Moses-support Digest, Vol 100, Issue 27

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: Native 64-bit Windows support (Tom Hoar)
2. Error while compile with nplm (Jianri Li)
3. optimizing lattice InputFeature weight (Jorg Tiedemann)
4. Re: optimizing lattice InputFeature weight (Hieu Hoang)


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

Message: 1
Date: Sun, 08 Feb 2015 00:24:34 +0700
From: Tom Hoar <tahoar@precisiontranslationtools.com>
Subject: Re: [Moses-support] Native 64-bit Windows support
To: moses-support@mit.edu
Message-ID: <54D64A52.8010109@precisiontranslationtools.com>
Content-Type: text/plain; charset="windows-1252"

Thanks. Yes, I know there many more. As you see, we're moving forward
with our Windows development. These are simply the ones we're
considering supporting on native Windows. I know you maintained some
support for a while. It looks like it has fallen behind. Thanks for the
offer for PhraseDictionaryOnDIsk. Let us dig a little deeper before you
burn hours on it.


On 02/08/2015 12:16 AM, Hieu Hoang wrote:
> hi tom
>
> Hieu Hoang
> Research Associate (until March 2015)
> ** searching for interesting commercial MT position **
> University of Edinburgh
> http://www.hoang.co.uk/hieu
>
>
> On 7 February 2015 at 16:33, Tom Hoar
> <tahoar@precisiontranslationtools.com
> <mailto:tahoar@precisiontranslationtools.com>> wrote:
>
> The most recent information I can find shows that following FF's were
> available:
>
> PhraseDictionaryMemory
> PhraseDictionaryBinary
> PhraseDictionaryOnDisk
> LexicalReordering
> KENLM
>
> There are many FF. Run the decoder without any arguments and you're
> get a list of them at the end.
>
> In R-3, processPhraseTable has been removed and PhraseDictionaryBinary
> been earmarked for end-of-life. Does R-3 include native Windows
> support
> for compact phrase and compact reordering formats? A quick review
> of the
> source shows they might not support native Win64. I also seem to
> remember some exchanges a few weeks ago about the compact
> phrase/reordering code needs major updates.
>
> If this is correct, what's the status on that and who's doing the
> work?
> We'd like to help to make sure we add Win64 support.
>
> The whole decoder isn't Win64 compatible. I maintained compatibility
> for years but it requires frequent small changes to the code. The pace
> and diversity of commits is such that I've given up. It needs a small
> community of WIndows users to chip in.
>
> If you're willing to maintain it now and in the future, I'll make my
> PhraseDictionaryOnDIsk support windows. Otherwise, just use cygwin/mingw
>
>
> _______________________________________________
> 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/20150208/e807faf2/attachment-0001.htm

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

Message: 2
Date: Sun, 08 Feb 2015 03:22:29 +0900
From: Jianri Li <skywalker@postech.ac.kr>
Subject: [Moses-support] Error while compile with nplm
To: " moses-support " <moses-support@mit.edu>
Message-ID: <1423333349846.79166.postech@postech.ac.kr>
Content-Type: text/plain; charset="us-ascii"

An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/mailman/private/moses-support/attachments/20150208/45b356ab/attachment-0001.htm

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

Message: 3
Date: Sat, 7 Feb 2015 21:05:25 +0100
From: Jorg Tiedemann <tiedeman@gmail.com>
Subject: [Moses-support] optimizing lattice InputFeature weight
To: moses-support <moses-support@mit.edu>
Message-ID: <AC6F12F5-B37B-4596-AF29-6CAA9652E72E@gmail.com>
Content-Type: text/plain; charset=iso-8859-1


I have a problem with lattice decoding and optimizing input-feature weights. I have edge weights in my lattice input (one per edge) and I defined one input feature that I'd like to optimize using MERT. However, my the input feature value is always 0 in the nbest lists even though none of the input edges has value 1 (or 0). What do I do wrong?

My initial config file includes:

[feature]
InputFeature num-features=1 num-input-features=1 real-word-count=0
...
# dense weights for feature functions
[weight]
InputFeature0= 1

The lattice input is valid and looks like this:

((('word',0.8,1),('word',0.6,1), ....

MERT tuning fails in the end especially because the input feature cannot be set.
Any help is very much appreciated.
Thanks,
J?rg





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

Message: 4
Date: Sat, 7 Feb 2015 20:21:46 +0000
From: Hieu Hoang <Hieu.Hoang@ed.ac.uk>
Subject: Re: [Moses-support] optimizing lattice InputFeature weight
To: Jorg Tiedemann <tiedeman@gmail.com>
Cc: moses-support <moses-support@mit.edu>
Message-ID:
<CAEKMkbi5oGDR4FtXg843B_OYSw-S_83E6Cq-fbr27s+w+Yr6xA@mail.gmail.com>
Content-Type: text/plain; charset="utf-8"

there's no reason why it shouldn't work.

the only thing i can think of is that the input type hasn't been set to
lattice. In the moses.ini, there should be something like
[inputtype]
2
or on the command line
moses -inputtype 2


Hieu Hoang
Research Associate (until March 2015)
** searching for interesting commercial MT position **
University of Edinburgh
http://www.hoang.co.uk/hieu


On 7 February 2015 at 20:05, Jorg Tiedemann <tiedeman@gmail.com> wrote:

>
> I have a problem with lattice decoding and optimizing input-feature
> weights. I have edge weights in my lattice input (one per edge) and I
> defined one input feature that I'd like to optimize using MERT. However, my
> the input feature value is always 0 in the nbest lists even though none of
> the input edges has value 1 (or 0). What do I do wrong?
>
> My initial config file includes:
>
> [feature]
> InputFeature num-features=1 num-input-features=1 real-word-count=0
> ...
> # dense weights for feature functions
> [weight]
> InputFeature0= 1
>
> The lattice input is valid and looks like this:
>
> ((('word',0.8,1),('word',0.6,1), ....
>
> MERT tuning fails in the end especially because the input feature cannot
> be set.
> Any help is very much appreciated.
> Thanks,
> J?rg
>
>
>
> _______________________________________________
> 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/20150207/ea9f15a7/attachment.htm

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

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


End of Moses-support Digest, Vol 100, Issue 27
**********************************************

Tweet