Hello,

I am trying to prune my phrase table using the SALM generated suffix array
indices but I don't see anything pruned.  I am using the
news-commentary-v10 data for english to french translation.

I followed the instructions on this wiki:
http://www.statmt.org/moses/?n=Advanced.RuleTables#ntoc5

Here is the command I used to run this:

I generated the suffix arrays using:

SALM/Bin/Linux/Index/IndexSA.O32 TARGET
SALM/Bin/Linux/Index/IndexSA.O32 SOURCE


[sbhadour@sbhadour-ld1 model]$ cat phrase-table |
/home/sbhadour/work/experiments/mosesdecoder/contrib/sigtest-filter/filter-pt
-e
/home/sbhadour/work/experiments/lang-training-data/salm-sufix-arr-index/en_fr/news-commentary-v10.fr-en.clean.fr.sa_suffix
-f
/home/sbhadour/work/experiments/lang-training-data/salm-sufix-arr-index/en_fr/news-commentary-v10.fr-en.clean.en.sa_suffix
-l 'a+e -n 30' > phrase-table.pruned

-l = a+e -n 30

Filtering using P(e|f) only. n=0

..................................................[n:500000]
..................................................[n:1000000]
..................................................[n:1500000]
..................................................[n:2000000]
..................................................[n:2500000]
..................................................[n:3000000]
..................................................[n:3500000]
..................................................[n:4000000]
..................................................[n:4500000]
..................................................[n:5000000]
..................................................[n:5500000]
..................................................[n:6000000]
............

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

  unfiltered phrases pairs: 6121925
     P(f|e) filter [first]: 0   (0%)
       significance filter: 0   (0%)
            TOTAL FILTERED: 0   (0%)
     FILTERED phrase pairs: 6121925   (100%)

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


Am I missing anything? Help appreciated!

-- 
Regards,
Sameer Bhadouria.
_______________________________________________
Moses-support mailing list
[email protected]
http://mailman.mit.edu/mailman/listinfo/moses-support

Reply via email to