zmert support

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@730 ec762483-ff6d-05da-a07a-a48fb63a330f

2 files changed, 32 insertions, 6 deletions

diff --git a/rescore/generate_zmert_params_from_weights.pl b/rescore/generate_zmert_params_from_weights.pl
new file mode 100755
index 00000000..a9287896
--- /dev/null
+++ b/rescore/generate_zmert_params_from_weights.pl
@@ -0,0 +1,26 @@
+#!/usr/bin/perl -w
+
+my %defaults;
+$defaults{'LanguageModel'} = "Opt\t0\t10\t0\t2.5";
+$defaults{'EgivenF'} = "Opt\t-5\t0.5\t-3\t0.5";
+$defaults{'LexEGivenF'} = "Opt\t-5\t0.5\t-3\t0.5";
+$defaults{'LexFGivenE'} = "Opt\t-5\t0.5\t-3\t0.5";
+$defaults{'PassThrough'} = "Opt\t-Inf\t+Inf\t-10\t0";
+$defaults{'WordPenalty'} = "Opt\t-Inf\t2\t-5\t0";
+my $DEFAULT = "Opt\t-Inf\t+Inf\t-1\t+1";
+
+while(<>) {
+  next if /^#/;
+  chomp;
+  next if /^\s*$/;
+  s/^\s+//;
+  s/\s+$//;
+  my ($a,$b) = split /\s+/;
+  next unless ($a && $b);
+  my $line = $DEFAULT;
+  if ($defaults{$a}) { $line = $defaults{$a}; }
+  print "$a\t|||\t$b\t$line\n";
+}
+
+print "normalization = none\n";
+
diff --git a/rescore/rerank.pl b/rescore/rerank.pl
index 296ae6d5..4a0c5750 100755
--- a/rescore/rerank.pl
+++ b/rescore/rerank.pl
@@ -39,7 +39,7 @@ while(<HYP>) {
   my ($id, $hyp, $feats) = split / \|\|\| /;
   unless (defined $cur) { $cur = $id; }
   if ($cur ne $id) {
-    extract_1best(\%hyps);
+    extract_1best($cur, \%hyps);
     $cur = $id;
     %hyps = ();
   }
@@ -53,15 +53,15 @@ while(<HYP>) {
   }
   $hyps{"$hyp ||| $feats"} = $tot;
 }
-extract_1best(\%hyps) if defined $cur;
+extract_1best($cur, \%hyps) if defined $cur;
 close HYP;
 
 sub extract_1best {
-  my $rh = shift;
+  my ($id, $rh) = @_;
   my %hyps = %$rh;
   if ($kbest) {
     for my $hyp (sort { $hyps{$b} <=> $hyps{$a} } keys %hyps) {
-      print "$hyp\n";
+      print "$id ||| $hyp\n";
     }
   } else {
     my $best_score = undef;
@@ -79,8 +79,8 @@ sub extract_1best {
 
 sub usage {
   print <<EOT;
-Usage: $0 -w weights.txt -h hyp.nbest.txt
-  Reranks n-best lists with new weights, extracting the new 1-best entries.
+Usage: $0 -w weights.txt -h hyp.nbest.txt [--kbest]
+  Reranks n-best lists with new weights, extracting the new 1/k-best entries.
 EOT
 }