From c807e0b514f21a80df0268c686c7ba70fe39611a Mon Sep 17 00:00:00 2001
From: "trevor.cohn" <trevor.cohn@ec762483-ff6d-05da-a07a-a48fb63a330f>
Date: Tue, 13 Jul 2010 20:09:22 +0000
Subject: Updated PR command line and output directory. Override for experiment
 directory.

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@240 ec762483-ff6d-05da-a07a-a48fb63a330f
---
 gi/pipeline/evaluation-pipeline.pl     |  1 +
 gi/pipeline/local-gi-pipeline.pl       | 15 ++++++++++-----
 gi/posterior-regularisation/prjava.jar |  2 +-
 3 files changed, 12 insertions(+), 6 deletions(-)

(limited to 'gi')

diff --git a/gi/pipeline/evaluation-pipeline.pl b/gi/pipeline/evaluation-pipeline.pl
index f5ad7b14..6e786a8a 100755
--- a/gi/pipeline/evaluation-pipeline.pl
+++ b/gi/pipeline/evaluation-pipeline.pl
@@ -110,6 +110,7 @@ if (GetOptions(
         "backoff_grammar" => \$bkoffgram,
         "data=s" => \$dataDir,
         "features=s@" => \@features,
+        "out-dir=s" => \$outdir,
 ) == 0 || @ARGV!=2 || $help) {
         print_help();
         exit;
diff --git a/gi/pipeline/local-gi-pipeline.pl b/gi/pipeline/local-gi-pipeline.pl
index 259dcd9c..bcf9c9be 100755
--- a/gi/pipeline/local-gi-pipeline.pl
+++ b/gi/pipeline/local-gi-pipeline.pl
@@ -20,9 +20,10 @@ my $TOPICS_CONFIG = "pyp-topics.conf";
 my $MODEL = "pyp";
 my $NUM_EM_ITERS = 100;
 my $NUM_PR_ITERS = 0;
-my $PR_SCALE_P = 10;
+my $PR_SCALE_P = 1;
 my $PR_SCALE_C = 0;
 my $PR_THREADS = 0;
+my $AGREE;
 
 my $EXTOOLS = "$SCRIPT_DIR/../../extools";
 die "Can't find extools: $EXTOOLS" unless -e $EXTOOLS && -d $EXTOOLS;
@@ -63,6 +64,7 @@ usage() unless &GetOptions('base_phrase_max_size=i' => \$BASE_PHRASE_MAX_SIZE,
                            'pr-scale-phrase=f' => \$PR_SCALE_P,
                            'pr-scale-context=f' => \$PR_SCALE_C,
                            'pr-threads=i' => \$PR_THREADS,
+                           'pr-agree' => \$AGREE,
                            'tagged_corpus=s' => \$TAGGED_CORPUS,
                           );
 
@@ -132,10 +134,12 @@ sub cluster_dir {
     if (lc($MODEL) eq "pyp") {
         return context_dir() . ".PYP.t$NUM_TOPICS.s$NUM_SAMPLES";
     } elsif (lc($MODEL) eq "prem") {
-        if ($NUM_PR_ITERS == 0) {
-            return context_dir() . ".PREM.t$NUM_TOPICS.ie$NUM_EM_ITERS.ip$NUM_PR_ITERS";
+        if (defined($AGREE)) {
+            return context_dir() . ".AGREE.t$NUM_TOPICS.i$NUM_EM_ITERS";
+	} elsif ($NUM_PR_ITERS == 0) {
+            return context_dir() . ".EM.t$NUM_TOPICS.i$NUM_EM_ITERS";
         } else {
-            return context_dir() . ".PREM.t$NUM_TOPICS.ie$NUM_EM_ITERS.ip$NUM_PR_ITERS.sp$PR_SCALE_P.sc$PR_SCALE_C";
+            return context_dir() . ".PR.t$NUM_TOPICS.ie$NUM_EM_ITERS.ip$NUM_PR_ITERS.sp$PR_SCALE_P.sc$PR_SCALE_C";
         }
     }
 }
@@ -205,7 +209,8 @@ sub prem_train {
   if (-e $OUT_CLUSTERS) {
     print STDERR "$OUT_CLUSTERS exists, reusing...\n";
   } else {
-    safesystem("$PREM_TRAIN --in $IN_CONTEXTS --topics $NUM_TOPICS --out $OUT_CLUSTERS --em $NUM_EM_ITERS --pr $NUM_PR_ITERS --scale-phrase $PR_SCALE_P --scale-context $PR_SCALE_C --threads $PR_THREADS") or die "Topic training failed.\n";
+    my $agree = ($AGREE) ? "--agree" : "";
+    safesystem("$PREM_TRAIN --in $IN_CONTEXTS --topics $NUM_TOPICS --out $OUT_CLUSTERS --em $NUM_EM_ITERS --pr $NUM_PR_ITERS --scale-phrase $PR_SCALE_P --scale-context $PR_SCALE_C --threads $PR_THREADS $agree") or die "Topic training failed.\n";
   }
 }
 
diff --git a/gi/posterior-regularisation/prjava.jar b/gi/posterior-regularisation/prjava.jar
index 7cd1a3ff..da8bf761 120000
--- a/gi/posterior-regularisation/prjava.jar
+++ b/gi/posterior-regularisation/prjava.jar
@@ -1 +1 @@
-prjava/prjava-20100707.jar
\ No newline at end of file
+prjava/prjava-20100708.jar
\ No newline at end of file
-- 
cgit v1.2.3