mxnet-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Meghna Baijal <meghnabaijal2...@gmail.com>
Subject Re: Publish MXNet images to DockerHub
Date Thu, 26 Jul 2018 06:08:31 GMT
Hi Anirudh,
Thanks for bringing this up.
The Python Images are being actively released for each MXNet version. Until
last release I was using the script Mu has pointed out but from 1.2.1 I
replaced these dockerfiles to use the pip binaries instead of building from
source.
Images for all other language bindings were being released only until MXNet
0.12.0 since they were not being maintained. I think there are a couple of
github issues open to track broken dockerfiles.

Kellen,

I can help you publish the docker images to dockerhub.


Thanks,

Meghna Baijal




On Tue, Jul 24, 2018 at 4:45 PM Anirudh Acharya <anirudhkrec@gmail.com>
wrote:

> Yes that would be good. Also I just noticed that in the Installation
> instructions page only python has docker image installation instruction
> here -
>
> http://mxnet.incubator.apache.org/install/index.html?platform=Linux&language=Python&processor=CPU
> Similar instructions need to be there for other bindings too.
>
> On Tue, Jul 24, 2018 at 4:04 PM kellen sunderland <
> kellen.sunderland@gmail.com> wrote:
>
> > I was actually interested in pushing a version of MXNet with TensorRT
> > enabled some time in the next few weeks just so that people can
> experiment
> > with the feature without worrying about installing the right protoc and
> > onnx versions.  If people here think it's a good idea I can open a PR
> with
> > a runtime-docker folder with the intent that this work could be a
> template
> > for others who want to contribute runtime Dockerfiles?  If a few
> > contributors do put together an Dockerfile with TensorRT enabled, would
> it
> > be possible to get that image pushed to the MXNet Dockerhub repo by a
> > committer?
> >
> > On Sun, Jul 22, 2018 at 3:57 PM Anirudh Acharya <anirudhkrec@gmail.com>
> > wrote:
> >
> > > @Naveen No, I meant in general, for all bindings. Irrespective of
> whether
> > > we use a package management repository, being able to pull an image
> from
> > > docker hub would be convenient for anyone wanting to get started on
> MXNet
> > > or run services( as Kellen said).
> > >
> > >
> > > On Sun, Jul 22, 2018 at 11:20 AM kellen sunderland <
> > > kellen.sunderland@gmail.com> wrote:
> > >
> > > > I think it's a good idea Anirudh.  It should help users easily get
> > MXNet
> > > up
> > > > and running whether they're running services, following tutorials,
> etc.
> > > >
> > > > On Sun, Jul 22, 2018 at 8:10 AM Naveen Swamy <mnnaveen@gmail.com>
> > wrote:
> > > >
> > > > > I don't think we need for JVM languages, they have a good
> dependency
> > > > > management through Maven Central. We weren't publishing regularly
> to
> > > > Maven,
> > > > > now we do.
> > > > >
> > > > > Anirudh, I am guessing you are interested docker for R language,
If
> > > the R
> > > > > packages were published to CRAN do you still see a need for docker
> ?
> > > > Could
> > > > > you elaborate how this would be helpful and easy if they were to
> use
> > > > other
> > > > > packages in CRAN?
> > > > >
> > > > > On Sat, Jul 21, 2018 at 10:51 PM, Anirudh Acharya <
> > > anirudhkrec@gmail.com
> > > > >
> > > > > wrote:
> > > > >
> > > > > > Yes, correct cu90 is indeed there, thanks for pointing it.
> > > > > >
> > > > > > So the question, should we be publishing to Docker Hub as part
of
> > the
> > > > > > release process so that bindings other than python are also
> > published
> > > > and
> > > > > > there is a policy on what cuda versions we publish?
> > > > > >
> > > > > >
> > > > > > Thanks
> > > > > > ANirudh
> > > > > >
> > > > > > On Sat, Jul 21, 2018 at 9:56 PM Mu Li <muli.cmu@gmail.com>
> wrote:
> > > > > >
> > > > > > > cu90 and cu90mkl are also available, see
> > > > > > > https://hub.docker.com/r/mxnet/python/tags/
> > > > > > >
> > > > > > > On Sat, Jul 21, 2018 at 9:51 PM, Anirudh Acharya <
> > > > > anirudhkrec@gmail.com>
> > > > > > > wrote:
> > > > > > >
> > > > > > > > The python binding that is actively maintained is
> > > > > > > >
> > > > > > > > mxnet-mkl  1.2.1
> > > > > > > >
> > > > > > > >
> > > > > > > > Other versions that use CUDA like mxnet-cu<xx>
and
> > > mxnet-cu<xx>mkl
> > > > > are
> > > > > > > not
> > > > > > > > actively maintained.
> > > > > > > >
> > > > > > > >
> > > > > > > >
> > > > > > > > -
> > > > > > > >
> > > > > > > > Anirudh
> > > > > > > >
> > > > > > > >
> > > > > > > >
> > > > > > > >
> > > > > > > >
> > > > > > > > On Sat, Jul 21, 2018 at 9:09 PM Mu Li <muli.cmu@gmail.com>
> > > wrote:
> > > > > > > >
> > > > > > > > > Surprisingly only the python binding is actively
> maintained.
> > I
> > > > > > remember
> > > > > > > > we
> > > > > > > > > can easily push all bindings into docker hub
through the
> > script
> > > > in
> > > > > > > > >
> https://github.com/apache/incubator-mxnet/tree/master/docker
> > .
> > > > > > > > >
> > > > > > > > > On Sat, Jul 21, 2018 at 5:03 PM, Anirudh Acharya
<
> > > > > > > anirudhkrec@gmail.com>
> > > > > > > > > wrote:
> > > > > > > > >
> > > > > > > > > > Hi,
> > > > > > > > > >
> > > > > > > > > > Docker Hub( https://hub.docker.com/u/mxnet/
) currently
> > > hosts
> > > > > > images
> > > > > > > > of
> > > > > > > > > > MXNet and its various bindings but it is
not actively
> > > > maintained.
> > > > > > > > Should
> > > > > > > > > we
> > > > > > > > > > publish MXNet images to Docker Hub as part
of the release
> > > > process
> > > > > > and
> > > > > > > > > > actively maintain it?
> > > > > > > > > >
> > > > > > > > > > The pros of publishing docker images would
be ease of use
> > and
> > > > > > access
> > > > > > > to
> > > > > > > > > our
> > > > > > > > > > users. Is this something that should be
included as part
> of
> > > the
> > > > > > > release
> > > > > > > > > > process? What does the community think?
> > > > > > > > > >
> > > > > > > > > > Thanks
> > > > > > > > > > Anirudh Acharya
> > > > > > > > > >
> > > > > > > > >
> > > > > > > >
> > > > > > >
> > > > > >
> > > > >
> > > >
> > >
> >
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message