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. Formatting source code (Jeroen Vermeulen)
2. Re: Formatting source code (Hieu Hoang)
3. Re: Formatting source code (Jeroen Vermeulen)
4. mgiza crashes when using Incremental training with baseline
model (Hegde, Sujay)
----------------------------------------------------------------------
Message: 1
Date: Fri, 15 May 2015 10:19:57 +0700
From: Jeroen Vermeulen <jtv@precisiontranslationtools.com>
Subject: [Moses-support] Formatting source code
To: "moses-support@mit.edu" <moses-support@mit.edu>
Message-ID: <555565DD.4070107@precisiontranslationtools.com>
Content-Type: text/plain; charset=utf-8
Would anyone mind if I gave beautify.pl some love?
I was thinking to:
* Re-do it in python (because I find it too easy to mess up in Perl).
* Locate source files to reformat based on a config file.
* Apply the same thing to mgiza++, so it's all one style.
If this sounds like yak-shaving, that's because it is. I started out by
fixing some warnings for reordered constructor initializers in mgiza++,
which was hard because of the formatting, but I couldn't run beautify.pl
because it's tied to Moses, which in turn... etc.
Jeroen
------------------------------
Message: 2
Date: Fri, 15 May 2015 07:33:49 +0400
From: Hieu Hoang <hieuhoang@gmail.com>
Subject: Re: [Moses-support] Formatting source code
To: Jeroen Vermeulen <jtv@precisiontranslationtools.com>
Cc: moses-support <moses-support@mit.edu>
Message-ID:
<CAEKMkbhzzKW9_x+ZvBUXLafZT+Gh42ph7D9Xzy+=+ag6ZDuMAw@mail.gmail.com>
Content-Type: text/plain; charset="utf-8"
Be my guest
On 15 May 2015 07:20, "Jeroen Vermeulen" <jtv@precisiontranslationtools.com>
wrote:
> Would anyone mind if I gave beautify.pl some love?
>
> I was thinking to:
>
> * Re-do it in python (because I find it too easy to mess up in Perl).
> * Locate source files to reformat based on a config file.
> * Apply the same thing to mgiza++, so it's all one style.
>
> If this sounds like yak-shaving, that's because it is. I started out by
> fixing some warnings for reordered constructor initializers in mgiza++,
> which was hard because of the formatting, but I couldn't run beautify.pl
> because it's tied to Moses, which in turn... etc.
>
>
> Jeroen
> _______________________________________________
> 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/20150515/cff85f89/attachment-0001.htm
------------------------------
Message: 3
Date: Fri, 15 May 2015 13:24:32 +0700
From: Jeroen Vermeulen <jtv@precisiontranslationtools.com>
Subject: Re: [Moses-support] Formatting source code
To: Hieu Hoang <hieuhoang@gmail.com>
Cc: moses-support <moses-support@mit.edu>
Message-ID: <55559120.1060709@precisiontranslationtools.com>
Content-Type: text/plain; charset=utf-8
On 15/05/15 10:33, Hieu Hoang wrote:
> Be my guest
Thanks. As a bonus, it'll be really easy to clean up trailing
whitespace in other types of files besides C/C++ source files. Provided
that we know it's absolutely definitely safe for those file types, of
course.
Any suggestions?
Jeroen
------------------------------
Message: 4
Date: Fri, 15 May 2015 06:40:38 +0000
From: "Hegde, Sujay" <Sujay.Hegde@xerox.com>
Subject: [Moses-support] mgiza crashes when using Incremental training
with baseline model
To: "moses-support@mit.edu" <moses-support@mit.edu>
Cc: "Venkatapathy, Sriram \(Calendar\)"
<sriram.venkatapathy@xrce.xerox.com>, "MudaliarMudaliar, Preeti J"
<preeti.mudaliarmudaliar@xerox.com>
Message-ID:
<586EA7C483504E48870F5BF54319B6EC39882D@USA7109MB006.na.xerox.net>
Content-Type: text/plain; charset="us-ascii"
HI,
We are getting following error in Giza training:
Starting MGIZA
Initializing Global Paras
DEBUG: EnterDEBUG: PrefixDEBUG: LogParsing Arguments
ERROR: parameter 'oldalprbs' does not exist.
WARNING: ignoring unrecognized option: -oldAlPrbs
ERROR: parameter 'mnthd1trainingworkingdirenestraininggiza1eneshhmm5' does not exist.
WARNING: ignoring unrecognized option: /mnt/hd1/training/working-dir-en-es/training/giza.1/en-es.hhmm.5
ERROR: parameter 'oldtrprbs' does not exist.
WARNING: ignoring unrecognized option: -oldTrPrbs
ERROR: parameter 'mnthd1trainingworkingdirenestraininggiza1enesthmm5' does not exist.
WARNING: ignoring unrecognized option: /mnt/hd1/training/working-dir-en-es/training/giza.1/en-es.thmm.5
ERROR: parameter 'stepk' does not exist.
WARNING: ignoring unrecognized option: -step_k
ERROR: parameter '1' does not exist.
WARNING: ignoring unrecognized option: 1
We initially ran Giza for a small corpus using EMS and generated a baseline model.The initial training was run with:
training-options = "-final-alignment-model hmm"
Then we ran Giza again with baseline model params set appropriately.Something similar to:
### use of baseline alignment model (incremental training)
#
baseline = 1
baseline-alignment-model = "$working-dir/training/prepared.$baseline/$input-extension.vcb \
$working-dir/training/prepared.$baseline/$output-extension.vcb \
$working-dir/training/giza.$baseline/${output-extension}-$input-extension.cooc \
$working-dir/training/giza-inverse.$baseline/${input-extension}-$output-extension.cooc \
$working-dir/training/giza.$baseline/${output-extension}-$input-extension.thmm.5 \
$working-dir/training/giza.$baseline/${output-extension}-$input-extension.hhmm.5 \
$working-dir/training/giza-inverse.$baseline/${input-extension}-$output-extension.thmm.5 \
$working-dir/training/giza-inverse.$baseline/${input-extension}-$output-extension.hhmm.5
But Giza crashes on 2nd run with the error as pasted above.
We referred the below link which reports a similar problem:
http://moses-support.mit.narkive.com/xXEngy5d/incremental-training
But it makes references to many links like: http://geek.kyloo.net/software/doku.php/mgiza:forcealignment
That does not open.
Please let us know what could be the problem.
We have worked on it for quite long.
Are we missing something in EMS configuration(training-options in TRAINING section) Or we need to explicitly change MGiza configurations.Or we need to use a new version of MGiza?
Regards,
Sujay
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/mailman/private/moses-support/attachments/20150515/b08529a5/attachment.htm
------------------------------
_______________________________________________
Moses-support mailing list
Moses-support@mit.edu
http://mailman.mit.edu/mailman/listinfo/moses-support
End of Moses-support Digest, Vol 103, Issue 36
**********************************************
Subscribe to:
Post Comments (Atom)
0 Response to "Moses-support Digest, Vol 103, Issue 36"
Post a Comment