From fd920188bd5f912a45ef862b2a2ed7b1062b564e Mon Sep 17 00:00:00 2001
From: redpony <redpony@ec762483-ff6d-05da-a07a-a48fb63a330f>
Date: Thu, 22 Jul 2010 20:52:22 +0000
Subject: add additional filtering step

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@368 ec762483-ff6d-05da-a07a-a48fb63a330f
---
 vest/dist-vest.pl | 17 +++++++++++++++--
 1 file changed, 15 insertions(+), 2 deletions(-)

(limited to 'vest')

diff --git a/vest/dist-vest.pl b/vest/dist-vest.pl
index 97b72f45..8251de30 100755
--- a/vest/dist-vest.pl
+++ b/vest/dist-vest.pl
@@ -56,6 +56,7 @@ my $maxsim=0;
 my $oraclen=0;
 my $oracleb=20;
 my $dirargs='';
+my $density_prune;
 my $usefork;
 
 # Process command-line options
@@ -63,6 +64,7 @@ Getopt::Long::Configure("no_auto_abbrev");
 if (GetOptions(
 	"decoder=s" => \$decoderOpt,
 	"decode-nodes=i" => \$decode_nodes,
+  "density-prune=f" => \$density_prune,
 	"dont-clean" => \$disable_clean,
         "use-fork" => \$usefork,
 	"dry-run" => \$dryrun,
@@ -93,6 +95,10 @@ if (GetOptions(
 	exit;
 }
 
+if (defined $density_prune) {
+  die "--density_prune n: n must be greater than 1.0\n" unless $density_prune > 1.0;
+}
+
 if ($usefork) { $usefork = "--use-fork"; } else { $usefork = ''; }
 
 if ($metric =~ /^(combi|ter)$/i) {
@@ -236,9 +242,12 @@ while (1){
 	my $im1 = $iteration - 1;
 	my $weightsFile="$dir/weights.$im1";
 	my $decoder_cmd = "$decoder -c $iniFile -w $weightsFile -O $dir/hgs";
+  if ($density_prune) {
+    $decoder_cmd .= " --density_prune $density_prune";
+  }
 	my $pcmd = "cat $srcFile | $parallelize $usefork -p $pmem -e $logdir -j $decode_nodes -- ";
-        if ($run_local) { $pcmd = "cat $srcFile |"; }
-        my $cmd = $pcmd . "$decoder_cmd 2> $decoderLog 1> $runFile";
+  if ($run_local) { $pcmd = "cat $srcFile |"; }
+  my $cmd = $pcmd . "$decoder_cmd 2> $decoderLog 1> $runFile";
 	print STDERR "COMMAND:\n$cmd\n";
 	my $result = 0;
 	$result = system($cmd);
@@ -547,6 +556,10 @@ Options:
 	--decoder <decoder path>
 		Decoder binary to use.
 
+  --density-prune <N>
+    Limit the density of the hypergraph on each iteration to N times
+    the number of edges on the Viterbi path.
+
 	--help
 		Print this message and exit.
 
-- 
cgit v1.2.3