Thanks,
> > Lane
> >
> >
> >
> > _______
> > Moses-support mailing list
> > Moses-support@mit.edu
> > http://mailman.mit.edu/mailman/listinfo/moses-support
> >
> >
UVD*
If you have any enquiries, please contact *Stephanie Kyle (
stephaniek...@wlv.ac.uk)* or *Emma Franklin (emma.frank...@wlv.ac.uk)*.
--
Wilker Aziz
http://pers-www.wlv.ac.uk/~in1676/
PhD candidate at The Research Group in Computational Linguistics
Research Institute of Information and
t; much, but I've never seen the evidence. If you want to verify this, I've
> created a phrase-penalty feature that you can use
>
> https://github.com/moses-smt/mosesdecoder/commit/e15a4fc882952be13efcdecc8284d19560229785
>
> On 24 June 2013 22:16, Wilker Aziz wrote:
&
Hello everybody,
I would like to share with you the results of a de-en hierarchical model
trained using this year's WMT constrained data. This model was trained
using Adam Lopez's hierarchical suffix arrays.
I patched some wrappers so hopefully anyone will be able to train such a
model using EMS n
S 20G"
# order of the language model
order = 3
Cheers,
Wilker.
On 24 June 2013 11:12, Philipp Koehn wrote:
> Hi,
>
> you will have to write a wrapper script, along the lines of
> scripts/generic/trainlm-irst2.perl
>
> -phi
>
> On Mon, Jun 24, 2013 at 11:01 AM, Wil
Hi all,
there is a small problem in using lmplz with EMS, its interface is
different from SRILM.
>From experiments.meta:
train
| in: split-corpus
| out: lm
| default-name: lm/lm
| ignore-if: rlm-training
| rerun-on-change: lm-training order settings
| template: $lm-training *-order* $
s_chart -f moses.filtered.ini.1 -i input.split.1 >& out
> let me know if it works for you
>
> it'll be good if you can push your EMS changes when you're ready
>
>
>
--
Wilker Aziz
http://pers-www.wlv.ac.uk/~in1676/
PhD candidate at The Research Group in Computa
/adam-suffix-array/suffix-array-extract.sh
cat $INPUT_FILE | python -m cdec.sa.extract -c $MODEL_DIR/extract.ini -g
$OUTPUT_DIR -j $njobs -z
Cheers,
Wilker
On 28 May 2013 13:13, Hieu Hoang wrote:
> hi wilker
>
>
> On 28 May 2013 12:53, Wilker Aziz wrote:
>
>> Hi Hieu,
I can certainly update Moses to
>> get it to work
>>
>>
>>
>>
>> On 28 May 2013 12:32, Wilker Aziz wrote:
>>
>>> Hi,
>>>
>>> I've been trying to train a hierarchical model using suffix arrays.
>>> The current versio
Hi,
I've been trying to train a hierarchical model using suffix arrays.
The current version of the training script declares the following feature
function to handle the weights of the "rule-table" (or the SCFG extracted
for each sentence individually).
*UnknownPtImplementation name=TranslationMod
proposal acceptance notification August 9, 2013: Demo
paper camera-ready version due October 15-17: Conference
6. Demonstrations Chairs
Kentaro Torisawa (NICT, Japan)
Hang Li (Huawei Technologies, Hong Kong) Any inquiry should be emailed to
d...@ijcnlp2013.org.
--
Wilker Aziz
http://pers-www.
pus. Can anyone comment a bit more on this?
>
> Thanks,
> Prashant
>
> ___
> Moses-support mailing list
> Moses-support@mit.edu
> http://mailman.mit.edu/mailman/listinfo/moses-support
>
--
Wilker Aziz
http://pers-www.wlv.ac.uk/~in1676/
PhD candidate at The Research Group
Workshops
Oct 18-19, 2013: Post-Conference Workshops
Workshop Co-Chairs
Naoaki Okazaki (Tohoku University, Japan)
Scott Wen-tau Yih (Microsoft Research, USA)
--
Wilker Aziz
http://pers-www.wlv.ac.uk/~in1676/
PhD candidate at The Research Group in Computational Linguistics
Research
)
Rovira i Virgili University
Av. Catalunya, 35
43002 Tarragona, Spain
Phone: +34-977-559543
Fax: +34-977-558386
ACKNOWLEDGEMENTS:
Diputació de Tarragona
Universitat Rovira i Virgili
University of Wolverhampton
--
Wilker Aziz
http://pers-www.wlv.ac.uk/~in1676/
PhD candidate at The
fford St.
Wolverhampton WV1 1SB
Telephone + 44 1902 322623
Fax +44 1902 323543
Email: n.konstantin...@wlv.ac.uk
--
Wilker Aziz
http://pers-www.wlv.ac.uk/~in1676/
PhD candidate at The Research Group in Computational Linguistics
Research Institute of Information and Language Proces
phone + 44 1902 322623
Fax +44 1902 323543
Email:
n.konstantin...@wlv.ac.uk<https://exchcas.unv.wlv.ac.uk/owa/redir.aspx?C=uwqwnllH40mKHtVVwGpTt6iLIDA1rc8IzjmanDmkMGYhvLVmkXyz1uCoRfhXBS5edn5yoiYyEXQ.&URL=mailto%3an.konstantinova%40wlv.ac.uk>
--
Wilker Aziz
http://pers-www.wlv.ac.uk/~in1676/
PhD cand
uistics (GRLMC)
Rovira i Virgili University
Av. Catalunya, 35
43002 Tarragona, Spain
Phone: +34‐977‐559543
Fax: +34‐977‐558386
*ACKNOWLEDGEMENTS:*
Diputació de Tarragona
Universitat Rovira i Virgili
University of Wolverhampton
--
Wilker Aziz
http://pers-www.wlv.ac.uk/~in1676/
PhD c
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" > <mailto:mo...@kheaf
y, so
> I have added those.
>
> Now works for me.
>
> Kenneth
>
> On 11/07/12 19:51, Wilker Aziz wrote:
> > Hello,
> >
> > I am not yet familiar with bjam, so please excuse me if I am missing
> > something obvious.
> > I am trying to compile Moses with *l
Hello,
I am not yet familiar with bjam, so please excuse me if I am missing
something obvious.
I am trying to compile Moses with *link=shared* and I am getting the error
below.
In case it matters, I managed to compile with the very same command line
some revisions ago.
./bjam -a -j4 link=shared
w
ration, and also give me some tips to get better results in
>> the
>> > translation of pt->zn and zn->pt?
>> >
>> >
>> > Best regards!
>> > Nelson from Portugal.
>> >
>> > ___
>> &
good, and the chinese corpus I just
> tokenized and cleaned it.
>
>
> (Sorry about my weak english)
>
> ___
> Moses-support mailing list
> Moses-support@mit.edu
> http://mailman.mit.edu/mailman/listinfo/moses-support
>
>
--
t;
> I managed to access the Moses libraries using boost python - to query the
> binarised phrase tables - contact me offline if you want me to send you the
> code.
>
> cheers - Barry
>
>
> On 10/09/12 17:40, Wilker Aziz wrote:
>
>> Hi everyone,
>>
>>
import the code I get an error message: ImportError:
./binpt.so: undefined symbol: _ZN5Moses20PhraseDictionaryTreeC1Em
I tried to grep that symbol from nm lib/libmoses.a and it wasn't there.
Would anyone know if I'm looking for it in the wrong place?
Is there any comments/directions you
- deadline for abstracts
- 24th July 2012 - all authors notified of decisions
- 12th November 2012 - speakers’ presentations to be submitted
- 29th-30th November 2012 – conference takes place in London
--
Wilker Aziz
http://pers-www.wlv.ac.uk/~in1676/
PhD candidate at The Researc
for zmert to be run with Joshua was
released last EMNLP.
I wonder whether MERT is already able to optimize towards METEOR.
Checking the mert directory I found BLEU and TER there, so I'm guessing we
don't have a version for METEOR.
Just wanted to confirm that before taking any action.
Tha
after moses was at least 2 years old, so it was retro-
> fitted onto the core features such as phrase table and language model.
>
> >
> > The second question (actually 2 in 1) would be:
> > Is there anything like a "Moses re-factoring team" and may I sign up to
&g
objects are always in valid
states (from the moment they are built) , being most of them immutable
whenever possible.
The second question (actually 2 in 1) would be:
Is there anything like a "Moses re-factoring team" and may I sign up to it?
Best regards,
--
Hi all,
after playing with factored models I came to realize that train-model.perl
uses the target side of the monolingual data to estimate the conditional
probability distributions used at generation steps... besides it seems
impossible to set a larger monolingual data (once all we need for gener
29 matches
Mail list logo
