia docker image that comes already with
>> cuda/cudnn, then we install singa-cuda.deb on it and push it to the
>> marketplace. Let me see if this can work.
>>
>> Moaz
>> On 16/02/2017 12:55, Wang Wei wrote:
>>
>> I think we can. Amazon is dong it in this way as mentioned in the video
>> meeting.
>>
>> On Thu, Feb 16, 2017 at 12:41 PM, Moaz Reyad <m...@comp.nus.edu.sg>
>> wrote:
>>
>>> Thanks.
>>>
>>> We can create GPU AMI, but I am not sure we can publish it in the Amazon
>>> marketplace.
>>>
>>> Moaz
>>>
>>>
>>>
>>
>
>

[ANNOUNCE] Apache SINGA (incubating) 1.1.0 release

[Wang Wei]

The Apache SINGA (incubating) team is pleased to announce the release of
SINGA 1.1.0.

SINGA is a general distributed deep learning platform for training big deep
learning models over large datasets.

The release is available at:
http://singa.apache.org/en/downloads.html

Features implemented in this release include,

   - Create Docker images (CPU and GPU versions)
   - Create Amazon AMI for SINGA (CPU version)
   - Integrate with Jenkins for automatically generating Wheel and Debian
   packages (for installation), and updating the website.
   - Enhance the FeedFowardNet, e.g., multiple inputs and verbose mode for
   debugging
   - Add Concat and Slice layers
   - Extend CrossEntropyLoss to accept instance with multiple labels
   - Add image_tool.py with image augmentation methods
   - Support model loading and saving via the Snapshot API
   - Compile SINGA source on Windows
   - Compile mandatory dependent libraries together with SINGA code
   - Enable Java binding (basic) for SINGA
   - Add version ID in checkpointing files
   - Add Rafiki toolkit for providing RESTFul APIs
   - Add examples pretrained from Caffe, including GoogleNet


See the release notes for more details:
http://singa.apache.org/en/releases/RELEASE_NOTES_1.1.0.html

We look forward to hearing your feedbacks, suggestions, and contributions
to the project (http://singa.apache.org/).

Regards,
The Apache SINGA (incubating) team

=

*Disclaimer*

Apache SINGA is an effort undergoing incubation at The Apache Software
Foundation (ASF), sponsored by the name of Apache Incubator PMC. Incubation
is required of all newly accepted projects until a further review indicates
that the infrastructure, communications, and decision making process have
stabilized in a manner consistent with other successful ASF projects. While
incubation status is not necessarily a reflection of the completeness or
stability of the code, it does indicate that the project has yet to be
fully endorsed by the ASF.

Re: [VOTE] Release apache-singa-incubating-1.1.0 (RC1)

[Wang Wei]

Hi John,

Thank you for checking the license of Glog.
The document building instructions are here
https://github.com/apache/incubator-singa/blob/master/doc/README.md
Glog is an optional dependency, which is declared in the installation page
http://singa.apache.org/en/docs/installation.html#from-source.
We would make it more clear in the LICENSE file and remove it from the
travis.yml file.
Can we pass this release considering Glog is an optional dependency?

Regards,
Wei


On Mon, Feb 6, 2017 at 10:49 AM, John D. Ament <johndam...@apache.org>
wrote:

> We actually just had a discussion recently on legal-discuss on this type of
> topic.  Specifically, Cat-X and optional vs required dependencies.  Henri
> and I settled on the wording you'll find at [1] as the final result.
> Basically, you can rely on Cat-X but only for optional features.
>
> We can probably follow up with legal on whether this does fall into the GPL
> bucket though.
>
> John
>
> [1]: https://www.apache.org/legal/resolved.html#optional
>
> On Sun, Feb 5, 2017 at 9:45 PM Niclas Hedhman <nic...@hedhman.org> wrote:
>
> > I think that ends up being a build time dependency in GLOG, i.e. the
> > equivalent of Systems Requirement, and not in itself viral to the ASF
> > software. I assume that Google is much more worried about this and may
> even
> > have checked with their Legal team...
> >
> > Want to check with legal-discuss@ ? Is my memory failing me, or hasn't
> FSF
> > stated that the build output is not bound by GPL of the build toolchain?
> > (otherwise they can't release their own LGPL stuff)
> >
> > Cheers
> > Niclas
> >
> > On Mon, Feb 6, 2017 at 10:27 AM, John D. Ament <johndam...@apache.org>
> > wrote:
> >
> > > -1 at least I think there's an issue.
> > >
> > > While the source code all looks good, the resulting binary is not
> valid.
> > > There's no how to build doc, so I looked at your .travis.yml.  It
> > confirmed
> > > what I suspected for make, but then I started looking at your required
> > > packages.
> > >
> > > You require glog [1], which I can't find a license for.  However, glog
> > > includes [2] which is GPL, which makes glog GPL and as a result, your
> > code.
> > >
> > > [1]: https://github.com/google/glog
> > > [2]: https://github.com/google/glog/blob/master/missing
> > >
> > > - John
> > >
> > > On Sun, Jan 29, 2017 at 8:24 PM Wang Wei <wang...@apache.org> wrote:
> > >
> > > > Hi all,
> > > >
> > > > The SINGA community has voted on and approved a proposal to release
> > > Apache
> > > > SINGA 1.1.0 (incubating).
> > > >
> > > > The vote thread is at:
> > > > http://mail-archives.apache.org/mod_mbox/singa-dev/201701.mbox/%
> > > > 3CCAJz0iLvAaEd5AgCqaWFD1M4-D_3EBh%3DUHbeZu-MRn%3DcKuQiX-Q%4
> > > 0mail.gmail.com
> > > > %3E
> > > >
> > > > and the result is at:
> > > >
> > > > http://mail-archives.apache.org/mod_mbox/singa-dev/201701.mbox/%
> > > 3ccajz0ilspborscstawuraowwetidb4cn14ak5sgdyb8umxoj...@mail.gmail.com
> %3e
> > > >
> > > > We ask the IPMC to vote on this release.
> > > >
> > > > The artifacts to be voted on are located here:
> > > > https://dist.apache.org/repos/dist/dev/incubator/singa/1.1.0/
> > > >
> > > > The hashes of the artifacts are as follows:
> > > > MD5:  08 CA 31 10 4E 79 02 16  A1 D7 3F 20 2D 60 21 BB
> > > >
> > > > Release artifacts are signed with the following key:
> > > > https://people.apache.org/keys/committer/dinhtta.asc
> > > >
> > > > and the signature file is:
> > > > https://dist.apache.org/repos/dist/dev/incubator/singa/1.1.0
> > > > /apache-singa-incubating-1.1.0-RC1.tar.gz.asc
> > > >
> > > > The Github tag is at:
> > > > https://github.com/apache/incubator-singa/releases/tag/v1.1.0-rc1
> > > > commit ID is: 59ca44a7ce38ce4f965511c805cda074d0b8e360
> > > >
> > > > To check the license, you can use the Apache Rat tool as follows:
> > > > 1. download & decompress apache rat from
> > http://creadur.apache.org/rat/
> > > > download_rat.cgi
> > > > 2. run the following command under singa folder:
> > > > java -jar /PATH/TO/RAT/apache-rat-0.11.jar -E rat-excludes -d .
> >
> > > > rat_check
> > > > 3. check the results in file named "rat_check"
> > > >
> > > > Please check and vote on releasing this package. The vote is open for
> > at
> > > > least 72 hours and passes if a majority of at least three +1 votes
> are
> > > > cast.
> > > >
> > > > [ ] +1 Release this package as Apache SINGA 1.0.0-incubating
> > > > [ ]  0 I don't feel strongly about it, but I'm okay with the release
> > > > [ ] -1 Do not release this package because...
> > > >
> > > > Best,
> > > > Wei
> > > >
> > >
> >
> >
> >
> > --
> > Niclas Hedhman, Software Developer
> > http://polygene.apache.org <http://zest.apache.org> - New Energy for
> Java
> >
>

[VOTE] Release apache-singa-incubating-1.1.0 (RC1)

[Wang Wei]

Hi all,

The SINGA community has voted on and approved a proposal to release Apache
SINGA 1.1.0 (incubating).

The vote thread is at:
http://mail-archives.apache.org/mod_mbox/singa-dev/201701.mbox/%
3CCAJz0iLvAaEd5AgCqaWFD1M4-D_3EBh%3DUHbeZu-MRn%3DcKuQiX-Q%40mail.gmail.com
%3E

and the result is at:
http://mail-archives.apache.org/mod_mbox/singa-dev/201701.mbox/%3ccajz0ilspborscstawuraowwetidb4cn14ak5sgdyb8umxoj...@mail.gmail.com%3e

We ask the IPMC to vote on this release.

The artifacts to be voted on are located here:
https://dist.apache.org/repos/dist/dev/incubator/singa/1.1.0/

The hashes of the artifacts are as follows:
MD5:  08 CA 31 10 4E 79 02 16  A1 D7 3F 20 2D 60 21 BB

Release artifacts are signed with the following key:
https://people.apache.org/keys/committer/dinhtta.asc

and the signature file is:
https://dist.apache.org/repos/dist/dev/incubator/singa/1.1.0
/apache-singa-incubating-1.1.0-RC1.tar.gz.asc

The Github tag is at:
https://github.com/apache/incubator-singa/releases/tag/v1.1.0-rc1
commit ID is: 59ca44a7ce38ce4f965511c805cda074d0b8e360

To check the license, you can use the Apache Rat tool as follows:
1. download & decompress apache rat from http://creadur.apache.org/rat/
download_rat.cgi
2. run the following command under singa folder:
java -jar /PATH/TO/RAT/apache-rat-0.11.jar -E rat-excludes -d . >
rat_check
3. check the results in file named "rat_check"

Please check and vote on releasing this package. The vote is open for at
least 72 hours and passes if a majority of at least three +1 votes are cast.

[ ] +1 Release this package as Apache SINGA 1.0.0-incubating
[ ]  0 I don't feel strongly about it, but I'm okay with the release
[ ] -1 Do not release this package because...

Best,
Wei

[Docker] Can we push docker images to https://hub.docker.com/u/apache/

[Wang Wei]

Hi all,

I am from the incubator-singa project.
We want to host our docker images under https://hub.docker.com/u/apache/.
https://issues.apache.org/jira/browse/INFRA-13156

Currently, there is no incubator project hosting docker images under the
Apache namespace.
I start this thread to get binding vote for this request.

Thank you!

Best,
Wei

Re: Apache Singa

[Wang Wei]

Hi Ram,

Welcome to use and contribute to SINGA.
Here is the installation doc:
http://singa.apache.org/en/docs/installation.html
The system software and python APIs are also available online at
http://singa.apache.org/en/docs/index.html
If you have any problems, feel free to contact us.

Best,
Wei

On Tue, Oct 18, 2016 at 10:59 PM, Sinthalapadi Ram Janarthana Raja <
e0012...@u.nus.edu> wrote:

>
> My name is Ram Janarthan, and I am an undergraduate student at NUS,
> pursuing a degree in Computer Science. I wanted to get in touch with you
> regarding your SINGA (Distributed Deep Learning) project, and the
> specifications on how to use it. As the website does not have any active
> links on how to install the system, I was hoping you could help me out with
> this, as well as help me better understand the working process of the
> system as whole (I have a hazy idea, after looking at your past
> presentation slides) . I'm looking forward to your response!
>
>
> Regards,
>
> Ram
>

Re: activity recognition using apache singa

[Wang Wei]

Hi Arash,

Make sure all tensors are copied to the same cuda device to avoid memory
error.
Please reinstall singa using the latest wheel file

http://comp.nus.edu.sg/~dbsystem/singa/assets/file/pb2.6-cuda7.5-cudnn5/singa-1.0.0-cp27-none-linux_x86_64.whl

which resolved the problem of int32<->float32.

Thanks.


On Sun, Oct 9, 2016 at 4:14 PM, Wang Wei <wangwei...@gmail.com> wrote:

> The label tensor should be of shape (2947,) like this one
> https://github.com/apache/incubator-singa/blob/master/
> python/singa/metric.py#L31
> I will check the problem from int32 and float32.
>
> On Sun, Oct 9, 2016 at 3:27 PM Arash Shafiei <arash.shaf...@gmail.com>
> wrote:
>
>> I am facing a problem concerning creating tensor from numpy:
>>
>> Suppose that I have an array of these values:
>> b = np.asarray([ [1., 0., 0., 0., 0., 0.], [0., 1., 0., 0., 0., 0.]])
>>
>> I transform it to a tensor:
>> t = tensor.from_numpy(b.astype(np.int32))
>>
>> Now I again take the numpy back:
>> a = tensor.to_numpy(t)
>>
>> The values of 'a' change and I lose accuracy:
>> >>> a
>> array([[  1.40129846e-45,   0.e+00,   0.e+00,
>>   0.e+00,   0.e+00,   0.e+00],
>>[  0.e+00,   1.40129846e-45,   0.e+00,
>>   0.e+00,   0.e+00,   0.e+00]],
>> dtype=float32)
>>
>> Now suppose that I create a tensor of float32:
>> t = tensor.from_numpy(b.astype(np.float32))
>>
>> Now I again take the numpy back:
>> a = tensor.to_numpy(t)
>>
>> Now the values of 'a' are correct:
>> array([[ 1.,  0.,  0.,  0.,  0.,  0.],
>>[ 0.,  1.,  0.,  0.,  0.,  0.]], dtype=float32)
>>
>> But in the second case where I am having float32, while calculating the
>> L2 norm:
>> lvalue = lossfun.forward(model_pb2.kTrain, act, labels)
>> batch_loss += lvalue.l2()
>>
>> The error message is:
>> [F d1009 t15:26:37 p29915:584 /home/wuwf/work/incubator-
>> singa/src/core/device/cuda_gpu.cc:112] Check failed: error ==
>> cudaSuccess (77 vs. 0) an illegal memory access was encountered
>> Aborted (core dumped)
>>
>> Thanks for the support.
>>
>>
>> On Sun, Oct 9, 2016 at 2:09 PM, Arash Shafiei <arash.shaf...@gmail.com>
>> wrote:
>>
>> Thanks much.
>>
>> I am trying to use Evaluate() function. I pass activations of the dense
>> layers and all the labels:
>>
>> evaluator.Evaluate(act,labels)
>>
>> The dimensions seem to be correct:
>>
>> act.shape == labels.shape == (2947, 6)
>>
>> But I am getting the following error:
>> TypeError: in method 'Metric_Evaluate', argument 2 of type 'singa::Tensor
>> const &'
>>
>> On Sun, Oct 9, 2016 at 1:17 PM, Wang Wei <wangwei...@gmail.com> wrote:
>>
>> No. Loss != Inaccuracy.
>> If you want to compute the accuracy, you need to create an
>> evaluator=singa.Accuracy(), and call evaluator.Evaluate(o, t), where o is
>> the output from the dense layer and t is the ground truth tensor. You can
>> follow the example here https://github.com/apache/
>> incubator-singa/blob/master/python/singa/metric.py#L67.
>>
>> Good Luck!
>>
>> On Sun, Oct 9, 2016 at 12:08 PM Arash Shafiei <arash.shaf...@gmail.com>
>> wrote:
>>
>> Thanks for the hint.
>>
>> I was sending it to the device but the problem turned out to be that I
>> did not cast labels to int32.
>>
>> Now it is working and I am getting:
>>
>> [... ] 96.4% training loss = 0.003444
>> Epoch 49, train loss is 0.003509
>> Epoch 49, evaluation loss is 0.003534
>>
>> Does this mean that after 50 epoch the evaluation has only 3.5%
>> inaccuracy?
>>
>> On Sun, Oct 9, 2016 at 11:29 AM, Wei Wang <wangwei...@gmail.com> wrote:
>>
>> Have you moved all tensor onto the same devices? Including the tensor for
>> the labels.
>>
>>
>> On 9 Oct 2016, at 11:02 AM, Arash Shafiei <arash.shaf...@gmail.com>
>> wrote:
>>
>> outputs = rnn.forward(model_pb2.kTrain, inputs)[0:-2]
>> grads = []
>> batch_loss = 0
>> g_dense_w.set_value(0.0)
>> g_dense_b.set_value(0.0)
>> print 'outputs len', len(outputs) // 128
>> output = outputs[-1]
>> act = dense.forward(model_pb2.kTrain, output)
>> print 'output shape', output.shape // (256, 28)
>> print 'activation shape', act.shape // (256, 6)
>> print 'labels shape', labels.shape // (256, 6)
>> lvalue = lossfun.forward(model_pb2.kTrain, act, labe

Re: activity recognition using apache singa

[Wang Wei]

The label tensor should be of shape (2947,) like this one
https://github.com/apache/incubator-singa/blob/master/python/singa/metric.py#L31
I will check the problem from int32 and float32.

On Sun, Oct 9, 2016 at 3:27 PM Arash Shafiei <arash.shaf...@gmail.com>
wrote:

> I am facing a problem concerning creating tensor from numpy:
>
> Suppose that I have an array of these values:
> b = np.asarray([ [1., 0., 0., 0., 0., 0.], [0., 1., 0., 0., 0., 0.]])
>
> I transform it to a tensor:
> t = tensor.from_numpy(b.astype(np.int32))
>
> Now I again take the numpy back:
> a = tensor.to_numpy(t)
>
> The values of 'a' change and I lose accuracy:
> >>> a
> array([[  1.40129846e-45,   0.e+00,   0.e+00,
>   0.e+00,   0.e+00,   0.e+00],
>[  0.e+00,   1.40129846e-45,   0.e+00,
>   0.e+00,   0.e+00,   0.e+00]],
> dtype=float32)
>
> Now suppose that I create a tensor of float32:
> t = tensor.from_numpy(b.astype(np.float32))
>
> Now I again take the numpy back:
> a = tensor.to_numpy(t)
>
> Now the values of 'a' are correct:
> array([[ 1.,  0.,  0.,  0.,  0.,  0.],
>[ 0.,  1.,  0.,  0.,  0.,  0.]], dtype=float32)
>
> But in the second case where I am having float32, while calculating the L2
> norm:
> lvalue = lossfun.forward(model_pb2.kTrain, act, labels)
> batch_loss += lvalue.l2()
>
> The error message is:
> [F d1009 t15:26:37 p29915:584
> /home/wuwf/work/incubator-singa/src/core/device/cuda_gpu.cc:112] Check
> failed: error == cudaSuccess (77 vs. 0) an illegal memory access was
> encountered
> Aborted (core dumped)
>
> Thanks for the support.
>
>
> On Sun, Oct 9, 2016 at 2:09 PM, Arash Shafiei <arash.shaf...@gmail.com>
> wrote:
>
> Thanks much.
>
> I am trying to use Evaluate() function. I pass activations of the dense
> layers and all the labels:
>
> evaluator.Evaluate(act,labels)
>
> The dimensions seem to be correct:
>
> act.shape == labels.shape == (2947, 6)
>
> But I am getting the following error:
> TypeError: in method 'Metric_Evaluate', argument 2 of type 'singa::Tensor
> const &'
>
> On Sun, Oct 9, 2016 at 1:17 PM, Wang Wei <wangwei...@gmail.com> wrote:
>
> No. Loss != Inaccuracy.
> If you want to compute the accuracy, you need to create an
> evaluator=singa.Accuracy(), and call evaluator.Evaluate(o, t), where o is
> the output from the dense layer and t is the ground truth tensor. You can
> follow the example here
> https://github.com/apache/incubator-singa/blob/master/python/singa/metric.py#L67
> .
>
> Good Luck!
>
> On Sun, Oct 9, 2016 at 12:08 PM Arash Shafiei <arash.shaf...@gmail.com>
> wrote:
>
> Thanks for the hint.
>
> I was sending it to the device but the problem turned out to be that I did
> not cast labels to int32.
>
> Now it is working and I am getting:
>
> [... ] 96.4% training loss = 0.003444
> Epoch 49, train loss is 0.003509
> Epoch 49, evaluation loss is 0.003534
>
> Does this mean that after 50 epoch the evaluation has only 3.5% inaccuracy?
>
> On Sun, Oct 9, 2016 at 11:29 AM, Wei Wang <wangwei...@gmail.com> wrote:
>
> Have you moved all tensor onto the same devices? Including the tensor for
> the labels.
>
>
> On 9 Oct 2016, at 11:02 AM, Arash Shafiei <arash.shaf...@gmail.com> wrote:
>
> outputs = rnn.forward(model_pb2.kTrain, inputs)[0:-2]
> grads = []
> batch_loss = 0
> g_dense_w.set_value(0.0)
> g_dense_b.set_value(0.0)
> print 'outputs len', len(outputs) // 128
> output = outputs[-1]
> act = dense.forward(model_pb2.kTrain, output)
> print 'output shape', output.shape // (256, 28)
> print 'activation shape', act.shape // (256, 6)
> print 'labels shape', labels.shape // (256, 6)
> lvalue = lossfun.forward(model_pb2.kTrain, act, labels)
> batch_loss += lvalue.l1() // [F d1009 t11:00:24 p23551:016
> /home/wuwf/work/incubator-singa/src/core/tensor/./tensor_math_cuda.h:344]
> Check failed: status == CUBLAS_STATUS_SUCCESS (11 vs. 0)
> CUBLAS_STATUS_MAPPING_ERROR
> Aborted (core dumped)
>
>
>
>
> On Sun, Oct 9, 2016 at 10:55 AM, Wei Wang <wangwei...@gmail.com> wrote:
>
> Could you please paste the relevant code leading to this error?
>
>
>
> On 9 Oct 2016, at 10:32 AM, Arash Shafiei <arash.shaf...@gmail.com> wrote:
>
> Thanks, it worked.
>
> So far, I managed to do rnn::forward(...) but now I am stuck somewhere
> else.
>
> rnn::forward(...) returns a tensor (denoted as lvalue). I have to obtain
> the L1 norm using lvalue.l1().
>
> But I get this error:
> [F d1009 t10:30:14 p23056:-56
> /home/wuwf/work

Re: activity recognition using apache singa

[Wang Wei]

No. Loss != Inaccuracy.
If you want to compute the accuracy, you need to create an
evaluator=singa.Accuracy(), and call evaluator.Evaluate(o, t), where o is
the output from the dense layer and t is the ground truth tensor. You can
follow the example here
https://github.com/apache/incubator-singa/blob/master/python/singa/metric.py#L67
.

Good Luck!

On Sun, Oct 9, 2016 at 12:08 PM Arash Shafiei <arash.shaf...@gmail.com>
wrote:

> Thanks for the hint.
>
> I was sending it to the device but the problem turned out to be that I did
> not cast labels to int32.
>
> Now it is working and I am getting:
>
> [... ] 96.4% training loss = 0.003444
> Epoch 49, train loss is 0.003509
> Epoch 49, evaluation loss is 0.003534
>
> Does this mean that after 50 epoch the evaluation has only 3.5% inaccuracy?
>
> On Sun, Oct 9, 2016 at 11:29 AM, Wei Wang <wangwei...@gmail.com> wrote:
>
> Have you moved all tensor onto the same devices? Including the tensor for
> the labels.
>
>
> On 9 Oct 2016, at 11:02 AM, Arash Shafiei <arash.shaf...@gmail.com> wrote:
>
> outputs = rnn.forward(model_pb2.kTrain, inputs)[0:-2]
> grads = []
> batch_loss = 0
> g_dense_w.set_value(0.0)
> g_dense_b.set_value(0.0)
> print 'outputs len', len(outputs) // 128
> output = outputs[-1]
> act = dense.forward(model_pb2.kTrain, output)
> print 'output shape', output.shape // (256, 28)
> print 'activation shape', act.shape // (256, 6)
> print 'labels shape', labels.shape // (256, 6)
> lvalue = lossfun.forward(model_pb2.kTrain, act, labels)
> batch_loss += lvalue.l1() // [F d1009 t11:00:24 p23551:016
> /home/wuwf/work/incubator-singa/src/core/tensor/./tensor_math_cuda.h:344]
> Check failed: status == CUBLAS_STATUS_SUCCESS (11 vs. 0)
> CUBLAS_STATUS_MAPPING_ERROR
> Aborted (core dumped)
>
>
>
>
> On Sun, Oct 9, 2016 at 10:55 AM, Wei Wang <wangwei...@gmail.com> wrote:
>
> Could you please paste the relevant code leading to this error?
>
>
>
> On 9 Oct 2016, at 10:32 AM, Arash Shafiei <arash.shaf...@gmail.com> wrote:
>
> Thanks, it worked.
>
> So far, I managed to do rnn::forward(...) but now I am stuck somewhere
> else.
>
> rnn::forward(...) returns a tensor (denoted as lvalue). I have to obtain
> the L1 norm using lvalue.l1().
>
> But I get this error:
> [F d1009 t10:30:14 p23056:-56
> /home/wuwf/work/incubator-singa/src/core/tensor/./tensor_math_cuda.h:344]
> Check failed: status == CUBLAS_STATUS_SUCCESS (11 vs. 0)
> CUBLAS_STATUS_MAPPING_ERROR
> Aborted (core dumped)
>
> On Sat, Oct 8, 2016 at 9:43 PM, Wang Wei <wang...@comp.nus.edu.sg> wrote:
>
> Actually, the char-rnn example is from type (4), where each rnn unit would
> generate a prediction and has a ground truth label.
>
> For your model (type 2), you only need to use the y128 (of shape 256, 28)
> from the rnn::forward() as the input to the dense layer. All other yi
> should be ignored.
> Consequently, you would have an output (denoted as o) of shape (256, 6)
> from the dense layer, which is the prediction for the whole sequence (of
> length 128).
> By feeding the prediction o and the label into the loss layer, you can
> compute the loss value and compute the gradient for o (denoted as o').
> Backward propagating the o through the dense layer, you would get the
> gradient for y128, denoted as y'128.
>
> *The input of the rnn::backward() would be <y'1, y'2, ...y'128, hy', cy'>,
> where only y'128 is a valid tensor. y'1, y'2 ... should be tensor with
> value 0.*
>
> Best,
> Wei
>
>
> On Sat, Oct 8, 2016 at 9:33 PM Arash Shafiei <arash.shaf...@gmail.com>
> wrote:
>
> Thanks. It worked.
>
> I am now at the phase of evaluating the loss.
>
> singa.loss.SoftmaxCrossEntropy has a forward function where it takes
> prediction tensors and ground truth.
>
> My problem now is that the prediction is a sequence and my label is not a
> sequence.
>
> Your char-rnn example is an application of type (1) in the figure bellow,
> but activity recognition is an application of type (2).
>
>
> 
> Therefore for each sequence in a batch I have only 1 label. (although this
> label can be of one dimension from the set of {1,2,3,4,5,6} or of 6
> dimension from the set of { [1,0,0,0,0,0], [0,1,0,0,0,0] , etc. }
>
> So now I need predictions and ground truth. The prediction for me is of
> shape
> (128, 256, 28)
> where 128 is the length of the sequence, 256 is the batch size and 28 is
> the hidden layer size.
>
> And my ground truth is of shape
> (256, 1) or (256, 6) -- depending on how you model it..
>
> But as I understood from the example of char-rnn my ground truth must be
> of shape:
> (128, 256)
>
> 

Re: activity recognition using apache singa

[Wang Wei]

Actually, the char-rnn example is from type (4), where each rnn unit would
generate a prediction and has a ground truth label.

For your model (type 2), you only need to use the y128 (of shape 256, 28)
from the rnn::forward() as the input to the dense layer. All other yi
should be ignored.
Consequently, you would have an output (denoted as o) of shape (256, 6)
from the dense layer, which is the prediction for the whole sequence (of
length 128).
By feeding the prediction o and the label into the loss layer, you can
compute the loss value and compute the gradient for o (denoted as o').
Backward propagating the o through the dense layer, you would get the
gradient for y128, denoted as y'128.

*The input of the rnn::backward() would be <y'1, y'2, ...y'128, hy', cy'>,
where only y'128 is a valid tensor. y'1, y'2 ... should be tensor with
value 0.*

Best,
Wei


On Sat, Oct 8, 2016 at 9:33 PM Arash Shafiei <arash.shaf...@gmail.com>
wrote:

> Thanks. It worked.
>
> I am now at the phase of evaluating the loss.
>
> singa.loss.SoftmaxCrossEntropy has a forward function where it takes
> prediction tensors and ground truth.
>
> My problem now is that the prediction is a sequence and my label is not a
> sequence.
>
> Your char-rnn example is an application of type (1) in the figure bellow,
> but activity recognition is an application of type (2).
>
>
> [image: Inline image 1]
> Therefore for each sequence in a batch I have only 1 label. (although this
> label can be of one dimension from the set of {1,2,3,4,5,6} or of 6
> dimension from the set of { [1,0,0,0,0,0], [0,1,0,0,0,0] , etc. }
>
> So now I need predictions and ground truth. The prediction for me is of
> shape
> (128, 256, 28)
> where 128 is the length of the sequence, 256 is the batch size and 28 is
> the hidden layer size.
>
> And my ground truth is of shape
> (256, 1) or (256, 6) -- depending on how you model it..
>
> But as I understood from the example of char-rnn my ground truth must be
> of shape:
> (128, 256)
>
> Would you have any insight about this?
> Thanks..
>
>
> On Sat, Oct 8, 2016 at 6:42 PM, Wang Wei <wang...@comp.nus.edu.sg> wrote:
>
> Currently, numpy array of dtype=np.float32 or np.int could be converted
> into singa tensor.
> Please convert the numpy array into np.float32 and then call
> tensor.from_numpy(t) (without dtype=np.float32).
>
> On Sat, Oct 8, 2016 at 6:36 PM Arash Shafiei <arash.shaf...@gmail.com>
> wrote:
>
> The values that I have are floating points [-1 1].
>
> While using tensor.from_numpy(...), I was getting this error:
>
> Not implemented yet for  float64
>
> I understood from the tutorial that we could pass the data type:
>
> y = tensor.from_numpy(..., dtype=np.float32)
>
> But using dtype, I am getting another error:
>
> TypeError: from_numpy() got an unexpected keyword argument 'dtype'
>
>
>
> On Sat, Oct 8, 2016 at 3:45 PM, Wang Wei <wang...@comp.nus.edu.sg> wrote:
>
> Hi
>
> According to the API of forward function:
> http://singa.apache.org/en/docs/layer.html#singa.layer.RNN.forward
> The input should be a vector of Tensors, <x1, x2, ... x128, hx, cx>, xi is
> of shape (1500, 9), hx and cx are optional whose shape should be (1500, 28).
> The output would be a vector of Tensors, <y1, y2, ..., y128, hy, cy>, yi
> is of shape (1500, 28), hy and cy are optional depending on the existence
> of hx and cx.
> If you want to put the dense layer on top of the last rnn unit (i.e. the
> 128-th), then you feed y128 to the dense layer.
>
> function convert just reshapes the raw data into a sequence of tensors
> <x1, x2, ..>.
>
> BTW, typically, people would use a smaller batchsize e.g. less than 256.
>
> May I forward our discussion to the incubator email list in case others
> have similar problems?
> Thanks.
>
> Best,
> Wei
>
> So here what I have:
>
> input batch of dimension (1500, 128, 9)
> This means a batch of 1500 windows each having 128 vector of 9 dimensions.
>
> input label of dimension (1500, 6)
> This means a label batch of 1500 vector of 6 dimensions. This is to label
> if the person is sitting ([1,0,0,0,0,0]) or standing ([0,1,0,0,0,0]), etc.
>
> I am creating an lstm layer with hidden_size=28 and
> input_sample_shape=(9,) and num_stacks=1
>
> Then I create a dense layer with num_output=6 and input_sample_shape=(28,)
>
> Now I would like to feed the data to the 'forward' function of lstm and
> dense layer. But I could not make it work and I could not quit understand
> from the example what 'convert' and 'numpy2tensors' are suppose to do...
>
> I would appreciate your comments..
>
> On Sun, Sep 25, 2016 at 12:23 PM, Arash Shafiei <arash.shaf

Re: activity recognition using apache singa

[Wang Wei]

Currently, numpy array of dtype=np.float32 or np.int could be converted
into singa tensor.
Please convert the numpy array into np.float32 and then call
tensor.from_numpy(t) (without dtype=np.float32).

On Sat, Oct 8, 2016 at 6:36 PM Arash Shafiei <arash.shaf...@gmail.com>
wrote:

> The values that I have are floating points [-1 1].
>
> While using tensor.from_numpy(...), I was getting this error:
>
> Not implemented yet for  float64
>
> I understood from the tutorial that we could pass the data type:
>
> y = tensor.from_numpy(..., dtype=np.float32)
>
> But using dtype, I am getting another error:
>
> TypeError: from_numpy() got an unexpected keyword argument 'dtype'
>
>
>
> On Sat, Oct 8, 2016 at 3:45 PM, Wang Wei <wang...@comp.nus.edu.sg> wrote:
>
> Hi
>
> According to the API of forward function:
> http://singa.apache.org/en/docs/layer.html#singa.layer.RNN.forward
> The input should be a vector of Tensors, <x1, x2, ... x128, hx, cx>, xi is
> of shape (1500, 9), hx and cx are optional whose shape should be (1500, 28).
> The output would be a vector of Tensors, <y1, y2, ..., y128, hy, cy>, yi
> is of shape (1500, 28), hy and cy are optional depending on the existence
> of hx and cx.
> If you want to put the dense layer on top of the last rnn unit (i.e. the
> 128-th), then you feed y128 to the dense layer.
>
> function convert just reshapes the raw data into a sequence of tensors
> <x1, x2, ..>.
>
> BTW, typically, people would use a smaller batchsize e.g. less than 256.
>
> May I forward our discussion to the incubator email list in case others
> have similar problems?
> Thanks.
>
> Best,
> Wei
>
> So here what I have:
>
> input batch of dimension (1500, 128, 9)
> This means a batch of 1500 windows each having 128 vector of 9 dimensions.
>
> input label of dimension (1500, 6)
> This means a label batch of 1500 vector of 6 dimensions. This is to label
> if the person is sitting ([1,0,0,0,0,0]) or standing ([0,1,0,0,0,0]), etc.
>
> I am creating an lstm layer with hidden_size=28 and
> input_sample_shape=(9,) and num_stacks=1
>
> Then I create a dense layer with num_output=6 and input_sample_shape=(28,)
>
> Now I would like to feed the data to the 'forward' function of lstm and
> dense layer. But I could not make it work and I could not quit understand
> from the example what 'convert' and 'numpy2tensors' are suppose to do...
>
> I would appreciate your comments..
>
> On Sun, Sep 25, 2016 at 12:23 PM, Arash Shafiei <arash.shaf...@gmail.com>
> wrote:
>
> Yes, I was thinking of batch size to be 32.
>
> Thanks. I am getting more how it works and I am thinking how RNN would be
> helpful. Because we do not want to predict a sequence. We just have a
> sequence (in raw data) and a set of features (in processed data) and we
> want to know the classification.
>
> So I was thinking of using other approaches with SINGA. I understood that
> there is also MLP. We could use MLP from SINGA to see the result first.
>
> In this case input would be a set of 561 values with a label.
> Then the MLP, given a set of test data with 561 features would predict the
> label.
>
> Thanks for advices..
>
>
>
> On Sun, Sep 25, 2016 at 12:03 PM, Wang Wei <wang...@comp.nus.edu.sg>
> wrote:
>
>
>
> On Sun, Sep 25, 2016 at 9:37 AM, Arash Shafiei <arash.shaf...@gmail.com>
> wrote:
>
> Hi Wang Wei,
>
> I am trying to understand the char-nn example, but there is still
> something that I am missing and cannot figure is out by myself.
>
> The convert function creates two numpy array x and y. As I understood the
> array x is the data and array y are labels.
>
> I checked the dimentions of these arrays.
> x.shape is (32, 100, 101)
> y.shape is (32, 100)
>
> 32 is the batch size
> 100 is the sequence size
> 101 is the vocabulary size, i.e. there ae 101 unique chars in the
> linux_input.txt.  each input from one sample and at one time step is a
> one-hot vector with all positions being 0 except the position of the
> character (set to 1).
>
>
> given a sequence of chars,   a,b,c,d,e,f
> if the input (x) is  a, b, c, d, e
> then the label is  b, c, d, e, f
>
>
>
> In my understanding you are taking a batch of 100 character and the next
> character must be the label. So according to my understanding
> x.shape must be (32, 100)
> y.shape must be (32, 1)
>
> I mean that you have a batch of 32 sample to train and each sample is a
> series of 100 character. For each sample, there must be a label, which says
> what character must follow this series. And that character is only 1.
>
> Is there anything that I do not quit understand?
>

Re: activity recognition using apache singa

[Wang Wei]

You may need use gcc4.8 or gcc4.9 to reinstall protobuf.

BTW, there is a compatibility problem between gcc5.x and cuda 7,
http://stackoverflow.com/questions/34996295/trying-to-get-cuda-7-5-to-work-with-gcc-5-x



On Thu, Sep 22, 2016 at 4:31 PM, Arash Shafiei <arash.shaf...@gmail.com>
wrote:

> Yes, we are using the same link.
>
> gcc version is 5.4.0
>
> On Thu, Sep 22, 2016 at 4:24 PM, Wang Wei <wang...@comp.nus.edu.sg> wrote:
>
>> Are you using the singa wheel for protobuf 2.6 (i.e. the following link)?
>>
>> http://comp.nus.edu.sg/~dbsystem/singa/assets/file/pb2.6-cuda7.5-cudnn5/singa-1.0.0-cp27-none-linux_x86_64.whl
>>
>> May I know your gcc version for compiling protobuf?
>>
>> On Thu, Sep 22, 2016 at 4:10 PM, Arash Shafiei <arash.shaf...@gmail.com>
>> wrote:
>>
>>> Hi Wang Wei,
>>>
>>> Here is the output of ldd /usr/local/lib/python2.7/dist-
>>> packages/singa/_singa_wrap.so
>>>
>>> linux-vdso.so.1 =>  (0x7ffc5abba000)
>>> libprotobuf.so.9 => /usr/local/lib/libprotobuf.so.9
>>> (0x7f20dd5c3000)
>>> libpthread.so.0 => /lib/x86_64-linux-gnu/libpthread.so.0
>>> (0x7f20dd3a5000)
>>> libcudnn.so.5 => /usr/lib/x86_64-linux-gnu/libcudnn.so.5
>>> (0x7f20d985a000)
>>> libcudart.so.7.5 => /usr/local/cuda/lib64/libcudart.so.7.5
>>> (0x7f20d95fc000)
>>> libcurand.so.7.5 => /usr/local/cuda/lib64/libcurand.so.7.5
>>> (0x7f20d5d93000)
>>> libcublas.so.7.5 => /usr/local/cuda/lib64/libcublas.so.7.5
>>> (0x7f20d44b4000)
>>> libopenblas.so.0 => /usr/lib/libopenblas.so.0 (0x7f20d242)
>>> libpython2.7.so.1.0 => /usr/local/lib/libpython2.7.so.1.0
>>> (0x7f20d2022000)
>>> libstdc++.so.6 => /usr/lib/x86_64-linux-gnu/libstdc++.so.6
>>> (0x7f20d1ca)
>>> libm.so.6 => /lib/x86_64-linux-gnu/libm.so.6 (0x7f20d1997000)
>>> libgcc_s.so.1 => /lib/x86_64-linux-gnu/libgcc_s.so.1
>>> (0x7f20d178)
>>> libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x7f20d13b7000)
>>> libz.so.1 => /lib/x86_64-linux-gnu/libz.so.1 (0x7f20d119d000)
>>> /lib64/ld-linux-x86-64.so.2 (0x55d5047c5000)
>>> librt.so.1 => /lib/x86_64-linux-gnu/librt.so.1 (0x7f20d0f94000)
>>> libdl.so.2 => /lib/x86_64-linux-gnu/libdl.so.2 (0x7f20d0d9)
>>> libgfortran.so.3 => /usr/lib/x86_64-linux-gnu/libgfortran.so.3
>>> (0x7f20d0a64000)
>>> libutil.so.1 => /lib/x86_64-linux-gnu/libutil.so.1
>>> (0x7f20d0861000)
>>> libquadmath.so.0 => /usr/lib/x86_64-linux-gnu/libquadmath.so.0
>>> (0x7f20d0621000)
>>>
>>>
>>> Although we installed libprotobuf2.6, but it shows that the
>>> libprotobuf.so.9 is installed.
>>>
>>> We can verify the version of protobuf using:
>>> protoc --version
>>> libprotoc 2.6.0
>>>
>>> We are installing the protobuf from the source:
>>> https://github.com/google/protobuf/releases/tag/v2.6.0
>>>
>>> Would you know where we could be doing wrong?
>>>
>>> Thanks.
>>>
>>>
>>> On Thu, Sep 22, 2016 at 3:34 PM, Wang Wei <wang...@comp.nus.edu.sg>
>>> wrote:
>>>
>>>> Hi Arash,
>>>>
>>>> Have you installed google protobuf, which is a dependent library of
>>>> SINGA?
>>>> Can you send me the output from running
>>>>
>>>> ldd /usr/local/lib/python2.7/dist-packages/singa/_singa_wrap.so
>>>>
>>>> It would display the version of the protobuf wheel is using.
>>>>
>>>> Please make sure the version of the protobuf matches the wheel version.
>>>> Use the singa wheel for 2.5-protobuf if you installed protobuf 2.5 on
>>>> your machine.
>>>> Use the singa wheel for 2.6-protobuf if you installed protobuf 2.6 on
>>>> your machine.
>>>>
>>>> You don't need to reset the path if If you use virtualenv.
>>>>
>>>>
>>>> On Thu, Sep 22, 2016 at 3:24 PM, Arash Shafiei <arash.shaf...@gmail.com
>>>> > wrote:
>>>>
>>>>> Hi Wang Wei, Thanks.
>>>>>
>>>>> We installed it from the wheel and that problem was solved.
>>>>>
>>>>> Now we are facing a new error message:
>>>>> No module name _singa_wrap
>>>>>
>>&g

Re: activity recognition using apache singa

[Wang Wei]

Are you using the singa wheel for protobuf 2.6 (i.e. the following link)?

http://comp.nus.edu.sg/~dbsystem/singa/assets/file/pb2.6-cuda7.5-cudnn5/singa-1.0.0-cp27-none-linux_x86_64.whl

May I know your gcc version for compiling protobuf?

On Thu, Sep 22, 2016 at 4:10 PM, Arash Shafiei <arash.shaf...@gmail.com>
wrote:

> Hi Wang Wei,
>
> Here is the output of ldd /usr/local/lib/python2.7/dist-
> packages/singa/_singa_wrap.so
>
> linux-vdso.so.1 =>  (0x7ffc5abba000)
> libprotobuf.so.9 => /usr/local/lib/libprotobuf.so.9
> (0x7f20dd5c3000)
> libpthread.so.0 => /lib/x86_64-linux-gnu/libpthread.so.0
> (0x7f20dd3a5000)
> libcudnn.so.5 => /usr/lib/x86_64-linux-gnu/libcudnn.so.5
> (0x7f20d985a000)
> libcudart.so.7.5 => /usr/local/cuda/lib64/libcudart.so.7.5
> (0x7f20d95fc000)
> libcurand.so.7.5 => /usr/local/cuda/lib64/libcurand.so.7.5
> (0x7f20d5d93000)
> libcublas.so.7.5 => /usr/local/cuda/lib64/libcublas.so.7.5
> (0x7f20d44b4000)
> libopenblas.so.0 => /usr/lib/libopenblas.so.0 (0x7f20d242)
> libpython2.7.so.1.0 => /usr/local/lib/libpython2.7.so.1.0
> (0x7f20d2022000)
> libstdc++.so.6 => /usr/lib/x86_64-linux-gnu/libstdc++.so.6
> (0x7f20d1ca)
> libm.so.6 => /lib/x86_64-linux-gnu/libm.so.6 (0x7f20d1997000)
> libgcc_s.so.1 => /lib/x86_64-linux-gnu/libgcc_s.so.1
> (0x7f20d178)
> libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x7f20d13b7000)
> libz.so.1 => /lib/x86_64-linux-gnu/libz.so.1 (0x7f20d119d000)
> /lib64/ld-linux-x86-64.so.2 (0x55d5047c5000)
> librt.so.1 => /lib/x86_64-linux-gnu/librt.so.1 (0x7f20d0f94000)
> libdl.so.2 => /lib/x86_64-linux-gnu/libdl.so.2 (0x7f20d0d9)
> libgfortran.so.3 => /usr/lib/x86_64-linux-gnu/libgfortran.so.3
> (0x7f20d0a64000)
> libutil.so.1 => /lib/x86_64-linux-gnu/libutil.so.1
> (0x7f20d0861000)
> libquadmath.so.0 => /usr/lib/x86_64-linux-gnu/libquadmath.so.0
> (0x7f20d0621000)
>
>
> Although we installed libprotobuf2.6, but it shows that the
> libprotobuf.so.9 is installed.
>
> We can verify the version of protobuf using:
> protoc --version
> libprotoc 2.6.0
>
> We are installing the protobuf from the source:
> https://github.com/google/protobuf/releases/tag/v2.6.0
>
> Would you know where we could be doing wrong?
>
> Thanks.
>
>
> On Thu, Sep 22, 2016 at 3:34 PM, Wang Wei <wang...@comp.nus.edu.sg> wrote:
>
>> Hi Arash,
>>
>> Have you installed google protobuf, which is a dependent library of SINGA?
>> Can you send me the output from running
>>
>> ldd /usr/local/lib/python2.7/dist-packages/singa/_singa_wrap.so
>>
>> It would display the version of the protobuf wheel is using.
>>
>> Please make sure the version of the protobuf matches the wheel version.
>> Use the singa wheel for 2.5-protobuf if you installed protobuf 2.5 on
>> your machine.
>> Use the singa wheel for 2.6-protobuf if you installed protobuf 2.6 on
>> your machine.
>>
>> You don't need to reset the path if If you use virtualenv.
>>
>>
>> On Thu, Sep 22, 2016 at 3:24 PM, Arash Shafiei <arash.shaf...@gmail.com>
>> wrote:
>>
>>> Hi Wang Wei, Thanks.
>>>
>>> We installed it from the wheel and that problem was solved.
>>>
>>> Now we are facing a new error message:
>>> No module name _singa_wrap
>>>
>>> Then we set the path to the singa folder located at:
>>> /usr/local/lib/python2.7/dist-packages/singa
>>>
>>> And we are getting this error message:
>>> mportError: /usr/local/lib/python2.7/dist-packages/singa/_singa_wrap.so:
>>> undefined symbol: _ZNK6google8protobuf7Message11GetTypeNameEv
>>>
>>> It seems there is a link problem. Would you know where the problem could
>>> be?
>>>
>>> Thanks.
>>>
>>>
>>>
>>> On Thu, Sep 22, 2016 at 2:23 PM, Wang Wei <wang...@comp.nus.edu.sg>
>>> wrote:
>>>
>>>> Hi Arash,
>>>>
>>>> Have you installed PySINGA via the following approaches (
>>>> http://singa.apache.org/en/docs/installation.html#install-pysinga)?
>>>>
>>>> # in build/ folder
>>>> cmake -DUSE_CUDA=ON -DUSE_PYTHON=ON ..
>>>> make
>>>> cd python
>>>> pip install .
>>>>
>>>> or using the python wheel
>>>>
>>>> On Thu, Sep 22, 2016 at 2:02 PM, Arash Shafiei <arash.shaf...@gmail.com
>>>> > wrote:
&g

Re: activity recognition using apache singa

[Wang Wei]

Hi Arash,

Have you installed google protobuf, which is a dependent library of SINGA?
Can you send me the output from running

ldd /usr/local/lib/python2.7/dist-packages/singa/_singa_wrap.so

It would display the version of the protobuf wheel is using.

Please make sure the version of the protobuf matches the wheel version.
Use the singa wheel for 2.5-protobuf if you installed protobuf 2.5 on your
machine.
Use the singa wheel for 2.6-protobuf if you installed protobuf 2.6 on your
machine.

You don't need to reset the path if If you use virtualenv.


On Thu, Sep 22, 2016 at 3:24 PM, Arash Shafiei <arash.shaf...@gmail.com>
wrote:

> Hi Wang Wei, Thanks.
>
> We installed it from the wheel and that problem was solved.
>
> Now we are facing a new error message:
> No module name _singa_wrap
>
> Then we set the path to the singa folder located at:
> /usr/local/lib/python2.7/dist-packages/singa
>
> And we are getting this error message:
> mportError: /usr/local/lib/python2.7/dist-packages/singa/_singa_wrap.so:
> undefined symbol: _ZNK6google8protobuf7Message11GetTypeNameEv
>
> It seems there is a link problem. Would you know where the problem could
> be?
>
> Thanks.
>
>
>
> On Thu, Sep 22, 2016 at 2:23 PM, Wang Wei <wang...@comp.nus.edu.sg> wrote:
>
>> Hi Arash,
>>
>> Have you installed PySINGA via the following approaches (
>> http://singa.apache.org/en/docs/installation.html#install-pysinga)?
>>
>> # in build/ folder
>> cmake -DUSE_CUDA=ON -DUSE_PYTHON=ON ..
>> make
>> cd python
>> pip install .
>>
>> or using the python wheel
>>
>> On Thu, Sep 22, 2016 at 2:02 PM, Arash Shafiei <arash.shaf...@gmail.com>
>> wrote:
>>
>>> Dear Wang Wei,
>>>
>>> We are still working on the activity recognition project and have some
>>> difficulties installing SINGA.
>>>
>>> The python2.7 is installed. Trying to do RNN training using "python2.7
>>> train.py linux_input.txt", we get the following error:
>>>
>>> ImportError: No module name singa.
>>>
>>> It seems that python cannot find the singa module. We tried to set the
>>> path of singa in PYTHONPATH but it didn't help.
>>>
>>> Python is installed in:
>>>
>>> /usr/include/python2.7
>>> /usr/lib/python2.7
>>> /usr/bin/python2.7
>>>
>>> Singa is installed in:
>>> /usr/local/include/singa
>>> /usr/local/bin/singa
>>>
>>> Thanks.
>>> Arash
>>>
>>>
>>> On Fri, Sep 16, 2016 at 3:24 PM, Wang Wei <wang...@comp.nus.edu.sg>
>>> wrote:
>>>
>>>> Sure.
>>>> You can use the API of RNN (http://singa.apache.org/en/do
>>>> cs/layer.html#singa.layer.RNN) to construct a supervised model and
>>>> train it with labels.
>>>> E.g., a simple model would be input->rnn->dense->softmax->cross-entropy
>>>> loss.
>>>> The loss is computed to reduce the prediction error against the ground
>>>> truth labels.
>>>>
>>>> You may need to read some papers on activity recognition and find
>>>> proper datasets, e.g. http://archive.ics.uci.edu/ml/
>>>> datasets/Heterogeneity+Activity+Recognition.
>>>> I think one problem is that you need to do some feature engineering for
>>>> the raw data, whose feature has few dimensions.
>>>> We have tried to use the original 3-dimendional feature with a MLP
>>>> model, the performance is not good (about 50% accuracy). But the
>>>> 500-dimensional feature in the above dataset has about 90% accuracy.
>>>>
>>>> Best,
>>>> Wei
>>>>
>>>>
>>>
>>
>

Re: activity recognition using apache singa

[Wang Wei]

Hi Arash,

Have you installed PySINGA via the following approaches (
http://singa.apache.org/en/docs/installation.html#install-pysinga)?

# in build/ folder
cmake -DUSE_CUDA=ON -DUSE_PYTHON=ON ..
make
cd python
pip install .

or using the python wheel

On Thu, Sep 22, 2016 at 2:02 PM, Arash Shafiei <arash.shaf...@gmail.com>
wrote:

> Dear Wang Wei,
>
> We are still working on the activity recognition project and have some
> difficulties installing SINGA.
>
> The python2.7 is installed. Trying to do RNN training using "python2.7
> train.py linux_input.txt", we get the following error:
>
> ImportError: No module name singa.
>
> It seems that python cannot find the singa module. We tried to set the
> path of singa in PYTHONPATH but it didn't help.
>
> Python is installed in:
>
> /usr/include/python2.7
> /usr/lib/python2.7
> /usr/bin/python2.7
>
> Singa is installed in:
> /usr/local/include/singa
> /usr/local/bin/singa
>
> Thanks.
> Arash
>
>
> On Fri, Sep 16, 2016 at 3:24 PM, Wang Wei <wang...@comp.nus.edu.sg> wrote:
>
>> Sure.
>> You can use the API of RNN (http://singa.apache.org/en/do
>> cs/layer.html#singa.layer.RNN) to construct a supervised model and train
>> it with labels.
>> E.g., a simple model would be input->rnn->dense->softmax->cross-entropy
>> loss.
>> The loss is computed to reduce the prediction error against the ground
>> truth labels.
>>
>> You may need to read some papers on activity recognition and find proper
>> datasets, e.g. http://archive.ics.uci.edu/ml/
>> datasets/Heterogeneity+Activity+Recognition.
>> I think one problem is that you need to do some feature engineering for
>> the raw data, whose feature has few dimensions.
>> We have tried to use the original 3-dimendional feature with a MLP model,
>> the performance is not good (about 50% accuracy). But the 500-dimensional
>> feature in the above dataset has about 90% accuracy.
>>
>> Best,
>> Wei
>>
>>
>

Re: [VOTE] Release apache-singa-incubating-1.0.0 (RC2)

[Wang Wei]

+1 (binding)

I checked the installation and running on Mac OS.(if dependent libs are not
in system folder, users need to export them).

regards,
wang wei


On Thu, Sep 1, 2016 at 11:12 AM, WANG Sheng <wan...@comp.nus.edu.sg> wrote:

> Hi all,
>
> I have created a build for Apache SINGA 1.0.0-incubating, release candidate
> RC2.
>
> The artifacts to be voted on are located here:
> https://dist.apache.org/repos/dist/dev/incubator/singa/1.0.0/
>
> The hashes of the artifacts are as follows:
> MD5:  AE 01 1F 67 D7 F0 F6 23  FF 26 F9 A9 F0 00 1C C3
>
> Release artifacts are signed with the following key:
> https://people.apache.org/keys/committer/dinhtta.asc
>
> and the signature file is:
> https://dist.apache.org/repos/dist/dev/incubator/singa/1.0.
> 0/apache-singa-incubating-1.0.0-RC2.tar.gz.asc
>
> The Github tag is at:
> https://github.com/apache/incubator-singa/releases/tag/v1.0.0-rc2
> commit ID is: 0416a0fdfa63cb1b4f54e438ff5b33d7eb4d8df0
>
> To check the license, you can use the Apache Rat tool as follows:
> 1. download & decompress apache rat from
> http://creadur.apache.org/rat/download_rat.cgi
> 2. run the following command under singa folder:
> java -jar /PATH/TO/RAT/apache-rat-0.11.jar -E rat-excludes -d . >
> rat_check
> 3. check the results in file named "rat_check"
>
> Please check and vote on releasing this package. The vote is open for at
> least 72 hours and passes if a majority of at least three +1 votes are
> cast.
>
> [ ] +1 Release this package as Apache SINGA 1.0.0-incubating
> [ ]  0 I don't feel strongly about it, but I'm okay with the release
> [ ] -1 Do not release this package because...
>
> Here is my vote:
> +1
>
> Regards,
> Wang Sheng
>

Re: Spiking NN

[Wang Wei]

Hi Erhard,

Please go ahead to add example models on top of signa.

For the Mac installation, @zhongle could install it successfully. Could you
please paste the error?

On Fri, Jul 29, 2016 at 1:30 PM, Erhard M. Dinhobl jun. <
e...@master-studios.net> wrote:

> Hi guys
>
> Is it interesting to implement the functionality of spiking für singa?
> if so I would begin it in my repo.
>
> BTW: mac installation is still not working! I studied the source code
> the last days and trying to install on mac over and over.
>
> cheers
>
> E
>
> --
>
> master-studios.net - Intelligent Software / Consulting / Research
>
> +43 (0)699 / 102 69 056, Vienna, Austria

Re: long short term memory in singe

[Wang Wei]

Welcome to join us!

Please follow the readme file
https://github.com/apache/incubator-singa/tree/dev for installation and a
CNN example.
We will add more documentation next week and prepare the release by the end
of this month.

Thanks.

regards,


On Tue, Jul 19, 2016 at 2:02 PM, Shayan Shams <ssha...@lsu.edu> wrote:

> Thats a good news that you are working on lstm and gru implementation and
> im more than glad to help in developing that. I have been following your
> development on bersion 1 and i really like it. Very good job. Just to be
> clear in version one you will have both GRU and also Lstm im asking because
> in the current version you just have the GRU
> Thanks alot
> And i start working on version1 asap
>
> Shayan Shams
>
> On Jul 19, 2016, at 00:58, Wang Wei <wangwei...@gmail.com> wrote:
>
> Yes. We will support CPU and GPU for CNN and RNN.
> The v0.3 supports CPU and GPU for CNN and GRU.
> We are working on v1.0 which would be much easier to use and to customize.
> The GPU part of RNN (including LSTM/GRU/TANH) would be ready next week.
> I am busy on the v1.0 development, and may not have time to work on LSTM
> for v0.3 until the end of this month.
> Sorry about that..
>
> I would encourage you to try v1.0 and help us to improve it as you are
> familiar with LSTM and GRU.
>
> regards,
> Wei
>
>
> On Tue, Jul 19, 2016 at 1:44 PM, Shayan Shams <ssha...@lsu.edu> wrote:
>
>>
>> Hi Wei
>> Thanks for reply i am using lstm on the top of cnn for some medical image
>> data I do have gpu but i prefer to have both gpu and cpu code and as far as
>> i know you guys have gru but not lstm implemented and another reason i have
>> changed the cnn codes and im running them on cpu so it would be great if i
>> can run both of them on cpu
>> Shayan Shams
>>
>> On Jul 19, 2016, at 00:22, Wang Wei <wangwei...@gmail.com> wrote:
>>
>> Hi Shanyan,
>>
>> May I know your purpose or application of running lstm?
>> We are working on RNN (LSTM/TANH/GRU) using cuDNN
>> https://github.com/apache/incubator-singa/pull/203
>>
>> If you have GPU, I think it would be more efficient to use cuDNN v5 for
>> RNN applications.
>>
>> regards,
>> Wei
>>
>> On Mon, Jul 18, 2016 at 12:53 PM, Shayan Shams <ssha...@lsu.edu> wrote:
>>
>>> Dear all,
>>>
>>> I am writing to ask for you consultation about a problem that I have
>>> been faced with. I tried to use the current GRU implementation and change
>>> it a little bit to create  LSTM on singa, I have created the following
>>> files in singe
>>>
>>> "lstm.cc"
>>>
>>> lstm section in "job.proto"
>>>
>>> lstm connection layer in "neuralnet.h"
>>>
>>> lstm layer in "neuron_layer.h"
>>>
>>> and also I created one function called mem to return mem_ in "layer.h"
>>>
>>> I have registered my layer in "driver.cc"
>>>
>>> I have attached all the files.
>>>
>>> I was able to recompile the singa but when I tried to run the char-rnn
>>> example for which the conf file is there with just changing the gru layer
>>> to lstm,
>>>
>>> it is compiled and running without any error but after some steps of
>>> training the gradvec[1] is getting so big  that I get nan for lost. I
>>> tried different parameter and if I make the learning rate small it works
>>> but it doesnt converge.
>>>
>>> I tried to debug it but i couldnt.
>>>
>>>
>>> so would you mind please take a look at it and guide me in correct
>>> direction, I know its a lot to ask but any help is very much appreciated.
>>>
>>> I sincerely appreciate your kindness and consideration.
>>>
>>> Yours Sincerely,
>>>
>>> Shayan Shams
>>>
>>
>>
>

Re: Problems on mac

[Wang Wei]

OK. I see. It works by installing from source.
It would be better to figure out the problem of homebrew as it is preferred
by MacOS users.
http://caffe.berkeleyvision.org/install_osx.html

On Sun, Jul 17, 2016 at 10:02 PM, Xiezhongle <xiezhon...@comp.nus.edu.sg>
wrote:

> He installed Protobuf_v2.6.1.
> I installed Profobuf on our mac with default settings and it worked. The
> version is also 2.6.1.
>
> -Original Message-
> From: wangwei...@gmail.com [mailto:wangwei...@gmail.com] On Behalf Of
> Wang Wei
> Sent: Sunday, 17 July, 2016 9:59 PM
> To: dev@singa.incubator.apache.org
> Subject: Re: Problems on mac
>
> brew install --build-from-source --with-python --fresh -vd protobuf
>
> May have a try.
>
> Or would it the problem of protobuf 3.0? hombrew would install protobuf
> 3.0 by default http://brewformulas.org/protobuf
>
> On Sun, Jul 17, 2016 at 9:43 PM, Xiezhongle <xiezhon...@comp.nus.edu.sg>
> wrote:
>
> > Cmake does find protobuf and the compilation turns out errors as the
> > protobuf library is installed with wrong settings.
> > I am not sure about how it happens.
> > Maybe homebrew causes the problem or perhaps wrong parameters are
> > specified during installing.
> >
> > -Original Message-
> > From: wangwei...@gmail.com [mailto:wangwei...@gmail.com] On Behalf Of
> > Wang Wei
> > Sent: Sunday, 17 July, 2016 9:35 PM
> > To: dev@singa.incubator.apache.org
> > Subject: Re: Problems on mac
> >
> > On Sun, Jul 17, 2016 at 9:25 PM, Xiezhongle
> > <xiezhon...@comp.nus.edu.sg>
> > wrote:
> >
> > > I think the problem is about the configuration when installing
> Protobuf.
> > > The make script sets the architecture as "x86_64" somehow.
> > >
> > Do you mean homebrew installs protobuf as 'x86_64'?
> > and then cmake cannot find protobuf?
> >
> > >
> > > -Original Message-
> > > From: wangwei...@gmail.com [mailto:wangwei...@gmail.com] On Behalf
> > > Of Wang Wei
> > > Sent: Sunday, 17 July, 2016 9:23 PM
> > > To: zhongle
> > > Cc: dev@singa.incubator.apache.org
> > > Subject: Re: Problems on mac
> > >
> > > What is the problem of homebrew?
> > > I think people using MacOS would typically use homebrew just like
> > > using 'sudo apt-get' for ubuntu OS.
> > >
> > > On Sun, Jul 17, 2016 at 7:28 PM, Erhard M. Dinhobl jun. <
> > > e...@master-studios.net> wrote:
> > >
> > > > I built protobuf now from source! I also reconfigured my whole mac
> > > > to get default configuration. Everything should be fine now!
> > > >
> > > > --
> > > >
> > > > master-studios.net - Intelligent Software / Consulting / Research
> > > >
> > > > +43 (0)699 / 102 69 056, Vienna, Austria
> > > >
> > > > Am 17.07.2016 13:27, schrieb Erhard M. Dinhobl jun.:
> > > >
> > > > > Hi
> > > > >
> > > > > I first removed protobuf from brew (no install). then:
> > > > >
> > > > > CPATH=/usr/local/include
> > > > >
> > > > > DYLD_LIBRARY_PATH=/usr/local/lib
> > > > >
> > > > > LD_LIBRARY_PATH=/usr/local/lib
> > > > >
> > > > > LIBRARY_PATH=/usr/local/lib
> > > > >
> > > > >
> > > > PATH=/usr/local/lib:/usr/local/bin:/usr/bin:/bin:/usr/sbin:/sbin:/
> > > > Li br
> > > > ary/TeX/texbin:/Users/Erhard/Library/Developer/android-ndk-r10e
> > > > >
> > > > > I removed the build
> > > > >
> > > > > created it with:
> > > > >
> > > > > $ cmake ..
> > > > >
> > > > > $ make
> > > > >
> > > > > again:
> > > > >
> > > > > Undefined symbols for architecture x86_64:
> > > > > "google::protobuf::Message::DebugString() const", referenced from:
> > > > > singa::Channel::Send(google::protobuf::Message const&) in
> > > > > channel.cc.o
> > > > >
> > > > "google::protobuf::Message::SerializeToOstream(std::__1::basic_ost
> > > > re am <char, std::__1::char_traits >*) const", referenced
> > > > from:
> > > > > singa::Channel::Send(google::protobuf::Message const&) in
> > > > > channel.cc.o
> > > > > ld: symbol(s) not found for architecture x86_64
> > 

Re: Problems on mac

[Wang Wei]

On Sun, Jul 17, 2016 at 9:25 PM, Xiezhongle <xiezhon...@comp.nus.edu.sg>
wrote:

> I think the problem is about the configuration when installing Protobuf.
> The make script sets the architecture as "x86_64" somehow.
>
Do you mean homebrew installs protobuf as 'x86_64'?
and then cmake cannot find protobuf?

>
> -Original Message-
> From: wangwei...@gmail.com [mailto:wangwei...@gmail.com] On Behalf Of
> Wang Wei
> Sent: Sunday, 17 July, 2016 9:23 PM
> To: zhongle
> Cc: dev@singa.incubator.apache.org
> Subject: Re: Problems on mac
>
> What is the problem of homebrew?
> I think people using MacOS would typically use homebrew just like using
> 'sudo apt-get' for ubuntu OS.
>
> On Sun, Jul 17, 2016 at 7:28 PM, Erhard M. Dinhobl jun. <
> e...@master-studios.net> wrote:
>
> > I built protobuf now from source! I also reconfigured my whole mac to
> > get default configuration. Everything should be fine now!
> >
> > --
> >
> > master-studios.net - Intelligent Software / Consulting / Research
> >
> > +43 (0)699 / 102 69 056, Vienna, Austria
> >
> > Am 17.07.2016 13:27, schrieb Erhard M. Dinhobl jun.:
> >
> > > Hi
> > >
> > > I first removed protobuf from brew (no install). then:
> > >
> > > CPATH=/usr/local/include
> > >
> > > DYLD_LIBRARY_PATH=/usr/local/lib
> > >
> > > LD_LIBRARY_PATH=/usr/local/lib
> > >
> > > LIBRARY_PATH=/usr/local/lib
> > >
> > >
> > PATH=/usr/local/lib:/usr/local/bin:/usr/bin:/bin:/usr/sbin:/sbin:/Libr
> > ary/TeX/texbin:/Users/Erhard/Library/Developer/android-ndk-r10e
> > >
> > > I removed the build
> > >
> > > created it with:
> > >
> > > $ cmake ..
> > >
> > > $ make
> > >
> > > again:
> > >
> > > Undefined symbols for architecture x86_64:
> > > "google::protobuf::Message::DebugString() const", referenced from:
> > > singa::Channel::Send(google::protobuf::Message const&) in
> > > channel.cc.o
> > >
> > "google::protobuf::Message::SerializeToOstream(std::__1::basic_ostream
> > <char, std::__1::char_traits >*) const", referenced from:
> > > singa::Channel::Send(google::protobuf::Message const&) in
> > > channel.cc.o
> > > ld: symbol(s) not found for architecture x86_64
> > > clang: error: linker command failed with exit code 1 (use -v to see
> > invocation)
> > > make[2]: *** [lib/libsinga_utils.dylib] Error 1
> > > make[1]: *** [src/CMakeFiles/singa_utils.dir/all] Error 2
> > >
> > > I checked every file. in the dylib files, there are evidences, that
> > those two methods are in.
> > >
> > > Anyone other a suggestion?
> > >
> > > --
> > >
> > > master-studios.net - Intelligent Software / Consulting / Research
> > >
> > > +43 (0)699 / 102 69 056, Vienna, Austria
> > >
> > > Am 16.07.2016 20:59, schrieb Xiezhongle:
> > >
> > >> Hi Erhard,
> > >>
> > >> I believe that the problem is caused by Protobuf, not SINGA.
> > >>
> > >> Could you please try to reinstall it?
> > >>
> > >> I suggest you compile Protobuf from source code instead of using
> > homebrew.
> > >>
> > >> Best,
> > >>
> > >> Zhongle
> > >>
> > >> From: Erhard M. Dinhobl jun. [mailto:e...@master-studios.net]
> > >> Sent: Sunday, 17 July, 2016 2:06 AM
> > >> To: dev@singa.incubator.apache.org
> > >> Cc: Xiezhongle
> > >> Subject: RE: Problems on mac
> > >>
> > >> BTW: it would be greate because to get it up running because I
> > >> invested
> > now more than 40 hours in this problem and my mac is currently my main
> > working station.
> > >>
> > >> --
> > >>
> > >> master-studios.net - Intelligent Software / Consulting / Research
> > >>
> > >> +43 (0)699 / 102 69 056, Vienna, Austria
> > >>
> > >> Am 16.07.2016 19:40, schrieb Erhard M. Dinhobl jun.:
> > >>
> > >> Hi xiezhongle
> > >>
> > >> I motivated myself trying another run on mac (to work also at home
> > >> and
> > not in my bureau only, at home I have more time). I did:
> > >>
> > >> $ mkdir build
> > >>
> > >> $ cd build
> > >>
> > >> $ cmake

Re: Problems on mac

[Wang Wei]

What is the problem of homebrew?
I think people using MacOS would typically use homebrew just like using
'sudo apt-get' for ubuntu OS.

On Sun, Jul 17, 2016 at 7:28 PM, Erhard M. Dinhobl jun. <
e...@master-studios.net> wrote:

> I built protobuf now from source! I also reconfigured my whole mac to
> get default configuration. Everything should be fine now!
>
> --
>
> master-studios.net - Intelligent Software / Consulting / Research
>
> +43 (0)699 / 102 69 056, Vienna, Austria
>
> Am 17.07.2016 13:27, schrieb Erhard M. Dinhobl jun.:
>
> > Hi
> >
> > I first removed protobuf from brew (no install). then:
> >
> > CPATH=/usr/local/include
> >
> > DYLD_LIBRARY_PATH=/usr/local/lib
> >
> > LD_LIBRARY_PATH=/usr/local/lib
> >
> > LIBRARY_PATH=/usr/local/lib
> >
> >
> PATH=/usr/local/lib:/usr/local/bin:/usr/bin:/bin:/usr/sbin:/sbin:/Library/TeX/texbin:/Users/Erhard/Library/Developer/android-ndk-r10e
> >
> > I removed the build
> >
> > created it with:
> >
> > $ cmake ..
> >
> > $ make
> >
> > again:
> >
> > Undefined symbols for architecture x86_64:
> > "google::protobuf::Message::DebugString() const", referenced from:
> > singa::Channel::Send(google::protobuf::Message const&) in channel.cc.o
> >
> "google::protobuf::Message::SerializeToOstream(std::__1::basic_ostream std::__1::char_traits >*) const", referenced from:
> > singa::Channel::Send(google::protobuf::Message const&) in channel.cc.o
> > ld: symbol(s) not found for architecture x86_64
> > clang: error: linker command failed with exit code 1 (use -v to see
> invocation)
> > make[2]: *** [lib/libsinga_utils.dylib] Error 1
> > make[1]: *** [src/CMakeFiles/singa_utils.dir/all] Error 2
> >
> > I checked every file. in the dylib files, there are evidences, that
> those two methods are in.
> >
> > Anyone other a suggestion?
> >
> > --
> >
> > master-studios.net - Intelligent Software / Consulting / Research
> >
> > +43 (0)699 / 102 69 056, Vienna, Austria
> >
> > Am 16.07.2016 20:59, schrieb Xiezhongle:
> >
> >> Hi Erhard,
> >>
> >> I believe that the problem is caused by Protobuf, not SINGA.
> >>
> >> Could you please try to reinstall it?
> >>
> >> I suggest you compile Protobuf from source code instead of using
> homebrew.
> >>
> >> Best,
> >>
> >> Zhongle
> >>
> >> From: Erhard M. Dinhobl jun. [mailto:e...@master-studios.net]
> >> Sent: Sunday, 17 July, 2016 2:06 AM
> >> To: dev@singa.incubator.apache.org
> >> Cc: Xiezhongle
> >> Subject: RE: Problems on mac
> >>
> >> BTW: it would be greate because to get it up running because I invested
> now more than 40 hours in this problem and my mac is currently my main
> working station.
> >>
> >> --
> >>
> >> master-studios.net - Intelligent Software / Consulting / Research
> >>
> >> +43 (0)699 / 102 69 056, Vienna, Austria
> >>
> >> Am 16.07.2016 19:40, schrieb Erhard M. Dinhobl jun.:
> >>
> >> Hi xiezhongle
> >>
> >> I motivated myself trying another run on mac (to work also at home and
> not in my bureau only, at home I have more time). I did:
> >>
> >> $ mkdir build
> >>
> >> $ cd build
> >>
> >> $ cmake
> -DPROTOBUF_INCLUDE_DIR:PATH=/usr/local/Cellar/protobuf/2.6.1/include \
> >>
> --DPROTOBUF_LIBRARY:FILEPATH=/usr/local/Cellar/protobuf/2.6.1/lib/libprotobuf.dylib
> \
> >>
> -DPROTOBUF_LIBRARY_DEBUG:FILEPATH=/usr/local/Cellar/protobuf/2.6.1/lib/libprotobuf.dylib
> \
> >>
> -DPROTOBUF_LITE_LIBRARY:FILEPATH=/usr/local/Cellar/protobuf/2.6.1/lib/libprotobuf-lite.dylib
> \
> >>
> -DPROTOBUF_LITE_LIBRARY_DEBUG:FILEPATH=/usr/local/Cellar/protobuf/2.6.1/lib/libprotobuf-lite.dylib
> \
> >>
> -DPROTOBUF_PROTOC_EXECUTABLE:FILEPATH=/usr/local/Cellar/protobuf/2.6.1/bin/protoc
> \
> >>
> -DPROTOBUF_PROTOC_LIBRARY:FILEPATH=/usr/local/Cellar/protobuf/2.6.1/lib/libprotoc.dylib
> \
> >>
> -DPROTOBUF_PROTOC_LIBRARY_DEBUG:FILEPATH=/usr/local/Cellar/protobuf/2.6.1/lib/libprotoc.dylib
> \
> >> -DUSE_CUDA=OFF -DUSE_CUDNN=OFF ..
> >>
> >> (those are all taken from the CMakeCache-File)
> >>
> >> and then:
> >>
> >> $ make
> >>
> >> result is again:
> >>
> >> Undefined symbols for architecture x86_64:
> >> "google::protobuf::Message::DebugString() const", referenced from:
> >> singa::Channel::Send(google::protobuf::Message const&) in channel.cc.o
> >>
> "google::protobuf::Message::SerializeToOstream(std::__1::basic_ostream std::__1::char_traits >*) const", referenced from:
> >> singa::Channel::Send(google::protobuf::Message const&) in channel.cc.o
> >> ld: symbol(s) not found for architecture x86_64
> >> clang: error: linker command failed with exit code 1 (use -v to see
> invocation)
> >> make[2]: *** [lib/libsinga_utils.dylib] Error 1
> >> make[1]: *** [src/CMakeFiles/singa_utils.dir/all] Error 2
> >> make: *** [all] Error 2
> >>
> >> I don't know where to define the lib that it finds the x86_64 arch.
> There exists every path and what ever. I really don't know.
> >>
> >> --
> >>
> >> master-studios.net - Intelligent Software / Consulting / Research
> >>
> >> +43 (0)699 / 102 69 056, Vienna, Austria
> >>
> >> Am 

Re: Install of Singa ... Doc

[Wang Wei]

Hi,

We are working on the documentation of the dev branch (i.e., v1.0)
https://github.com/moazreyad/incubator-singa/tree/master/doc.
Welcome to contribute!

regards,


On Sat, Jul 9, 2016 at 11:52 PM, Erhard M. Dinhobl <e...@master-studios.net>
wrote:

> Btw: I am currently writing a documentation for my own understanding and
> for tha lulz.
> --
> Von: Xiangrui CAI <caix...@gmail.com>
> Gesendet: ‎09.‎07.‎2016 10:18
> An: e...@master-studios.net; dev@singa.incubator.apache.org
> Cc: wangwei...@gmail.com
> Betreff: Re: Install of Singa (2)
>
> Hi Erhard,
>
> Thanks for your feedback. We fixed the problem in textfile_reader.cc.
> Please pull and check it again.
>
> Best Regards!
>
> Wang Wei <wangwei...@gmail.com>于2016年7月8日周五 下午10:24写道：
>
>> could you pls check it..
>> -- Forwarded message --
>> From: Erhard M. Dinhobl jun. <e...@master-studios.net>
>> Date: Fri, Jul 8, 2016 at 7:03 PM
>> Subject: Re: Install of Singa (2)
>> To: dev@singa.incubator.apache.org
>> Cc: wangwei...@gmail.com
>>
>>
>> Hi wang!
>>
>> the problems with lib cnmem are gone. another problem now:
>>
>> In file included from
>> /home/erhard/apache-singa/src/io/textfile_reader.cc:20:0:
>> /home/erhard/apache-singa/src/io/textfile_reader.cc: In member function
>> ‘virtual void singa::io::TextFileReader::SeekToFirst()’:
>> /home/erhard/apache-singa/src/io/textfile_reader.cc:63:15: error: no
>> match for ‘operator!=’ (operand types are ‘std::ifstream {aka
>> std::basic_ifstream}’ and ‘std::nullptr_t’)
>>CHECK(fdat_ != nullptr);
>>^
>> /home/erhard/apache-singa/include/singa/utils/logging.h:96:9: note: in
>> definition of macro ‘CHECK’
>>if (!(condition)) \
>>
>>
>>
>> --
>>
>> master-studios.net - Intelligent Software / Consulting / Research
>>
>> +43 (0)699 / 102 69 056, Vienna, Austria
>>
>>
>>
>> Am 08.07.2016 11:46, schrieb Wang Wei:
>>
>> It should be fixed now.
>> Pls try again https://github.com/apache/incubator-singa/tree/dev.
>> Thanks.
>>
>> regards,
>>
>> On Fri, Jul 8, 2016 at 4:27 PM, Wang Wei <wang...@comp.nus.edu.sg> wrote:
>>
>> Hi Erhard,
>>
>> We have noticed that problem, and will send a PR soon today.
>> cnmem is only required when using cuda.
>>
>> regards,
>>
>> On Fri, Jul 8, 2016 at 4:20 PM, Erhard M. Dinhobl jun. <
>> e...@master-studios.net> wrote:
>>
>> And if I remove all cnmem references from the CMakeList.txt file, it
>> works until:
>>
>> In file included from
>> /home/erhard/apache-singa/src/io/textfile_reader.cc:20:0:
>> /home/erhard/apache-singa/src/io/textfile_reader.cc: In member function
>> 'virtual void singa::io::TextFileReader::SeekToFirst()':
>> /home/erhard/apache-singa/src/io/textfile_reader.cc:63:15: error: no
>> match for 'operator!=' (operand types are 'std::ifstream {aka
>> std::basic_ifstream}' and 'std::nullptr_t')
>>
>> --
>>
>> master-studios.net - Intelligent Software / Consulting / Research
>>
>> +43 (0)699 / 102 69 056, Vienna, Austria
>>
>>
>>
>>
>>
>>

Re: Contribution

[Wang Wei]

Hi Erhard,

Here are some papers for v0.3, http://singa.apache.org/.
For the v1.0, we do not have much documentation currently.
You can take a look at the jira tickets (created after SINGA-162), and try
to get started by implementing one ticket that is not assigned to anyone
yet.
You can add comments there if you need further information about the
tickets.

regards,

On Mon, Jun 27, 2016 at 11:10 PM, Erhard M. Dinhobl jun. <
e...@master-studios.net> wrote:

> Hi!
>
> Big thanks. I will have a look at it. Is there some initial paper I can
> read or some other information which is good to get into the code? I also
> got in contact with a PhD student from louisiana. We will have a skype
> call. Also my aim is to get some knowledge to prepare my PhD and get a
> mentor somewhere.
>
>
>
> --
>
> master-studios.net - Intelligent Software / Consulting / Research
>
> +43 (0)699 / 102 69 056, Vienna, Austria
>
>
>
> Am 27.06.2016 04:23, schrieb Wang Wei:
>
> Hi Erhard,
>
> Welcome to join us!
> We are currently working on the version V1.0
> https://issues.apache.org/jira/browse/SINGA-162.
> You can work on any features that you are interested in.
> If you need any clarifications, please let us know.
>
> regards,
> wang wei
>
> On Mon, Jun 27, 2016 at 2:43 AM, Erhard M. Dinhobl jun. <
> e...@master-studios.net> wrote:
>
>> Hi Singa devs!
>>
>> I am very interested in participating in developing apache singa. Is it
>> possible, useful and how to start?
>>
>> I am a software developer/architect with more than 17 years experience
>> and knowledge about AI.
>>
>> Thanks
>>
>> Erhard
>>
>> --
>>
>> master-studios.net - Intelligent Software / Consulting / Research
>>
>> +43 (0)699 / 102 69 056, Vienna, Austria
>
>

Re: Contribution

[Wang Wei]

Hi Erhard,

Welcome to join us!
We are currently working on the version V1.0
https://issues.apache.org/jira/browse/SINGA-162.
You can work on any features that you are interested in.
If you need any clarifications, please let us know.

regards,
wang wei


On Mon, Jun 27, 2016 at 2:43 AM, Erhard M. Dinhobl jun. <
e...@master-studios.net> wrote:

> Hi Singa devs!
>
> I am very interested in participating in developing apache singa. Is it
> possible, useful and how to start?
>
> I am a software developer/architect with more than 17 years experience
> and knowledge about AI.
>
> Thanks
>
> Erhard
>
> --
>
> master-studios.net - Intelligent Software / Consulting / Research
>
> +43 (0)699 / 102 69 056, Vienna, Austria

Re: Error while running singa on mesos

[Wang Wei]

Hi Venkat,

It should be the problem of the node address.
Pls replace node0 and node2 with their IP addresses.

regards,
wei

On Wed, Jun 22, 2016 at 2:40 AM, Venkat Katta <ska...@adobe.com> wrote:

> i tried running without mesos i got the same error
>
>
> root@node0:~/incubator-singa# ./bin/singa-run.sh -conf
> examples/cifar10/hybrid.conf
> Unique JOB_ID is 4
> Record job information to /tmp/singa-log/job-info/job-4-20160621-183305
> Executing @ node2 : cd /root/incubator-singa; source
> /root/incubator-singa/conf/profile; ./singa -singa_conf
> /root/incubator-singa/conf/singa.conf -singa_job 4 -conf
> /root/incubator-singa/examples/cifar10/hybrid.conf
> Executing @ node0 : cd /root/incubator-singa; source
> /root/incubator-singa/conf/profile; ./singa -singa_conf
> /root/incubator-singa/conf/singa.conf -singa_job 4 -conf
> /root/incubator-singa/examples/cifar10/hybrid.conf
> F0621 18:33:24.171468   725 socket.cc:98] Check failed: port != -1 (-1 vs.
> -1) tcp://node2:*
> *** Check failure stack trace: ***
> @ 0x7f10d0a6b9fd  google::LogMessage::Fail()
> @ 0x7f10d0a6d89d  google::LogMessage::SendToLog()
> @ 0x7f10d0a6b5ec  google::LogMessage::Flush()
> @ 0x7f10d0a6e1be  google::LogMessageFatal::~LogMessageFatal()
> @ 0x7f10d0e05d79  singa::Router::Bind()
> @ 0x7f10d0d7a8bc  singa::Driver::Train()
> @ 0x7f10d0d7f48b  singa::Driver::Train()
> @   0x40c915  main
> @ 0x7f10c5f13f45  (unknown)
> @   0x40cb7e  (unknown)
> F0621 18:33:06.244278  1042 socket.cc:98] Check failed: port != -1 (-1 vs.
> -1) tcp://node0:*
> *** Check failure stack trace: ***
> @ 0x7f6d4516d9fd  google::LogMessage::Fail()
> @ 0x7f6d4516f89d  google::LogMessage::SendToLog()
> @ 0x7f6d4516d5ec  google::LogMessage::Flush()
> @ 0x7f6d451701be  google::LogMessageFatal::~LogMessageFatal()
> @ 0x7f6d45507d79  singa::Router::Bind()
> @ 0x7f6d4547c8bc  singa::Driver::Train()
> @ 0x7f6d4548148b  singa::Driver::Train()
> @   0x40c915  main
> @ 0x7f6d3a615f45  (unknown)
> @   0x40cb7e  (unknown)
> bash: line 1:   725 Aborted (core dumped) ./singa
> -singa_conf /root/incubator-singa/conf/singa.conf -singa_job 4 -conf
> /root/incubator-singa/examples/cifar10/hybrid.conf -host node2
> bash: line 1:  1042 Aborted (core dumped) ./singa
> -singa_conf /root/incubator-singa/conf/singa.conf -singa_job 4 -conf
> /root/incubator-singa/examples/cifar10/hybrid.conf -host node0
> E0621 18:33:07.467438  1067 job_manager.cc:156] job 4 not exists
>
>
> --
> *From:* Wang Wei <wang...@apache.org>
> *Sent:* Tuesday, June 21, 2016 7:09:46 PM
> *To:* Venkat Katta
> *Cc:* dev@singa.incubator.apache.org
> *Subject:* Re: Error while running singa on mesos
>
> Hi,
>
> Can you try to run it without Mesos?
> 1. Compile singa with enable-dist
> 2. change conf/singa.conf to set the zookeeper host
> 3. update the conf/hostfile one line per machine
> 4. update the conf/profile to export LD_LIBRARY_PATH
>
> regards,
> Wei
>
> On Tue, Jun 21, 2016 at 8:52 PM, Venkat Katta <ska...@adobe.com> wrote:
>
>> Hi,
>>
>>
>> I am actually trying to run singa on mesos in fully distributed
>> architecture. I built the docker images as given in the documentation. I am
>> using mesos 0.28.2 and singa 0.3-rc3.I am running each docker container
>> using --net=host flag so that they take the ip of the system. Singa works
>> as long as the workers are all in one machine .
>> When I try to use two machines for training it shows error
>>
>>
>> F0617 10:00:43.862246 2742 socket.cc:98] Check failed: port != -1 (-1 vs.
>> -1) tcp://localhost:*
>>
>>
>>   so while running the scheduler do we need to give it hostfile
>> containing all the hosts. How does it know the remaining hosts in cluster.
>>
>>
>> Thanks,
>>
>>
>> Venkat Satish Katta.
>>
>
>

Re: Error while running singa on mesos

[Wang Wei]

Hi,

Can you try to run it without Mesos?
1. Compile singa with enable-dist
2. change conf/singa.conf to set the zookeeper host
3. update the conf/hostfile one line per machine
4. update the conf/profile to export LD_LIBRARY_PATH

regards,
Wei

On Tue, Jun 21, 2016 at 8:52 PM, Venkat Katta  wrote:

> Hi,
>
>
> I am actually trying to run singa on mesos in fully distributed
> architecture. I built the docker images as given in the documentation. I am
> using mesos 0.28.2 and singa 0.3-rc3.I am running each docker container
> using --net=host flag so that they take the ip of the system. Singa works
> as long as the workers are all in one machine .
> When I try to use two machines for training it shows error
>
>
> F0617 10:00:43.862246 2742 socket.cc:98] Check failed: port != -1 (-1 vs.
> -1) tcp://localhost:*
>
>
>   so while running the scheduler do we need to give it hostfile
> containing all the hosts. How does it know the remaining hosts in cluster.
>
>
> Thanks,
>
>
> Venkat Satish Katta.
>

Re: Feedback from the podling report

[Wang Wei]

Hi Alan,

Thanks for reminding us this issue.
We have some candidate new committers and would like to call for vote when
the next release is ready (scheduled in July).

regards,
wang wei

On Fri, Jun 3, 2016 at 8:49 AM, Pravin Chandrasekaran <pra...@opalcrest.com>
wrote:

> Hi Alan,
>
> I am curious to understand more by what you mean. Can we have a chat?
>
> Regards,
> Pravin
>
>
>
>  Original message 
> From: Alan Gates <alanfga...@gmail.com>
> Date: 02/06/2016 19:46 (GMT-05:00)
> To: dev@singa.incubator.apache.org
> Subject: Feedback from the podling report
>
> I wanted to raise a concern that Singa has been in the incubator for over
> a year now and not added any committers.  Growing the community is one of
> the measures of maturity that Apache looks for in graduating projects from
> the incubator.  Are there any contributors that look like they might be
> ready for committership soon?  If not, do you have any thoughts on how to
> attract some?
>
> Alan.
>

Re: Question about data parallelism within group

[Wang Wei]

Hi Richard,

Yes, you are correct.
Each worker loads 4 records independently from the same data path.
To make them load different records, we do a little trick by assigning
different starting offsets to the two workers, which is set as
```
store_conf {
  ...
  random_skip: 5000  // the starting offset is random number in [0, 5000)
}
```

regards,
Wei



On Thu, May 12, 2016 at 7:07 AM, Richard Platania 
wrote:

> It is my understanding that data parallelism within a group should split
> the batch evenly among the workers in the group. However, I noticed that
> each worker is loading the exact same records. For example, consider a
> batch size of 10, two workers in a group, and partition dimension of 0
> (batch dimension) on the network. I would expect the first and second
> worker to be given records 0-4 and 5-9 respectively. Instead, this is
> resulting in both workers loading a copy of records 0-4.
>
> If this is intended, it would be great if someone could clear up why data
> parallelism configuration is causing multiple workers in a group to have
> the same records.
>
> Regards,
> Richard Platania
>

[ANNOUNCE] Apache SINGA (incubating) 0.3.0 release

[Wang Wei]

 The Apache SINGA (incubating) team is pleased to announce the release of
SINGA 0.3.0.

SINGA is a general distributed deep learning platform for training big deep
learning models over large datasets. It is designed with an intuitive
programming model based on the layer abstraction. SINGA supports a wide
variety of popular deep learning models.

The release is available at:
http://singa.apache.org/downloads.html

The major features of this release include:

   - Training on GPU cluster enables training of deep learning models over
   a GPU cluster.
   - Improve Python wrapper.
   - New SGD updaters are added, including Adam, AdaDelta and AdaMax.
   - Remove dependency on some libraries for single node training.
   - Enable heterogeneous training with CPU and GPU.
   - Support cuDNN V4.

We look forward to hearing your feedbacks, suggestions, and contributions
to the project (http://singa.apache.org/develop/schedule.html).

Regards,
The Apache SINGA (incubating) team



*Disclaimer*

Apache SINGA is an effort undergoing incubation at The Apache Software
Foundation (ASF), sponsored by the name of Apache Incubator PMC. Incubation
is required of all newly accepted projects until a further review indicates
that the infrastructure, communications, and decision making process have
stabilized in a manner consistent with other successful ASF projects. While
incubation status is not necessarily a reflection of the completeness or
stability of the code, it does indicate that the project has yet to be
fully endorsed by the ASF.

Re: [VOTE] Release apache-singa-incubating-0.3.0 (RC3)

[Wang Wei]

+1.

Checked MD5, signature, licenses, 'make' and 'make test'.
Thanks.

regards,
wang wei

On Sun, Apr 17, 2016 at 2:29 PM, Anh Dinh <dinh...@comp.nus.edu.sg> wrote:

> Hi all,
>
> I have created a build for Apache SINGA 0.3.0-incubating, release candidate
> RC3.
>
> The artifacts to be voted on are located here:
> https://dist.apache.org/repos/dist/dev/incubator/singa/0.3.0/
>
> The hashes of the artifacts are as follows:
> MD5:  45 4C 7A BB 17 C7 D6 47  77 85 92 58 59 DF B7 F5
>
> Release artifacts are signed with the following key:
> https://people.apache.org/keys/committer/dinhtta.asc
>
> and the signature file is:
>
> https://dist.apache.org/repos/dist/dev/incubator/singa/0.3.0/apache-singa-incubating-0.3.0-RC3.tar.gz.asc
>
> The Github tag is at:
> https://github.com/apache/incubator-singa/releases/tag/v0.3.0-rc3
> commit ID is: d547a861068973db7d8fc82f9c733e95307a40f1
> Please check and vote on releasing this package. The vote is open for at
> least 72 hours and passes if a majority of at least three +1 votes are
> cast.
>
> To check the license, you can use the Apache Rat tool following
>
> ./configure
> make rat
>
> The result is in rat_check file.
> To install and test the new features, please read the README file.
>
> [ ] +1 Release this package as Apache SINGA 0.3.0-incubating
> [ ]  0 I don't feel strongly about it, but I'm okay with the release
> [ ] -1 Do not release this package because...
>
> Here is my vote:
> +1
>
> Regards,
> Anh.
>

[VOTE] Release apache-singa-incubating-0.3.0 (RC2)

[Wang Wei]

Hi all,

I have created a build for Apache SINGA 0.3.0-incubating, release candidate
RC2.

The artifacts to be voted on are located here:
https://dist.apache.org/repos/dist/dev/incubator/singa/0.3.0/

The hashes of the artifacts are as follows:
MD5:  15 BC 23 07 EF C1 71 BA  4D 55 30 A0 DE E3 36 73

Release artifacts are signed with the following key:
https://people.apache.org/keys/committer/dinhtta.asc

and the signature file is:
https://dist.apache.org/repos/dist/dev/incubator/singa/0.3.0/apache-singa-incubating-0.3.0-RC2.tar.gz.asc

The github tag is at
https://github.com/apache/incubator-singa/releases/tag/v0.3.0-rc2. The
commit ID is d547a861068973db7d8fc82f9c733e95307a40f1

Please check and vote on releasing this package. The vote is open for at
least 72 hours and passes if a majority of at least three +1votes are cast.

To check the license, you can use the Apache Rat tool following

./configure
make rat

The result is in rat_check file.
To install and test the new features, please read the README file.

[ ] +1 Release this package as Apache SINGA 0.3.0-incubating
[ ]  0 I don't feel strongly about it, but I'm okay with the release
[ ] -1 Do not release this package because...


Here is my vote:
+1


Regards,
Wang Wei

[RESULT][VOTE] Release apache-singa-incubating-0.3.0 (RC1)

[Wang Wei]

Thanks to everyone who has voted and given their comments.
The tally is as follows.
0 binding +1s

5 non-binding +1s:
Wei Wang
Zhongle Xie
Anh Dinh
Zhaojing Luo
Beng Chin Ooi

No 0s or -1s.

I am delighted to announce that the proposal to release Apache SINGA
0.3.0-incubating has passed. We'll now start a vote on the general list.
Those of you in the IPMC, please recast your vote on the new thread.

Regards
Wei Wang

Re: [VOTE] Release apache-singa-incubating-0.3.0 (RC1)

[Wang Wei]

Thanks to everyone who has voted and given their comments.

The tally is as follows.
0 binding +1s

5 non-binding +1s:
Wei Wang
Zhongle Xie
Anh Dinh
Zhaojing Luo
Beng Chin Ooi

No 0s or -1s.

I am delighted to announce that the proposal to release Apache SINGA
0.3.0-incubating has passed.

We'll now start a vote on the general list. Those of you in the IPMC,
please recast your vote on the new thread.

Regards
Wei Wang


On Sat, Apr 9, 2016 at 12:33 PM, Wang Wei <wang...@comp.nus.edu.sg> wrote:

> Hi all,
>
> I have created a build for Apache SINGA 0.3.0-incubating, release candidate
> RC1.
>
> The artifacts to be voted on are located here:
> https://dist.apache.org/repos/dist/dev/incubator/singa/0.3.0/
>
> The hashes of the artifacts are as follows:
> MD5:  F8 F9 81 D7 60 B9 3E F9  29 80 33 B1 18 8B D7 05
>
> Release artifacts are signed with the following key:
> https://people.apache.org/keys/committer/dinhtta.asc
>
> and the signature file is:
>
> https://dist.apache.org/repos/dist/dev/incubator/singa/0.3.0/apache-singa-incubating-0.3.0-RC1.tar.gz.asc
>
> The github tag is at
> https://github.com/apache/incubator-singa/releases/tag/v0.3.0-rc1. The
> commit ID is fb5b2184160f3c2d3b422e5ed58657b6f460d090
>
> Please check and vote on releasing this package. The vote is open for at
> least 72 hours and passes if a majority of at least three +1votes are cast.
>
> To check the license, you can use the Apache Rat tool following
>
> ./configure
> make rat
>
> The result is in rat_check file.
> To install and test the new features, please read the README file and refer
> to SINGA website http://singa.apache.org/docs/index.html.
>
> [ ] +1 Release this package as Apache SINGA 0.3.0-incubating
> [ ]  0 I don't feel strongly about it, but I'm okay with the release
> [ ] -1 Do not release this package because...
>
>
> Here is my vote:
> +1
>
>
> Regards,
> Wang Wei
>

[VOTE] Release apache-singa-incubating-0.3.0 (RC1)

[Wang Wei]

Hi all,

I have created a build for Apache SINGA 0.3.0-incubating, release candidate
RC1.

The artifacts to be voted on are located here:
https://dist.apache.org/repos/dist/dev/incubator/singa/0.3.0/

The hashes of the artifacts are as follows:
MD5:  F8 F9 81 D7 60 B9 3E F9  29 80 33 B1 18 8B D7 05

Release artifacts are signed with the following key:
https://people.apache.org/keys/committer/dinhtta.asc

and the signature file is:
https://dist.apache.org/repos/dist/dev/incubator/singa/0.3.0/apache-singa-incubating-0.3.0-RC1.tar.gz.asc

The github tag is at
https://github.com/apache/incubator-singa/releases/tag/v0.3.0-rc1. The
commit ID is fb5b2184160f3c2d3b422e5ed58657b6f460d090

Please check and vote on releasing this package. The vote is open for at
least 72 hours and passes if a majority of at least three +1votes are cast.

To check the license, you can use the Apache Rat tool following

./configure
make rat

The result is in rat_check file.
To install and test the new features, please read the README file and refer
to SINGA website http://singa.apache.org/docs/index.html.

[ ] +1 Release this package as Apache SINGA 0.3.0-incubating
[ ]  0 I don't feel strongly about it, but I'm okay with the release
[ ] -1 Do not release this package because...


Here is my vote:
+1


Regards,
Wang Wei

Re: [VOTE] Release apache-singa-incubating-0.3.0 (RC1)

[Wang Wei]

I will check with other guys.

On Sat, Apr 9, 2016 at 10:10 AM, Anh Dinh <dinh...@comp.nus.edu.sg> wrote:

> maybe it's just my browser rendering it differently.
>
> On 9 April 2016 at 09:56, Wang Wei <wang...@comp.nus.edu.sg> wrote:
>
>> Hi Anh,
>>
>> The links are to v0.3.0-rc1.
>> which link is to v0.2?
>>
>>
>> On Sat, Apr 9, 2016 at 9:40 AM, Anh Dinh <dinh...@comp.nus.edu.sg> wrote:
>>
>> > The VOTE email contains links to v0.2 version (both to the path and to
>> the
>> > MD5 file)
>> >
>> > Anh.
>> >
>> >
>> > On 9 April 2016 at 01:21, zhaojing <zhaoj...@comp.nus.edu.sg> wrote:
>> >
>> > > +1
>> > >
>> > > I checked:
>> > > + build successful
>> > > + Python Binding for Interactive Training (train_mnist.py and
>> > > train_cifar10.py)
>> > >
>> > > Best regards,
>> > > Zhaojing.
>> > >
>> > >
>> > > On 2016-04-08 22:30, Wang Wei wrote:
>> > >
>> > >> Hi all,
>> > >>
>> > >> I have created a build for Apache SINGA 0.3.0-incubating, release
>> > >> candidate
>> > >> RC1.
>> > >>
>> > >> The artifacts to be voted on are located here:
>> > >> https://dist.apache.org/repos/dist/dev/incubator/singa/0.3.0/
>> > >> <https://dist.apache.org/repos/dist/dev/incubator/singa/0.2.0/>
>> > >>
>> > >> <https://dist.apache.org/repos/dist/dev/incubator/singa/0.2.0/>
>> > >> The hashes of the artifacts are as follows:
>> > >> <http://apache-singa-incubating-0.2.0-RC2.tar.gz.md>
>> > >> MD5:  F8 F9 81 D7 60 B9 3E F9  29 80 33 B1 18 8B D7 05
>> > >>
>> > >> Release artifacts are signed with the following key:
>> > >> https://people.apache.org/keys/committer/dinhtta.asc
>> > >>
>> > >> and the signature file is:
>> > >>
>> > >>
>> >
>> https://dist.apache.org/repos/dist/dev/incubator/singa/0.3.0/apache-singa-incubating-0.3.0-RC1.tar.gz.asc
>> > >>
>> > >> The github tag is at
>> > >> https://github.com/apache/incubator-singa/releases/tag/v0.3.0-rc1.
>> The
>> > >> commit ID is fb5b2184160f3c2d3b422e5ed58657b6f460d090
>> > >>
>> > >>
>> > >> Please check and vote on releasing this package. The vote is open
>> for at
>> > >> least 72 hours and passes if a majority of at least three +1votes are
>> > >> cast.
>> > >>
>> > >> To check the license, you can use the Apache Rat tool following
>> > >>
>> > >> ./configure
>> > >> make rat
>> > >>
>> > >> The result is in rat_check file.
>> > >> To install and test the new features, please read the README file and
>> > >> refer
>> > >> to SINGA website http://singa.apache.org/docs/index.html.
>> > >>
>> > >>
>> > >> [ ] +1 Release this package as Apache SINGA 0.3.0-incubating
>> > >> [ ]  0 I don't feel strongly about it, but I'm okay with the release
>> > >> [ ] -1 Do not release this package because...
>> > >>
>> > >>
>> > >> Here is my vote:
>> > >> +1
>> > >>
>> > >>
>> > >> Regards,
>> > >> Wang Wei
>> > >>
>> > >
>> > >
>> >
>>
>
>

[VOTE] Release apache-singa-incubating-0.3.0 (RC1)

[Wang Wei]

Hi all,

I have created a build for Apache SINGA 0.3.0-incubating, release candidate
RC1.

The artifacts to be voted on are located here:
https://dist.apache.org/repos/dist/dev/incubator/singa/0.3.0/
<https://dist.apache.org/repos/dist/dev/incubator/singa/0.2.0/>

<https://dist.apache.org/repos/dist/dev/incubator/singa/0.2.0/>
The hashes of the artifacts are as follows:
<http://apache-singa-incubating-0.2.0-RC2.tar.gz.md>
MD5:  F8 F9 81 D7 60 B9 3E F9  29 80 33 B1 18 8B D7 05

Release artifacts are signed with the following key:
https://people.apache.org/keys/committer/dinhtta.asc

and the signature file is:
https://dist.apache.org/repos/dist/dev/incubator/singa/0.3.0/apache-singa-incubating-0.3.0-RC1.tar.gz.asc

The github tag is at
https://github.com/apache/incubator-singa/releases/tag/v0.3.0-rc1. The
commit ID is fb5b2184160f3c2d3b422e5ed58657b6f460d090


Please check and vote on releasing this package. The vote is open for at
least 72 hours and passes if a majority of at least three +1votes are cast.

To check the license, you can use the Apache Rat tool following

./configure
make rat

The result is in rat_check file.
To install and test the new features, please read the README file and refer
to SINGA website http://singa.apache.org/docs/index.html.


[ ] +1 Release this package as Apache SINGA 0.3.0-incubating
[ ]  0 I don't feel strongly about it, but I'm okay with the release
[ ] -1 Do not release this package because...


Here is my vote:
+1


Regards,
Wang Wei

Re: [jira] [Commented] (SINGA-146) Implement speech recognition examples

[Wang Wei]

Hi Vishnu,

Please read some papers on speech recognition and learn the programming
model of SINGA.
If you have any idea of implementing these algorithms on SINGA, then just
let us know.

speech recognition papers: thesis of http://www.cs.toronto.edu/~gdahl/
(including CTC), deep speech from baidu http://arxiv.org/abs/1512.02595v1
SINGA: http://singa.apache.org/docs/programming-guide.html

Regard,
Wei

On Mon, Mar 7, 2016 at 3:51 PM, Vishnu Teja (JIRA)  wrote:

>
> [
> https://issues.apache.org/jira/browse/SINGA-146?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15182714#comment-15182714
> ]
>
> Vishnu Teja commented on SINGA-146:
> ---
>
> Hi Sheng,
>
> I have no experience in speech recognition but I have good grip on Machine
> Learning. And I can learn things in a appropriate speed. Please let me know
> if it's considerable.
>
> Thank you :)
>
> > Implement speech recognition examples
> > -
> >
> > Key: SINGA-146
> > URL: https://issues.apache.org/jira/browse/SINGA-146
> > Project: Singa
> >  Issue Type: New Feature
> >Reporter: wangwei
> >  Labels: gsoc, gsoc2016, mentor
> >
> > Speech recognition is a major application of deep learning models. The
> other two major applications are about computer vision and NLP.
> > SINGA has provided examples for image classification and language
> modelling. We are going to provide a few examples about speech recognition
> application, e.g., CTC and DBN-DNN.
>
>
>
> --
> This message was sent by Atlassian JIRA
> (v6.3.4#6332)
>

Re: is it possible to use MPI HPC cluster in singa?

[Wang Wei]

Hi Li Li,

MPI is included in our schedule for v0.3
https://issues.apache.org/jira/browse/SINGA-133.
Please stay tuned, or join us to implement this ticket.

Zookeeper is important for distributed training, as it keeps running data
for failure recovery.

Best,
Wei



On Mon, Jan 18, 2016 at 4:56 PM, Li Li  wrote:

> I have access to a HPC cluster with hundreds of nodes with thousands
> cores. I found there are many deep learning framework can train using
> multiple gpu in a single node. very few frameworks can train in
> multiple nodes using gpu. Is there any framework can train with MPI? I
> can't install long run service like zookeeper or spark in the cluster.
>

Re: New features to be included in v0.3

[Wang Wei]

Hi Mohamed

Yes. Only GRU has been added in v0.2. I think Zhongming (cc'ed) is
implementing LSTM.
It would be interesting to provide an example for acoustic modeling using
CTC.
If you have experience of using CTC for acoustic modelling , it would be
appreciated if you can help to add it in v0.3.

regards,
Wei

On Wed, Jan 13, 2016 at 5:41 PM, Mohamed Yousef <mohamed.ma...@fci.au.edu.eg
> wrote:

> I would suggest implementing an LSTM layer and a CTC  layer (LSTM was
> planned for v0.2, but only GRU was added)
> a CTC layer would greatly simplify acoustic modeling and many other
> applications
> 
> From: wangwei...@gmail.com <wangwei...@gmail.com> on behalf of Wang Wei <
> wang...@comp.nus.edu.sg>
> Sent: Wednesday, January 13, 2016 8:51 AM
> To: dev@singa.incubator.apache.org
> Subject: New features to be included in v0.3
>
> Hi all,
>
> I have proposed some features for v0.3, which is scheduled to be released
> in March 2016.
> The major goals include usability and performance improvements.
>
> I have assigned some of the tasks to some developers/committers.
> If you are interested in any of them, please join the development by
> replying to that ticket.
>
> I will add new tickets continuously.
> You are also welcomed to create new tickets for other features about
> usability and performance improvements.
> Let's get started!
>
> Thanks!
>
> regards,
> Wei
> SINGA-136 <https://issues.apache.org/jira/browse/SINGA-136>
>
> Support cuDNN v4 <https://issues.apache.org/jira/browse/SINGA-136>
> Haibo Chen
> <https://issues.apache.org/jira/secure/ViewProfile.jspa?name=seaokcs>
> *Actions*
> <
> https://issues.apache.org/jira/rest/api/1.0/issues/12929715/ActionsAndOperations?atl_token=A5KQ-2QAV-T4JA-FDED%7Cdab6635ed244de6f2d2e77e2220bfc953beea22a%7Clin
> >
> SINGA-135 <https://issues.apache.org/jira/browse/SINGA-135>
>
> Improve hybrid partitioning and benchmark performance
> <https://issues.apache.org/jira/browse/SINGA-135>
> Sheng Wang
> <https://issues.apache.org/jira/secure/ViewProfile.jspa?name=wangsh>
> *Actions*
> <
> https://issues.apache.org/jira/rest/api/1.0/issues/12929708/ActionsAndOperations?atl_token=A5KQ-2QAV-T4JA-FDED%7Cdab6635ed244de6f2d2e77e2220bfc953beea22a%7Clin
> >
> SINGA-134 <https://issues.apache.org/jira/browse/SINGA-134>
>
> Extend SINGA to run over a GPU cluster
> <https://issues.apache.org/jira/browse/SINGA-134>
> *Unassigned*
> *Actions*
> <
> https://issues.apache.org/jira/rest/api/1.0/issues/12929371/ActionsAndOperations?atl_token=A5KQ-2QAV-T4JA-FDED%7Cdab6635ed244de6f2d2e77e2220bfc953beea22a%7Clin
> >
> SINGA-133 <https://issues.apache.org/jira/browse/SINGA-133>
>
> Implement message passing using MPI
> <https://issues.apache.org/jira/browse/SINGA-133>
> *Unassigned*
> *Actions*
> <
> https://issues.apache.org/jira/rest/api/1.0/issues/12929370/ActionsAndOperations?atl_token=A5KQ-2QAV-T4JA-FDED%7Cdab6635ed244de6f2d2e77e2220bfc953beea22a%7Clin
> >
> SINGA-132 <https://issues.apache.org/jira/browse/SINGA-132>
>
> Optimize training on a single node with GPUs
> <https://issues.apache.org/jira/browse/SINGA-132>
> Haibo Chen
> <https://issues.apache.org/jira/secure/ViewProfile.jspa?name=seaokcs>
> *Actions*
> <
> https://issues.apache.org/jira/rest/api/1.0/issues/12929368/ActionsAndOperations?atl_token=A5KQ-2QAV-T4JA-FDED%7Cdab6635ed244de6f2d2e77e2220bfc953beea22a%7Clin
> >
> SINGA-131 <https://issues.apache.org/jira/browse/SINGA-131>
>
> Implement and optimize hybrid training using both CPU and GPU
> <https://issues.apache.org/jira/browse/SINGA-131>
> *Unassigned*
> *Actions*
> <
> https://issues.apache.org/jira/rest/api/1.0/issues/12929354/ActionsAndOperations?atl_token=A5KQ-2QAV-T4JA-FDED%7Cdab6635ed244de6f2d2e77e2220bfc953beea22a%7Clin
> >
> SINGA-130 <https://issues.apache.org/jira/browse/SINGA-130>
>
> Implement a layer subclass for data prefetching
> <https://issues.apache.org/jira/browse/SINGA-130>
> *Unassigned*
> *Actions*
> <
> https://issues.apache.org/jira/rest/api/1.0/issues/12929190/ActionsAndOperations?atl_token=A5KQ-2QAV-T4JA-FDED%7Cdab6635ed244de6f2d2e77e2220bfc953beea22a%7Clin
> >
> SINGA-129 <https://issues.apache.org/jira/browse/SINGA-129>
>
> Add new SGD updaters <https://issues.apache.org/jira/browse/SINGA-129>
> Jinyang Gao
> <https://issues.apache.org/jira/secure/ViewProfile.jspa?name=jinyang>
> *Actions*
> <
> https://issues.apache.org/jira/rest/api/1.0/issues/12929184/ActionsAndOperations?atl_token=A5KQ-2QAV-T4JA-FDED%7Cdab6635ed

New features to be included in v0.3

[Wang Wei]

Hi all,

I have proposed some features for v0.3, which is scheduled to be released
in March 2016.
The major goals include usability and performance improvements.

I have assigned some of the tasks to some developers/committers.
If you are interested in any of them, please join the development by
replying to that ticket.

I will add new tickets continuously.
You are also welcomed to create new tickets for other features about
usability and performance improvements.
Let's get started!

Thanks!

regards,
Wei
SINGA-136 

Support cuDNN v4 
Haibo Chen

*Actions*

SINGA-135 

Improve hybrid partitioning and benchmark performance

Sheng Wang

*Actions*

SINGA-134 

Extend SINGA to run over a GPU cluster

*Unassigned*
*Actions*

SINGA-133 

Implement message passing using MPI

*Unassigned*
*Actions*

SINGA-132 

Optimize training on a single node with GPUs

Haibo Chen

*Actions*

SINGA-131 

Implement and optimize hybrid training using both CPU and GPU

*Unassigned*
*Actions*

SINGA-130 

Implement a layer subclass for data prefetching

*Unassigned*
*Actions*

SINGA-129 

Add new SGD updaters 
Jinyang Gao

*Actions*

SINGA-128 

Update compilation options and scripts for installing SINGA on MacOS

Xie Zhongle

*Actions*

SINGA-127 

Clear optimization level of compilation for debugging

Xie Zhongle

*Actions*

SINGA-126 

Improve Python Binding for interactive training

*Unassigned*
*Actions*

SINGA-125 

Improve Python Helper - additional features

*Unassigned*

Re: New features to be included in v0.3

[Wang Wei]

Yes. R binding will be included.
I will create a new ticket after getting more knowledge about R:-)

regards,
Wei

On Wed, Jan 13, 2016 at 2:55 PM, ooibc <oo...@comp.nus.edu.sg> wrote:

>
> R binding?
>
> On 2016-01-13 14:51, Wang Wei wrote:
>
>> Hi all,
>>
>> I have proposed some features for v0.3, which is scheduled to be released
>> in March 2016.
>> The major goals include usability and performance improvements.
>>
>> I have assigned some of the tasks to some developers/committers.
>> If you are interested in any of them, please join the development by
>> replying to that ticket.
>>
>> I will add new tickets continuously.
>> You are also welcomed to create new tickets for other features about
>> usability and performance improvements.
>> Let's get started!
>>
>> Thanks!
>>
>> regards,
>> Wei
>> SINGA-136 <https://issues.apache.org/jira/browse/SINGA-136>
>>
>> Support cuDNN v4 <https://issues.apache.org/jira/browse/SINGA-136>
>> Haibo Chen
>> <https://issues.apache.org/jira/secure/ViewProfile.jspa?name=seaokcs>
>> *Actions*
>> <
>> https://issues.apache.org/jira/rest/api/1.0/issues/12929715/ActionsAndOperations?atl_token=A5KQ-2QAV-T4JA-FDED%7Cdab6635ed244de6f2d2e77e2220bfc953beea22a%7Clin
>> >
>> SINGA-135 <https://issues.apache.org/jira/browse/SINGA-135>
>>
>> Improve hybrid partitioning and benchmark performance
>> <https://issues.apache.org/jira/browse/SINGA-135>
>> Sheng Wang
>> <https://issues.apache.org/jira/secure/ViewProfile.jspa?name=wangsh>
>> *Actions*
>> <
>> https://issues.apache.org/jira/rest/api/1.0/issues/12929708/ActionsAndOperations?atl_token=A5KQ-2QAV-T4JA-FDED%7Cdab6635ed244de6f2d2e77e2220bfc953beea22a%7Clin
>> >
>> SINGA-134 <https://issues.apache.org/jira/browse/SINGA-134>
>>
>> Extend SINGA to run over a GPU cluster
>> <https://issues.apache.org/jira/browse/SINGA-134>
>> *Unassigned*
>> *Actions*
>> <
>> https://issues.apache.org/jira/rest/api/1.0/issues/12929371/ActionsAndOperations?atl_token=A5KQ-2QAV-T4JA-FDED%7Cdab6635ed244de6f2d2e77e2220bfc953beea22a%7Clin
>> >
>> SINGA-133 <https://issues.apache.org/jira/browse/SINGA-133>
>>
>> Implement message passing using MPI
>> <https://issues.apache.org/jira/browse/SINGA-133>
>> *Unassigned*
>> *Actions*
>> <
>> https://issues.apache.org/jira/rest/api/1.0/issues/12929370/ActionsAndOperations?atl_token=A5KQ-2QAV-T4JA-FDED%7Cdab6635ed244de6f2d2e77e2220bfc953beea22a%7Clin
>> >
>> SINGA-132 <https://issues.apache.org/jira/browse/SINGA-132>
>>
>> Optimize training on a single node with GPUs
>> <https://issues.apache.org/jira/browse/SINGA-132>
>> Haibo Chen
>> <https://issues.apache.org/jira/secure/ViewProfile.jspa?name=seaokcs>
>> *Actions*
>> <
>> https://issues.apache.org/jira/rest/api/1.0/issues/12929368/ActionsAndOperations?atl_token=A5KQ-2QAV-T4JA-FDED%7Cdab6635ed244de6f2d2e77e2220bfc953beea22a%7Clin
>> >
>> SINGA-131 <https://issues.apache.org/jira/browse/SINGA-131>
>>
>> Implement and optimize hybrid training using both CPU and GPU
>> <https://issues.apache.org/jira/browse/SINGA-131>
>> *Unassigned*
>> *Actions*
>> <
>> https://issues.apache.org/jira/rest/api/1.0/issues/12929354/ActionsAndOperations?atl_token=A5KQ-2QAV-T4JA-FDED%7Cdab6635ed244de6f2d2e77e2220bfc953beea22a%7Clin
>> >
>> SINGA-130 <https://issues.apache.org/jira/browse/SINGA-130>
>>
>> Implement a layer subclass for data prefetching
>> <https://issues.apache.org/jira/browse/SINGA-130>
>> *Unassigned*
>> *Actions*
>> <
>> https://issues.apache.org/jira/rest/api/1.0/issues/12929190/ActionsAndOperations?atl_token=A5KQ-2QAV-T4JA-FDED%7Cdab6635ed244de6f2d2e77e2220bfc953beea22a%7Clin
>> >
>> SINGA-129 <https://issues.apache.org/jira/browse/SINGA-129>
>>
>> Add new SGD updaters <https://issues.apache.org/jira/browse/SINGA-129>
>> Jinyang Gao
>> <https://issues.apache.org/jira/secure/ViewProfile.jspa?name=jinyang>
>> *Actions*
>> <
>> https://issues.apache.org/jira/rest/api/1.0/issues/12929184/ActionsAndOperations?atl_token=A5KQ-2QAV-T4JA-FDED%7Cdab6635ed244de6f2d2e77e2220bfc953beea22a%7Clin
>> >
>> SINGA-128 <https://issues.apache.org/jira/browse/SINGA-128>
>>
>> Update compilation options and scripts for installing SINGA on MacOS
>> <https://issues.apache.org/jira/browse/SINGA-128>
>> Xie Zhongle
>

[VOTE] Release apache-singa-incubating-0.2.0 (RC2)

[Wang Wei]

Hi all,

I have created a build for Apache SINGA 0.2.0-incubating, release candidate
RC2.
The license issues (
http://mail-archives.apache.org/mod_mbox/incubator-general/201601.mbox/%3Ce596cf92fdc46e111538c010a7fe0acb%40comp.nus.edu.sg%3E)
are fixed in RC2.

The artifacts to be voted on are located here:
<https://dist.apache.org/repos/dist/dev/incubator/singa/0.2.0/>
https://dist.apache.org/repos/dist/dev/incubator/singa/0.2.0/
The hashes of the artifacts are as follows:
apache-singa-incubating-0.2.0-RC2.tar.gz.md5:
41 4F 39 EE B0 25 68 38  C1 3A F0 9F 02 82 B2 9D
apache-singa-incubating-0.2.0-RC2.tar.gz.sha256:  42C13D9D D23C6179
7C1D50CA B11947E0 D260342D D37C1B61 7818C9ED B3964BEF

Release artifacts are signed with the following key:
https://people.apache.org/keys/committer/dinhtta.asc

and the signature file is:
apache-singa-incubating-0.2.0-RC2.tar.gz.asc


Please check and vote on releasing this package. The vote is open for at
least 72 hours and passes if a majority of at least three +1votes are cast.

To check the license, you can use the Apache Rat tool following

./configure
make rat

The result is in rat_check file.
To install and test the new features, please read the README file and refer
to SINGA website http://singa.apache.org/docs/index.html.


[ ] +1 Release this package as Apache SINGA 0.2.0-incubating
[ ]  0 I don't feel strongly about it, but I'm okay with the release
[ ] -1 Do not release this package because...


Here is my vote:
+1


Regards,
Wang Wei

SINGA docs repo

[Wang Wei]

Hi All,

I have just setup a github repo for editing signa docs,
https://github.com/nusdbsystem/singa-docs

Steps to update SINGA docs,

1. clone/fetch the latest master
2. edit your docs
3. commit (better describe what has been changed)
4. send pull request

Everyone (committers and non-committers) is now able to help edit the docs.
The admins will merge the pull request and copy the files into Apache SVN
repo.
Github can display the changes of each pull request clearly, which is
helpful.

regards,
wang wei

Re: Contribute

[Wang Wei]

Hi Mohit,

On Mon, Sep 28, 2015 at 6:29 PM, Mohit Singh <mohit1...@gmail.com> wrote:

> Hi Wang Wei,
>   Thanks for quick response and sharing with the next release cycle.
> I  would be interested in:
> 1) Getting up to speed on machine learning algorithms part ( so feed
> forward networks or RNN etc)
>
For feed forward networks, SINGA has provided layers for implementing
AlexNet (imagene2012 paper), VGG (http://arxiv.org/pdf/1409.1556.pdf).
We need to implement some other layers to support the training of GoogleNet
(http://arxiv.org/abs/1409.4842).
For RNN, http://karpathy.github.io/2015/05/21/rnn-effectiveness/ is a good
application to work on. Both LSTM and GRU are on our plan.

I would suggest you to work on GRU and GoogleNet firstly.
You can follow the vanilla rnn example http://singa.apache.org/docs/rnn.html
to implement GRU.

2) Maybe looking into hdfs as input or output sources
>
We are going to use HDFS for storing large training datasets.
It has not yet been decided. Maybe some in-memory storage would be
considered.

> 3) Python bindings?
>
This is to create some wrappers for the Layer, NeuralNet, etc classes. Then
users can use python scripts to run the SINGA functions.
For example, after training, you can load the model parameters to extract
features by passing the data through the network using python scripts.

But right now, want to start by taking something very simple.
> So far, I have been able to build singa and run CIFAR example locally.
>
> Thanks
>
>
> On Mon, Sep 28, 2015 at 3:13 AM, Wang Wei <wang...@comp.nus.edu.sg> wrote:
>
> > Hi Mohit,
> >
> > Thanks for your interest in SINGA.
> >
> > Attached is a schedule for our second release.
> > If you are interested in any topic/feature, then we can discuss in detail
> > on the development.
> >
> > Cheers,
> > Wei Wang
> >
> > On Mon, Sep 28, 2015 at 5:31 PM, Mohit Singh <mohit1...@gmail.com>
> wrote:
> >
> >> Hi,
> >>   I am new to C++ and deep learning..
> >> But want to learn both.
> >> How do I get started in contributing to singa.
> >>
> >> --
> >> Mohit
> >>
> >> "When you want success as badly as you want the air, then you will get
> it.
> >> There is no other secret of success."
> >> -Socrates
> >>
> >
> >
>
>
> --
> Mohit
>
> "When you want success as badly as you want the air, then you will get it.
> There is no other secret of success."
> -Socrates
>

Re: Contribute

[Wang Wei]

Hi Mohit,

Thanks for your interest in SINGA.

Attached is a schedule for our second release.
If you are interested in any topic/feature, then we can discuss in detail
on the development.

Cheers,
Wei Wang

On Mon, Sep 28, 2015 at 5:31 PM, Mohit Singh  wrote:

> Hi,
>   I am new to C++ and deep learning..
> But want to learn both.
> How do I get started in contributing to singa.
>
> --
> Mohit
>
> "When you want success as badly as you want the air, then you will get it.
> There is no other secret of success."
> -Socrates
>


schedule-v2.pptx
Description: MS-Powerpoint 2007 presentation

[UPDATE] [VOTE] Release apache-singa-0.1.0-incubating (RC2)

[Wang Wei]

Hi all,

The hashes of the artifacts are updated to

MD5 (apache-singa-incubating-0.1.0-RC2.tar.gz.md5): 63 0F DF E0 74 E0 E1
1F  89 F6 0E DF 9E 66 50 73

SHA256 (apache-singa-incubating-0.1.0-RC2.tar.gz.sha256): EE7CF820 70DB46F5
20FC39A1 F85B5B73 865503BD 36280917 5369EB9F 5FA7199E
Please check the package and vote. Thanks.

Regards,
Wei Wang



On Sun, Sep 27, 2015 at 5:36 PM, Wang Wei <wang...@comp.nus.edu.sg> wrote:

> Hi all,
>
> I have created a build for Apache SINGA 0.1.0-incubating, release
> candidate RC2.
> RC2 addresses the installation issue reported in the last candidate by
> Alan Gates (Thanks!).
>
> I have filed the JIRA tickets for tracking the addressing of the issues.
> https://issues.apache.org/jira/browse/SINGA-76
> https://issues.apache.org/jira/browse/SINGA-77
>
> The artifacts to be voted on are located here:
> https://dist.apache.org/repos/dist/dev/incubator/singa/
>
> The hashes of the artifacts are as follows:
> apache-singa-incubating-0.1.0-RC2.tar.gz.md5: 7C A8 F7 62 2B 9A EA 3D  D3
> 7B 81 03 90 FD B1 A8
> apache-singa-incubating-0.1.0-RC2.tar.gz.sha256: 200CB022 515AFB22
> 11BD46DA CEAB3DAC 478B40FE 2E5601E8 F57CA52D FAE80CF7
>
> Release artifacts are signed with the following key:
> https://people.apache.org/keys/committer/dinhtta.asc
>
> and the signature file is:
> apache-singa-incubating-0.1.0-RC2.tar.gz.asc
>
> Please check and vote on releasing this package. The vote is open for at
> least 72 hours  or until the necessary number of votes (3 +1) has been
> reached.
>
> [ ] +1 Release this package as Apache SINGA 0.1.0-incubating
> [ ]  0 I don't feel strongly about it, but I'm okay with the release
> [ ] -1 Do not release this package because...
>
> Here is my vote:
>
> +1
> Regards,
>
> Wei Wang
>

[CANCEL] [VOTE] Release apache-singa-0.1.0-incubating (RC1)

[Wang Wei]

Hi all,

I am going to cancel this vote as RC2 vote will be called in a moment,
which addresses the issues reported in this thread.

Regards,
Wei Wang


On Fri, Sep 25, 2015 at 11:20 AM, ooibc  wrote:

>
> Thanks.
>
> The fortran compiler is a dependency of OpenBLAS.
> We will update the README file for the time being, and as a longer term
> solution, we will replace OpenBLAS with our own codes.
>
> regards
> beng chin
>
>
>
> On 2015-09-25 01:12, Alan Gates wrote:
>
>> +0
>>
>>  I checked the LICENSE, NOTICE, and DISCLAIMER files, as well as the
>> signatures. All these looked good. I checked that there did not appear
>> to be in binaries in the distribution. I also checked for all of the
>> source files having appropriate Apache headers.
>>
>>  Issues I saw:
>>  * There is one source file that's missing a header,
>> ./src/test/test_common.cc. This should be fixed but it can be done
>> subsequent to the release.
>>  * I could not get the system to build. When I ran ./configure it said
>> "configure: error: unable to find cblas_sgemm() function" I could not
>> find a package for openblas, and when I downloaded the source for that
>> and tried to build it said I needed a fortran compiler. That's when I
>> bailed out.
>>
>>  I am voting +0 because I don't think the build issue is sufficient to
>> block the release. But if I had trouble building I suspect others will
>> as well. If people can't build your software they are much less likely
>> to use it. So I think it would be in your best interest to resolve the
>> build issue soon.
>>
>>  If there are simple steps to resolve the build issue then documenting
>> them in the README would be very helpful. For example pointers to
>> where a binary version of openblas can be obtained, if such a place
>> exists (and any other libraries not readily obtainable by yum or
>> apt-get). If openblas must be built then some help on how to do so
>> (hopefully without installing a fortran compiler) would be good.
>>
>>  Other miscellaneous advice:
>>  * It would be very helpful to have an Apache Rat target in your
>> makefile. Rat gives you a quick way to check for file missing licenses
>> and some other issues. I had to do it by hand this time.
>>
>>  Alan.
>>
>> zhaojing
>>> September 23, 2015 at 2:25
>>> +1
>>>
>>
>

[VOTE] Release apache-singa-0.1.0-incubating (RC2)

[Wang Wei]

Hi all,

I have created a build for Apache SINGA 0.1.0-incubating, release candidate
RC2.
RC2 addresses the installation issue reported in the last candidate by Alan
Gates (Thanks!).

I have filed the JIRA tickets for tracking the addressing of the issues.
https://issues.apache.org/jira/browse/SINGA-76
https://issues.apache.org/jira/browse/SINGA-77

The artifacts to be voted on are located here:
https://dist.apache.org/repos/dist/dev/incubator/singa/

The hashes of the artifacts are as follows:
apache-singa-incubating-0.1.0-RC2.tar.gz.md5: 7C A8 F7 62 2B 9A EA 3D  D3
7B 81 03 90 FD B1 A8
apache-singa-incubating-0.1.0-RC2.tar.gz.sha256: 200CB022 515AFB22 11BD46DA
CEAB3DAC 478B40FE 2E5601E8 F57CA52D FAE80CF7

Release artifacts are signed with the following key:
https://people.apache.org/keys/committer/dinhtta.asc

and the signature file is:
apache-singa-incubating-0.1.0-RC2.tar.gz.asc

Please check and vote on releasing this package. The vote is open for at
least 72 hours  or until the necessary number of votes (3 +1) has been
reached.

[ ] +1 Release this package as Apache SINGA 0.1.0-incubating
[ ]  0 I don't feel strongly about it, but I'm okay with the release
[ ] -1 Do not release this package because...

Here is my vote:

+1
Regards,

Wei Wang

[VOTE] Release apache-singa-0.1.0-incubating (RC1)

[Wang Wei]

Hi all,

I have created a build for Apache SINGA 0.1.0-incubating, release candidate
RC1.

The artifacts to be voted on are located here:

https://dist.apache.org/repos/dist/dev/incubator/singa/


The hashes of the artifacts are as follows:

apache-singa-0.1.0-incubating.tar.gz.md5: AB 66 B1 C9 FC 25 C8 0F  05 07 A7
D1 93 9D 1C A6

apache-singa-0.1.0-incubating.tar.gz.sha256: 6820E74A 84D8AF49 E519145E
F44B0E4D 58140327 659B170B 715A7D65 3068E0E3

Release artifacts are signed with the following key:

https://people.apache.org/keys/committer/dinhtta.asc

and the signature file is:

apache-singa-0.1.0-incubating.tar.gz.asc

Please check and vote on releasing this package. The vote is open for at
least 72 hours and passes if a majority of at least three +1 votes are cast.

[ ] +1 Release this package as Apache SINGA 0.1.0-incubating

[ ]  0 I don't feel strongly about it, but I'm okay with the release

[ ] -1 Do not release this package because...


Here is my vote:

+1

Regards,

Wang Wei

compile test code

[Wang Wei]

I tried to run the test code, but they are not included in the compilation
files.
We may need to update the Makefile.am or other related files to include a
target for test, e.g., make test.

regards,
wang wei

Fwd: [DISCUSS] Horn Incubation Proposal

[Wang Wei]

Dear Mentors,

We noticed a new proposal (Horn
https://wiki.apache.org/incubator/HornProposal) for developing a
distributed deep learning system.
Horn shares similar design ideas as SINGA. Beng Chin has given some details
below.
Maybe we can invite Horn developers to collaborate on SINGA?
If you know developers from Horn, you can introduce them to us.
Or if you have any other suggestions, please let us know.

BTW, we are going to release the first version of SINGA.
Do you have any suggestions on the release? (Since this is the first
release, we are not quite clear on the process)

Best,
Wei

-- Forwarded message --
From: ooibc oo...@comp.nus.edu.sg
Date: Fri, Aug 21, 2015 at 12:13 PM
Subject: Re: [DISCUSS] Horn Incubation Proposal
To: gene...@incubator.apache.org
Cc: Edward J. Yoon edwardy...@apache.org



Hi,

I am an initial committer of Apache(incubating) SINGA (
http://singa.incubator.apache.org/)

Both SINGA and the proposal follow the general parameter-server
architecture:
workers for computing gradients; servers for parameter updating.

SINGA has implemented the model and data parallelism discussed in the Horn'
proposal:
multiple worker groups for asynchronous training---data parallelism; and
multiple workers in one group for synchronous training---model parallelism.

One feature of SINGA's architecture is that it can be extended to organize
the
servers in a hierarchical topology, which may help to reduce the
communication bottleneck
of servers organized in a flat topology.

For the programming model, currently Horn proposes to support feed-forward
models,
e.g., MLP, auto-encoder, while SINGA supports all three categories of the
known models,
feed-forward models (eg MLP, CNN), energy models (eg RBM, DBM),
and recurrent models (eg. RNN).
SINGA provides good support for users to code, e.g., implement new
parameter updating
protocols or layers, and is being integrated with HDFS as well.

We will submit the first release and full documentation to the mentors this
weekend, and if
ok, we will announce the first full release soon.  The GPU version is
scheduled for
October release.

Technical papers:
  http://www.comp.nus.edu.sg/~ooibc/singa-mm15.pdf
  http://www.comp.nus.edu.sg/~ooibc/singaopen-mm15.pdf

and project website (which has more details than the Apache web site):
  http://www.comp.nus.edu.sg/~dbsystem/singa/


There are plenty of rooms for collaborations indeed...

regards
beng chin
www.comp.nus.edu.sg/~ooibc




On 2015-08-21 08:27, Edward J. Yoon wrote:

 Hi all,

 We'd like to propose Horn (혼), a fully distributed system for
 large-scale deep learning as an Apache Incubator project and start the
 discussion. The complete proposal can be found at:
 https://wiki.apache.org/incubator/HornProposal

 Any advices and helps are welcome! Thanks, Edward.

 = Horn Proposal =

 == Abstract ==

 (tentatively named Horn [hɔ:n], korean meaning of Horn is a
 Spirit) is a neuron-centric programming APIs and execution framework
 for large-scale deep learning, built on top of Apache Hama.

 == Proposal ==

 It is a goal of the Horn to provide a neuron-centric programming APIs
 which allows user to easily define the characteristic of artificial
 neural network model and its structure, and its execution framework
 that leverages the heterogeneous resources on Hama and Hadoop YARN
 cluster.

 == Background ==

 The initial ANN code was developed at Apache Hama project by a
 committer, Yexi Jiang (Facebook) in 2013. The motivation behind this
 work is to build a framework that provides more intuitive programming
 APIs like Google's MapReduce or Pregel and supports applications
 needing large model with huge memory consumptions in distributed way.

 == Rationale ==

 While many of deep learning open source softwares such as Caffe,
 DeepDist, and NeuralGiraph are still data or model parallel only, we
 aim to support both data and model parallelism and also fault-tolerant
 system design. The basic idea of data and model parallelism is use of
 the remote parameter server to parallelize model creation and
 distribute training across machines, and the BSP framework of Apache
 Hama for performing asynchronous mini-batches. Within single BSP job,
 each task group works asynchronously using region barrier
 synchronization instead of global barrier synchronization, and trains
 large-scale neural network model using assigned data sets in BSP
 paradigm. Thus, we achieve data and model parallelism. This
 architecture is inspired by Google's !DistBelief (Jeff Dean et al,
 2012).

 == Initial Goals ==

 Some current goals include:
  * builds new community
  * provides more intuitive programming APIs
  * needs both data and model parallelism support
  * must run natively on both Hama and Hadoop2
  * needs also GPUs and InfiniBand support (FPGAs if possible)

 == Current Status ==

 === Meritocracy ===

 The core developers understand what it means to have a process based
 on meritocracy. We will provide 

Problems on using CMS

[Wang Wei]

Hi Infra,

I am from Incubator-SINGA project.
We are using CMS to manage our documentations. Here are some problems we
have encountered.

1. The web page cannot list all the files. I have tried to force new
working copy, but some files are still missing from the list.
[image: Inline image 1]
I can see all files in the svn folder
https://svn.apache.org/repos/asf/incubator/singa/site/trunk/content/markdown/docs/
.

2. We cannot receive emails notification from anonymous' updates although
they do tell CMS to send an email to dev@singa

3. I cannot commit the updates from anonymous users, the CMS reports:
 Conflict

   A conflict in the working copy obstructs the current operation:
Commit failed (details follow)::  Aborting commit:
'/usr/local/cms/wc/singa/wangwei-anonymous-f0ZIJv/trunk/content/markdown/docs/
data.md' remains in conflict at
/usr/local/cms/webgui/lib/ASF/CMS/WC/Commit.pm line 64

 Please visit the following *uri for recovery steps*.

When I click the uri for recovery steps, it reports:
 Can't resolve this resource as it doesn't appear to be in a conflicted
state.

4. Does the gitpubsub support CMS? Can we switch to gitsubpub without
changing the building tool (we are using maven). Using git with github, we
can review and comment non-committer's documentation patch easily.

Thank you.

Regards,
Wei

Re: Caffe con Troll and SINGA?

[Wang Wei]

Hi Ce,

Thanks for the clarification!
I think I have got the high-level abstractions of CcT.
We will come back to get your help (e.g., transfer some code) after
finishing the GPU version.

Regards,
Wei

On Sat, Jul 25, 2015 at 12:13 PM, Ce Zhang czh...@cs.wisc.edu wrote:

  How do you parallelize the training on CPU and GPU, by creating two
 threads each with one (Caffe) Solver?
  Or does each Layer partition the mini-batch and dispatch them onto two
 threads (one using GPU Driver and one using CPU  Driver) after receiving
 the Pointer?

 The abstraction is like this. A Layer is an abstraction that takes as
 input a Pointer and fills an output Pointer. This allows us to have
 two types of Layers:
1. Single-Device Layer, which corresponding to one worker that only
 works on a single device.
2. Multi-Device Layer, which creates multiple single-device layers,
 starts multiple threads
 correspond to these single-device layers, call their forward/backward
 function
 with p_thread, and join these threads.

 Both types of Layers are the subclass of the same Layer class so they share
 the same interface.

 With this abstraction, I think this corresponds to your second strategy.

  Do I miss any other features of CcT?

 I think these 4-steps are exactly right, and this should be enough
 to integrate all features of CcT!

 One very minor thing to notice is that the parallelization we talked about
 maintains the replica of models on different devices and partition the
 data.
 It is also possible to partition the model instead, which might work
 better
 for layers like FC--I think a similar tradeoff is also in SINGA's
 technical
 report, so I am pretty sure this case will be handled efficiently! :)

 Thanks again for integrating CcT into SINGA! Let us know if you have any
 questions!

 Ce

 On Fri, Jul 24, 2015 at 10:40 PM, Wang Wei wang...@comp.nus.edu.sg
 wrote:

 Hi Ce,

 Thanks for your explanation.
 I read the example training configuration of CcT on github, and have some
 further questions on the parallelization.

 How do you parallelize the training on CPU and GPU, by creating two
 threads each with one (Caffe) Solver?
 Or does each Layer partition the mini-batch and dispatch them onto two
 threads (one using GPU Driver and one using CPU Driver) after receiving the
 Pointer?

 According to the configuration file (
 https://github.com/nudles/CaffeConTroll/blob/master/tests/imagenet_train/train_val/alexnet_train_val_1GPU_CPU.prototxt),

 it seems the parallelization is at the Layer level (i.e, the second one).
 In specific, given two connected layers A-B (e.g., conv and relu).
 If the batch partitioning for A is 0.6 on GPU and 0.4 on CPU, while the
 partitioning for B is 1.0 (ie.. all) on GPU.
 Then the computation for B should be blocked until the computation for A
 is finished.
 Consequently, the synchronization is done for every layer.

 For SINGA, we use the worker-server architecture (similar to parameter
 server).
 Currently, we support both synchronous and asynchronous training on CPU.
 We call the training framework implemented by CcT as synchronous training.
 SINGA implements this training framework by partitioning the neural
 network among one worker group.  One worker runs in a thread.
 The partitioning is currently done at Layer level. Users can configure it
 to be on dimension 0  or 1 of the feature blob.
 For dimension 0, it means partitioning one mini-batch onto workers like
 CcT (but we use equal partitioning).
 For dimension 1, it means partitioning one layer (with 4096 neurons) into
 sub-layers (with 1024 neurons if there are 4 workers).
 After partitioning, each (sub) layer will be assigned a location ID,
 i.e., the ID of the worker to which the (sub) layer will be dispatched.
 We will support partitioning at the neural network level, i.e., let users
 to configure the location ID of one layer.
 During training, each worker it has the full neural network structure,
 but it only visits (e.g., forward pass) the layers that are dispatched to
 him (based on Location ID and worker ID).


 To Support CcT in SINGA,
 1. We first need to support GPU training (should be done in August).
 2. Update the neural network partitioning function to integrate CcT's
 scheduling strategy.
 2. Make the Worker class a template, and create GPU workers and CPU
 workers after partitioning the neural network.
 3. I think the Lowering, Shifting techniques are easy to integrate as a
 library if it is independent of the devices.

 Do I miss any other features of CcT?

 Regards,
 Wei

 A Layer takes as input a Pointer, and calls the `dereference` function of
 Pointer to
 obtain local-copy (w.r.t. the driver) of the data. Therefore, the Layer
 object does not
 know where the input data comes from.

 To run operations like GEMM, Lowering, a Layer will call Driver, which
 provided
 a unified interface across devices.

 The Layer, Pointer and Driver abstractions are clear and easy to
 understand.

 I think it is possible

Re: zookeeper API

[Wang Wei]

In the current implementation, all processes will register itself to
zookeeper including hostname and port number.
It should work well for a cluster of nodes connected using local network.
Because every node can be identified by its hostname.
Registering hostname instead of IP is for further optimization on multiple
network cards. For example, if there are two ethernet cards per node, we
may do some optimization on selecting which card to use at real-time.
If we use IP directly, then we fix the ethernet card and cannot do this
kind of optimization.

For the cluster.conf and model.conf, we assume they are accessible for each
node (e.g., on NFS, or local disk). Hence, each process will load these two
configuration files from disk instead of reading it from Zookeeper.

Do you want to put the content of hostfile and cluster.conf into zookeeper,
and let Mesos read these information from zookeeper?

regards,
wang wei

On Tue, Jun 23, 2015 at 7:24 PM, Anh Dinh ug93...@gmail.com wrote:

 Thanks WangSheng,

 Should I also assume that the distributed version will read both hostfile
 and cluster.conf file from Zookeeper service? It'd make it easier for
 Mesos to manage Singa.

 Cheers,
 Anh.


 On 23 June 2015 at 11:25, WANG Sheng wan...@comp.nus.edu.sg wrote:

  Hi Anh,
 
  You are correct. The start/stop zookeeper service is only required in the
  standalone version.
  In the distributed version, the zookeeper service is always on and
 managed
  by users themselves.
 
  In this case, before running a singa job, zookeeper need to be
 initialized
  (e.g. clean old data, create missing path).
  This initialization phase should only be executed once. In the current
  singa architecture, there is no master node,
  hence we need an external tool to do the job before launching singa. (We
  are planning to implement it next.)
 
  For your request of writing hostfile into zookeeper, it could also be
  done by the same tool.
  Could you write your code as a new file /support/main.cc which we can
 later
  extend to be the tool.
 
  For the zookeeper API, it is included in the ZKClusterRT class in
  /utils/cluster_rt.h.
  I will make the create_zk_node a public function for you to use.
 
  By the way, for each commit we need to attach a jira ticket.
  Please create a jira ticket according to the guide on singa website. (You
  need to create an jira account first)
 
  Best Regards,
 
  Sheng
 
 
 
  On Tue, Jun 23, 2015 at 9:54 AM, Anh Dinh ug93...@gmail.com wrote:
 
   Hi guys,
  
   Currently the standalone version (I followed the Quickstart guide)
 would
   start Zookeeper service  every time I ran singa-run.sh.
  
   I assume that on the distributed version, the ZK service will be
 started
   only once by a master node, and the rest of the cluster will know the
 ZK
   master address?
  
   In this case, since I'm writing Mesos support for singa, could I have
 the
   following API from the SingaZooKeeperService (or any class that
  implements
   ZK service for Singa)?
  
   /**
   * create a node with given name and content.
   * the node can be located at $ZK_PREFIX/filename
   */
   static SingaZooKeeperService::create_zk_node(string filename, string
   content);
  
   I'm using this to write content of the hostfile so that all nodes can
   see.
  
   Anh.
  
 

Code Review

[Wang Wei]

Hi guys,

Please do a code review when you are getting familiar with the current code.
It is important to make the code clean and easy to read for our first
release.

We tried to follow the Google's C++ code style initially. But didn't do it
exactly.
If no other suggestions for the coding style, please still follow the
Google's style when doing the code review.
If you have any suggestions, please let us know.

Thanks.

regards,
Wei