Moses-support Digest, Vol 113, Issue 20

Send Moses-support mailing list submissions to
moses-support@mit.edu

To subscribe or unsubscribe via the World Wide Web, visit
http://mailman.mit.edu/mailman/listinfo/moses-support
or, via email, send a message with subject or body 'help' to
moses-support-request@mit.edu

You can reach the person managing the list at
moses-support-owner@mit.edu

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Moses-support digest..."


Today's Topics:

1. Re: Scripts for n-best-list rescoring (Marcin Junczys-Dowmunt)
2. philosophical question ....NMT/SMT (Vincent Nguyen)


----------------------------------------------------------------------

Message: 1
Date: Tue, 8 Mar 2016 16:04:41 +0000
From: Marcin Junczys-Dowmunt <junczys@amu.edu.pl>
Subject: Re: [Moses-support] Scripts for n-best-list rescoring
To: moses-support@mit.edu
Message-ID: <56DEF819.60002@amu.edu.pl>
Content-Type: text/plain; charset="windows-1252"

Great. Thanks for that.

W dniu 08.03.2016 o 16:02, Michael Denkowski pisze:
> I recently checked in the N-best re-scorer I wrote since I couldn't
> find a good existing one either:
> https://github.com/moses-smt/mosesdecoder/tree/master/scripts/nbest-rescore.
> Given an N-best list and references, it uses K-best MIRA to learn
> re-ranking weights relatively quickly. It's agnostic to original
> decoder and optimizer as long as entries are in the right format. The
> readme includes more detailed instructions.
>
> Best,
> Michael
>
> On Tue, Mar 8, 2016 at 8:25 AM, Philipp Koehn <phi@jhu.edu
> <mailto:phi@jhu.edu>> wrote:
>
> Hi,
>
> there is this mysterious check-in:
>
> Commit: c6314d927d8b7b638eca387f31ccfab7facb6624
> https://github.com/moses-smt/mosesdecoder/commit/c6314d927d8b7b638eca387f31ccfab7facb6624
> Author: Michael Denkowski <mdenkows@amazon.com
> <mailto:mdenkows@amazon.com>>
> Date: 2016-02-23 (Tue, 23 Feb 2016)
>
> Changed paths:
> A scripts/nbest-rescore/README.md
> A scripts/nbest-rescore/rescore.py
> A scripts/nbest-rescore/topbest.py
> A scripts/nbest-rescore/train.py
>
> -phi
>
> On Tue, Mar 8, 2016 at 8:18 AM, Lane Schwartz <dowobeha@gmail.com
> <mailto:dowobeha@gmail.com>> wrote:
>
> I don't think there is. At my previous lab, I believe we had
> to build our own in-house script. It would be nice to have one
> in moses.
>
> On Sat, Oct 31, 2015 at 12:56 PM, Marcin Junczys-Dowmunt
> <junczys@amu.edu.pl <mailto:junczys@amu.edu.pl>> wrote:
>
> Hi,
> does moses include scripts for n-best-list
> rescoring/resorting after a
> new feature has been added to the list?
>
> I guess, this can probably be achieved by running a single
> parameter
> tuning step on the extended n-best-list, but then I still
> need to fiddle
> around with calculating model scores with the new weights
> etc. Is there
> anything public and working with the moses n-best-list format?
>
> Cheers,
> Marcin
> _______________________________________________
> Moses-support mailing list
> Moses-support@mit.edu <mailto:Moses-support@mit.edu>
> http://mailman.mit.edu/mailman/listinfo/moses-support
>
>
>
>
> --
> When a place gets crowded enough to require ID's, social
> collapse is not
> far away. It is time to go elsewhere. The best thing about
> space travel
> is that it made it possible to go elsewhere.
> -- R.A. Heinlein, "Time Enough For Love"
>
> _______________________________________________
> Moses-support mailing list
> Moses-support@mit.edu <mailto:Moses-support@mit.edu>
> http://mailman.mit.edu/mailman/listinfo/moses-support
>
>
>
> _______________________________________________
> Moses-support mailing list
> Moses-support@mit.edu <mailto:Moses-support@mit.edu>
> http://mailman.mit.edu/mailman/listinfo/moses-support
>
>
>
>
> _______________________________________________
> Moses-support mailing list
> Moses-support@mit.edu
> http://mailman.mit.edu/mailman/listinfo/moses-support

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/mailman/private/moses-support/attachments/20160308/157c9062/attachment-0001.html

------------------------------

Message: 2
Date: Tue, 8 Mar 2016 17:49:24 +0100
From: Vincent Nguyen <vnguyen@neuf.fr>
Subject: [Moses-support] philosophical question ....NMT/SMT
To: moses-support <moses-support@mit.edu>
Message-ID: <56DF0294.8020006@neuf.fr>
Content-Type: text/plain; charset=utf-8; format=flowed

Guys,

I got a question to the mathematicians that you all are :)

I have been working and testing Moses as well as Groundhog for months now.
When I compare results (when comparability is possible, using same
corpus, in-domain, blablabla, ...) I do not see much difference in both
systems.

So when I read the statement below (from :
http://www.cis.uni-muenchen.de/~fraser/intensive_nmt_2015/)

I am confused. Is this really a new paradigm?

I have the feeling that at the end of the day, it ends up with
probabilities to get a word after another word after another word, given
some sequence input.
Whether machine learning gets deeply neural or in huge tables, don't we
get the same results if the concepts are similar ?


I would much appreciate you're feedback.
Cheers,
Vincent



Neural Machine Translation (NMT) is a new paradigm in data-driven
machine translation. Previous generation Statistical Machine Translation
(SMT) systems are built using a collection of heuristic models,
typically combined in a log-linear model with a small number of
parameters. In Neural Machine Translation, the entire translation
process is posed as an end-to-end supervised classification problem,
where the training data is pairs of sentences. While in SMT systems,
word-alignment is carried out, and then fixed, and then various
sub-models are estimated from the word-aligned data, this is not the
case in NMT. In NMT, fixed word-alignments are not used, and instead the
full sequence to sequence task is handled in one model.

The course will work backwards from the current state of the art in NMT,
which is the "ensemble" system submitted by the Bengio group in Montreal
to the 2015 shared task on machine translation (Jean et al. 2015, see
below, with some additional details to be published). Depending on the
background of the participants, some basics of SMT may also be covered.


------------------------------

_______________________________________________
Moses-support mailing list
Moses-support@mit.edu
http://mailman.mit.edu/mailman/listinfo/moses-support


End of Moses-support Digest, Vol 113, Issue 20
**********************************************

Tweet