Hi Guchun
-text-type is connected with some experimental domain adaptation features.
-t is an abbreviation for -report-segmentation
cheers - Barry
On 10/11/12 09:38, Guchun Zhang wrote:
Hi Barry,
Just a quick question that what -t and -text-type in the decoder are for?
Many thanks,
Thanks, Barry.
On 12 November 2012 09:24, Barry Haddow bhad...@staffmail.ed.ac.uk wrote:
Hi Guchun
-text-type is connected with some experimental domain adaptation features.
-t is an abbreviation for -report-segmentation
cheers - Barry
On 10/11/12 09:38, Guchun Zhang wrote:
Hi Barry,
Hi Hieu,
I have added mgizapp (MGIZA++) in MacPorts, and a port for GIZA++ is not gizapp
but giza-pp. There is a notational inconsistency.
If you try to use mgizapp or giza-pp, please run the following commands:
$ sudo port selfupdate
$ sudo port install mgizapp
or
$ sudo port install giza-pp
Hi all,
These days I've been stuck with a very fuzzy error.
I've been coding an IBM models training toolkit (1-3) and now finishing IBM
Model 4.
The output of my toolkit is exactly as the MOSES specification as
described: http://www.statmt.org/moses/?n=FactoredTraining.RunGIZA
For example:
#
Dear everyone,
moses_chart segfaults for me if -n-best-list is given, right before
generating the n-best list. Log output below. Without n-best list
generation translation works fine. Does anyone have the same
experience? Tree-to-tree translation.
Thanks and best regards,
Mark
.
Dear Prof. Koehn,
The fuzzy thing is there is no mismatch between those files!
The only mismatch is the output of the error!
!alignment point (42,23) out of range (0-31,0-25) in line 1, ignoring
alignment point (33,24) out of range (0-31,0-25) in line 1, ignoring
alignment point (28,26) out of
Hi,
one (not completely satisfying) solution is to break up
the corpus and run MGIZA++ separately on each part.
-phi
On Mon, Nov 12, 2012 at 4:56 AM, Jelita Asian
jelitay...@stkipsurya.ac.id wrote:
Hi Barry,
Actually how do we solve the more than 2 GB memory problem? Thanks.
Best regards,
Maybe it because of the mismatch since I don't know French.
I will check it,
tks!
On Tue, Nov 13, 2012 at 12:06 AM, Cuong Hoang hoangcuong2...@gmail.comwrote:
Dear Prof. Koehn,
The fuzzy thing is there is no mismatch between those files!
The only mismatch is the output of the error!
Hi Jelita,
mgiza claims to support native Windows via visual studio but it appears
to only have been tested with 32-bit. You're on your own to try and get
it working in 64-bit windows and to integrate it into the scripts.
Kenneth
On 11/12/12 16:09, Philipp Koehn wrote:
Hi,
one
hi mark
do you have an example model i can run to debug? I'm running hiero models
at the moment and they're doing ok
On 12 November 2012 16:04, Mark Fishel fis...@ut.ee wrote:
Dear everyone,
moses_chart segfaults for me if -n-best-list is given, right before
generating the n-best list. Log
I don't know if Berkeley Aligner works under Windows, but since it's
written in Java I strongly suspect that it would. If so, you could try
doing the word alignment with it instead of mgiza.
Cheers,
Lane
On Mon, Nov 12, 2012 at 7:56 AM, Jelita Asian
jelitay...@stkipsurya.ac.id wrote:
Hi Barry,
Hi Barry,
While installing IRSTLM file version irstlm-5.70.04.
the command line ./regenerate-makefiles.sh generated the error below
mrodoje@ubuntu:~/irstlm-5.70.04$ ./regenerate-makefiles.sh
CallingĀ
Calling ...
./regenerate-makefiles.sh: line 52: -I: command not found aclocal failed
what do
MGIZA++ can be compiled using VC++ both 32 and 64bit, however 64bit version
occasionally crashes during final clean up. So YMMV. Also, I believe
mingw-w64 is out there, maybe you can try that (msys + mingw-w64)
http://stackoverflow.com/questions/9942923/mingw-as-a-reliable-64-bit-gcc-compiler
Just now, I cut it down to 4 libraries produced by Moses:
moses mert_lib mira_lib pcfg_common
For decoding, then you should only need to link against moses.
Kenneth
On 11/07/12 23:30, Wilker Aziz wrote:
Thanks Ken!
Wilker
On Nov 7, 2012 11:32 PM, Kenneth Heafield mo...@kheafield.com
on cygwin, there is no way to solve the problem. 2GB is the maximum any
process can use on cygwin.
If you have large model files, you should move to 64-bit linux or
macOSX, with plenty of memory.
Or try and compile mgiza and moses without cygwin. For example, use
mingw or visual studio.
Hi Clement,
The installation instructions for IRSTLM can be found here:
http://sourceforge.net/apps/mediawiki/irstlm/index.php?title=Installation_Guidelines
also, here is the instruction manual:
http://sourceforge.net/apps/mediawiki/irstlm/index.php?title=User_Manual
First, try calling the
Thanks for everyone who answer! I guess the solution is either to break up
the corpus or run it using mingw-w64. Thanks.
Best regards,
Jelita
On Tue, Nov 13, 2012 at 6:08 AM, Hieu Hoang hieuho...@gmail.com wrote:
on cygwin, there is no way to solve the problem. 2GB is the maximum any
17 matches
Mail list logo