Moses-support Digest, Vol 109, Issue 61

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. Adding feature to moses (haouarid@iro.umontreal.ca)
2. ems: PhraseDictionaryOnDisk binarization (Nicholas Ruiz)
3. continue partial translation (He He)
4. how to cross compile moses binary on Raspberry Pi 2 model B
(Apurva Joshi)
5. Re: ems: PhraseDictionaryOnDisk binarization (Barry Haddow)


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

Message: 1
Date: Tue, 24 Nov 2015 17:46:53 -0500
From: haouarid@iro.umontreal.ca
Subject: [Moses-support] Adding feature to moses
To: moses-support@mit.edu
Message-ID:
<b44e69f3c64e13e95382b91df4e740b8.squirrel@webmail.iro.umontreal.ca>
Content-Type: text/plain;charset=iso-8859-1

Hi,

I would like to add a word by word translation model to moses as a new
feature. I didn't understand how it is possible to have access to the
source and target words. As it is explained in the documentation,
StringPiece is accessible by Factor class which is accessible by Word
class which is accessible by TargetPhrase which is given as argument in
the evaluation method like EvaluateWithSourceContext. I don't use factors
with my translation model, I just use surface form of words.
How is it possible to read the string source and target words without
using Factors?

Thank you,
Dorra


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

Message: 2
Date: Tue, 24 Nov 2015 20:44:34 -0500
From: Nicholas Ruiz <nicruiz@fbk.eu>
Subject: [Moses-support] ems: PhraseDictionaryOnDisk binarization
To: moses-support@mit.edu
Message-ID:
<CAKa+0YPr2Cko+4DnoEdQn81-3Er-SzxUK6JP72tZN9+u+agScg@mail.gmail.com>
Content-Type: text/plain; charset="utf-8"

Hi all,

I'm a bit behind on my moses versions and I'm using EMS for the first time.
I trained a toy model, which gave me a PhraseTableMemory translation table.
I'd like to binarize the phrase table and reordering models. I'm still
operating back in the PhraseTableBinary days, but obviously the codebase
has changed quite a bit since then.

How do I binarize the phrase table as a PhraseDictionaryOnDisk? I had tried
uncommenting the binarize-all setting, but that gave me
a PhraseDictionaryBitextSampling. However, I don't need to do incremental
training -- and the tuning phase is failing anyway. Help would be
appreciated about how to do a simple binarization in EMS.

Thanks!
Nick

z?n? ???u
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/mailman/private/moses-support/attachments/20151124/31701d7b/attachment-0001.html

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

Message: 3
Date: Tue, 24 Nov 2015 22:19:45 -0500
From: He He <hhe.xiy@gmail.com>
Subject: [Moses-support] continue partial translation
To: moses-support@mit.edu
Message-ID:
<CAMdMQUNJWzeYG=wZat08aTJGM7rNx+7u1+UEQe29bYErFbkczA@mail.gmail.com>
Content-Type: text/plain; charset="utf-8"

Hi there,

I'm trying to do translation conditioned on some already translated prefix
(essentially what -continue-partial-translation was supposed to do). I'm
using -xml-input exclusive to pass in the prefix source and translation.

However, when the prefix becomes long, this doesn't work, e.g.
<p translation="Britain 's trade house E D & F Man said on the amount of
money in eastern europe , sugar beet output both Ukraine and Russia in"> ??
? ED & F ?? ? ? ?? ? , 96 / 97 ?? ? ?? ? ??? ?? ? , ????? ???? ? ?? ??
???</p> ?? ? ?? ? ?? ? ? , ??? ?? ? ??> 0 ||| ?? ? ED & F?? ? ED & F ?? ? ?
?? ? / 96 , 97 ?? ? ?? ? ??? ?? ? ????? , ? ??? ? ?? ?? ??? substantial
decline was expeted to be tough ||| LexicalReordering0= -4.48185 -7.01678
-1.48808 -4.3759 -6.89465-0.942918 Distortion0= -12 LM0= -227.918
WordPenalty0= -40 PhrasePenalty0= 36 TransltionModel0= -7.25771 -34.5474
-2.80336 -22.3651 ||| -3322.64"

It just copies the source prefix. I suspect it's because many words now
becomes UNK due to ignoring entries in phrase table that overlaps the
prefix.

Is there a way around this? Thanks a lot in advance!

Best,
He
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/mailman/private/moses-support/attachments/20151124/9d85bffa/attachment-0001.html

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

Message: 4
Date: Wed, 25 Nov 2015 13:57:27 +0530
From: Apurva Joshi <apurvajoshi1992@gmail.com>
Subject: [Moses-support] how to cross compile moses binary on
Raspberry Pi 2 model B
To: moses-support@mit.edu
Message-ID:
<CAMfCXbXiy6jAie0PcYji1bt7Mp5pjNrstTCi2pc-PXt0Tq4pEw@mail.gmail.com>
Content-Type: text/plain; charset="utf-8"

hello all,

i run moses on pc which does not contain moses successfully , by copying

moses.ini
phrase table
reordering table
LM
moses( binary from ~/mosesdecoder/bin/moses)

i copied above files to pc which does not contain moses

and then on that pc i ran

moses -f moses.ini

its worked...

NOW

same procedure i used for running on raspberry pi..

on raspberry pi board , i copied all above 5 files

and ran moses -f moses.ini

its showing errror that "




*cant run binary"-bash: /home/ubuntu/Apurva/moses: cannot execute binary
file: Exec format error*
so i need to cross-compile moses (binary) for ARM on raspberry pi board

i dont know how to do cross compilation of binary for ARM

plz help and reply on same id

apurvajoshi1992@gmail.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/mailman/private/moses-support/attachments/20151125/d86da7a5/attachment-0001.html

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

Message: 5
Date: Wed, 25 Nov 2015 09:13:29 +0000
From: Barry Haddow <bhaddow@staffmail.ed.ac.uk>
Subject: Re: [Moses-support] ems: PhraseDictionaryOnDisk binarization
To: Nicholas Ruiz <nicruiz@fbk.eu>, moses-support@mit.edu
Message-ID: <56557BB9.9000709@staffmail.ed.ac.uk>
Content-Type: text/plain; charset="utf-8"

Hi Nick

The best solution is to use the compact phrase table, and for this just add

ttable-binarizer = $moses-bin-dir/processPhraseTableMin

to the general section.

If you need to use the ondisk phrase table (sparse features, properties
etc.) then replace the above with

ttable-binarizer = "$moses-bin-dir/CreateOnDiskPt 1 1 4 100 2"

where the "4" is the number of dense features in your phrase table.

It's a bit strange that binarize-all gives you
PhraseDictionaryBitextSampling, maybe that's the default these days?
Anyway, if you just want to decode a dev & test set then you don't want
to set binarize-all

cheers - Barry



On 25/11/15 01:44, Nicholas Ruiz wrote:
> Hi all,
>
> I'm a bit behind on my moses versions and I'm using EMS for the first
> time. I trained a toy model, which gave me a PhraseTableMemory
> translation table. I'd like to binarize the phrase table and
> reordering models. I'm still operating back in the PhraseTableBinary
> days, but obviously the codebase has changed quite a bit since then.
>
> How do I binarize the phrase table as a PhraseDictionaryOnDisk? I had
> tried uncommenting the binarize-all setting, but that gave me
> a PhraseDictionaryBitextSampling. However, I don't need to do
> incremental training -- and the tuning phase is failing anyway. Help
> would be appreciated about how to do a simple binarization in EMS.
>
> Thanks!
> Nick
>
> z?n? ???u
>
>
> _______________________________________________
> 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/20151125/5eb2de4d/attachment.html
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: not available
Url: http://mailman.mit.edu/mailman/private/moses-support/attachments/20151125/5eb2de4d/attachment.pl

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

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


End of Moses-support Digest, Vol 109, Issue 61
**********************************************

Related Posts :

0 Response to "Moses-support Digest, Vol 109, Issue 61"

Post a Comment