mxnet-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Mu Li <muli....@gmail.com>
Subject Re: Remove Caffe functionality
Date Thu, 19 Jul 2018 20:58:13 GMT
We cannot rely on #issue as a popularity score. No issue means either no
one is using it or it's well documented and has no bug.

Removing functionality is equivalent to API breaking, we should be very
careful about it.

I don't suggest the caffe2->onnx approach for 2 reasons:

1. it's complicated, people need to install pytorch and onnx, since caffe2
is merged into pytorch
2. both caffe2 and onnx are under rapid developing, it's hard to debug if
users submitted issues to us

If putting my amazon hat on, we must keep the caffe converter, since some
major customers are using it.

On Thu, Jul 19, 2018 at 1:42 PM, Afrooze, Sina <sina.beh@gmail.com> wrote:

> I think it'd be more sustainable long term to expect users to convert
> caffe -> caffe2 ->onnx->mxnet rather than caffe->mxnet. Perhaps an MXNet
> tutorial can show how to do this with minimum pain. - Sina
>
>
> ´╗┐On 7/19/18, 1:27 PM, "Steffen Rochel" <steffenrochel@gmail.com> wrote:
>
>     Hi Mu - do we have any indication how many users we do have on the
>     functionality? I see only two github issues from 2017 open.
>
>     On Thu, Jul 19, 2018 at 12:54 PM Mu Li <muli.cmu@gmail.com> wrote:
>
>     > Hi Anton,
>     >
>     > It's understandable that Caffe is old and its community is
> shrinking. For
>     > this very reasons, existing caffe users are looking for
> alternatives. The
>     > current converter is important for such users to transit to MXNet. It
>     > addresses an important issue that: how I can reuse my previous caffe
> models
>     > in MXNet.
>     >
>     > On Thu, Jul 19, 2018 at 11:53 AM, Anton Chernov <mechernov@gmail.com
> >
>     > wrote:
>     >
>     > > Dear community,
>     > >
>     > > Currently MXNet has a Caffe framework integration (translator and
>     > > converter) [1].
>     > >
>     > > There were some issues discovered with it, for example some tests
> were
>     > > failing [2]. Since we decided to remove the flaky tests and
> proceed to
>     > > making them stable I propose completely removing this functionality
>     > > instead.
>     > >
>     > > There are multiple reasons to this:
>     > >
>     > > * Mind that this is Caffe 1 (not 2)
>     > > * Some people mentioned: "Caffe is soooooo 2015."
>     > > * Keeping functionality that is both unstable and old is a burden
> for
>     > > maintenance.
>     > > * Keeping functionality that nobody needs is not necessary overall
>     > >
>     > > Please let me know your thoughts.
>     > >
>     > > Best
>     > > Anton
>     > >
>     > > [1]
>     > > https://github.com/apache/incubator-mxnet/commits/
>     > > master/tools/caffe_converter/convert_caffe_modelzoo.py
>     > > [2]
>     > > http://jenkins.mxnet-ci.amazon-ml.com/blue/organizations/jenkins/
>     > > incubator-mxnet/detail/master/1207/pipeline/
>     > >
>     >
>
>
>
>

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