Hi
I've used Lrsocre for tuning moses, with this command:mert-moses.pl
--mertargs="--sctype KENDALL,BLEU --scconfig weights:0.2+0.8 refalign: ... "It
works well and gives a bleu score on tuning set.
Now I'm going to use Lrscore to evaluate the test set with Lrscore but I don't
know how.
ThanksA
HiThanks Philipp for the reply.
1- when I use the suggested command, after the first run of tuning , It gives
"segmentation fault" error.
the exact command that I use:nohup nice $SCRIPTS_ROOTDIR/training/mert-moses.plĀ
--mertargs="--sctype KENDALL,BLEU --scconfig weights:0.6+0.4" m.en m.fa
\$MOSE
Hi
Thanks Philipp for the reply.
1- when I use the suggested command, after the first run of tuning , It
gives "segmentation fault" error.
the exact command that I use:
nohup nice $SCRIPTS_ROOTDIR/training/mert-moses.pl --mertargs="--sctype
KENDALL,BLEU --scconfig weights:0.6+0.4" m.en m.fa \
$MO
Hi,
you can tell mert the scorer type with "--sctype":
http://www.statmt.org/moses/?n=FactoredTraining.Tuning
The KENDALL scorer is likely doing what you want.
You can specify a combination of scorers
--sctype KENDALL,BLEU --scconfig weights:0.6+0.4
-phi
2015-09-19 10:11 GMT+02:00 Arefeh Kaz
Hi everyone
I'm testing a reordering model for hierarchical moses. I want to
interpolate bleu with kendall's tau permutation distance and use it as my
tuning metric. can anyone help me how to do this?
Regards
--
*Email Disclaimer"This e-mail and any files transmitted with it are
confidentia