Merge branch 'master' of github.com:moses-smt/mosesdecoder

4 files changed, 9 insertions, 6 deletions

diff --git a/scripts/training/clean-corpus-n.perl b/scripts/training/clean-corpus-n.perl
index bea32052a..2865fe391 100755
--- a/scripts/training/clean-corpus-n.perl
+++ b/scripts/training/clean-corpus-n.perl
@@ -47,7 +47,7 @@ my $l1input = "$corpus.$l1";
 if (-e $l1input) {
   $opn = $l1input;
 } elsif (-e $l1input.".gz") {
-  $opn = "zcat $l1input.gz |";
+  $opn = "gunzip -c $l1input.gz |";
 } else {
     die "Error: $l1input does not exist";
 }
@@ -57,7 +57,7 @@ my $l2input = "$corpus.$l2";
 if (-e $l2input) {
   $opn = $l2input;
 } elsif (-e $l2input.".gz") {
-  $opn = "zcat $l2input.gz |";
+  $opn = "gunzip -c $l2input.gz |";
 } else  {
  die "Error: $l2input does not exist";
 }
@@ -160,3 +160,4 @@ sub word_count {
   my @w = split(/ /,$line);
   return scalar @w;
 }
+
diff --git a/scripts/training/filter-rule-table.py b/scripts/training/filter-rule-table.py
index 8bef034de..86c8b300e 100755
--- a/scripts/training/filter-rule-table.py
+++ b/scripts/training/filter-rule-table.py
@@ -40,7 +40,8 @@ def printUsage():
 def main():
     parser = optparse.OptionParser()
     parser.add_option("-c", "--min-non-initial-rule-count",
-                      action="store", dest="minCount", type="int", default="1",
+                      action="store", dest="minCount",
+                      type="float", default="0.0",
                       help="prune non-initial rules where count is below N",
                       metavar="N")
     (options, args) = parser.parse_args()
diff --git a/scripts/training/mert-moses.pl b/scripts/training/mert-moses.pl
index 688e8ce55..9f5f25f15 100755
--- a/scripts/training/mert-moses.pl
+++ b/scripts/training/mert-moses.pl
@@ -1,4 +1,4 @@
-#!/usr/bin/perl -w
+#!/usr/bin/perl -w 
 # $Id$
 # Usage:
 # mert-moses.pl <foreign> <english> <decoder-executable> <decoder-config>
@@ -371,7 +371,7 @@ my $pro_optimizer = File::Spec->catfile($mertdir, "megam_i686.opt");  # or set t
 
 if (($___PAIRWISE_RANKED_OPTIMIZER || $___PRO_STARTING_POINT) && ! -x $pro_optimizer) {
   print "Could not find $pro_optimizer, installing it in $mertdir\n";
-  my $megam_url = "http://www.umiacs.umd.edu/~hal/megam/";
+  my $megam_url = "http://hal3.name/megam";
   if (&is_mac_osx()) {
     die "Error: Sorry for Mac OS X users! Please get the source code of megam and compile by hand. Please see $megam_url for details.";
   }
diff --git a/scripts/training/train-model.perl b/scripts/training/train-model.perl
index 5b0553581..e4292007e 100755
--- a/scripts/training/train-model.perl
+++ b/scripts/training/train-model.perl
@@ -38,8 +38,9 @@ my($_EXTERNAL_BINDIR, $_ROOT_DIR, $_CORPUS_DIR, $_GIZA_E2F, $_GIZA_F2E, $_MODEL_
    $_MEMSCORE, $_FINAL_ALIGNMENT_MODEL,
    $_CONTINUE,$_MAX_LEXICAL_REORDERING,$_DO_STEPS,
    @_ADDITIONAL_INI,$_ADDITIONAL_INI_FILE,
-   $_SPARSE_TRANSLATION_TABLE, @_BASELINE_ALIGNMENT_MODEL, $_BASELINE_EXTRACT, $_BASELINE_CORPUS, $_BASELINE_ALIGNMENT,
+   $_SPARSE_TRANSLATION_TABLE, @_BASELINE_ALIGNMENT_MODEL, $_BASELINE_EXTRACT, $_BASELINE_ALIGNMENT,
    $_DICTIONARY, $_SPARSE_PHRASE_FEATURES, $_EPPEX, $_INSTANCE_WEIGHTS_FILE, $_LMODEL_OOV_FEATURE, $IGNORE);
+my $_BASELINE_CORPUS = "";
 my $_CORES = 1;
 
 my $debug = 0; # debug this script, do not delete any files in debug mode