From 27db9d8c05188f64c17d61c394d3dafe8b8e93d8 Mon Sep 17 00:00:00 2001
From: Chris Dyer <redpony@gmail.com>
Date: Sat, 19 Dec 2009 14:32:28 -0500
Subject: cool new alignment feature

---
 training/cluster-ptrain.pl | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

(limited to 'training')

diff --git a/training/cluster-ptrain.pl b/training/cluster-ptrain.pl
index 33aab25d..8944ae34 100755
--- a/training/cluster-ptrain.pl
+++ b/training/cluster-ptrain.pl
@@ -104,7 +104,21 @@ if ($restart) {
   } else {
     `cp $initial_weights $dir/weights.1.gz`;
   }
+  open T, "<$training_corpus" or die "Can't read $training_corpus: $!";
+  open TO, ">$dir/training.in";
+  my $lc = 0;
+  while(<T>) {
+    chomp;
+    s/^\s+//;
+    s/\s+$//;
+    die "Expected A ||| B in input file" unless / \|\|\| /;
+    print TO "<seg id=\"$lc\">$_</seg>\n";
+    $lc++;
+  }
+  close T;
+  close TO;
 }
+$training_corpus = "$dir/training.in";
 
 my $iter_attempts = 1;
 while ($iter < $max_iteration) {
-- 
cgit v1.2.3