multimodel with compact pt

8 files changed, 94 insertions, 0 deletions

diff --git a/models/multimodel/model1/phrase-table.minphr b/models/multimodel/model1/phrase-table.minphr
new file mode 100644
index 0000000..ec86fbb
--- /dev/null
+++ b/models/multimodel/model1/phrase-table.minphr
diff --git a/models/multimodel/model2/phrase-table.minphr b/models/multimodel/model2/phrase-table.minphr
new file mode 100644
index 0000000..aa72322
--- /dev/null
+++ b/models/multimodel/model2/phrase-table.minphr
diff --git a/tests/phrase.multimodel-compactptable/filter-stderr.pl b/tests/phrase.multimodel-compactptable/filter-stderr.pl
new file mode 100755
index 0000000..4c279a1
--- /dev/null
+++ b/tests/phrase.multimodel-compactptable/filter-stderr.pl
@@ -0,0 +1,22 @@
+#!/usr/bin/env perl
+
+BEGIN { use Cwd qw/ abs_path /; use File::Basename; $script_dir = dirname(abs_path($0)); push @INC, "$script_dir/../perllib"; }
+use RegTestUtils;
+
+$x=0;
+while (<>) {
+  chomp;
+
+  if (/^Finished loading LanguageModels/) {
+    my $time = RegTestUtils::readTime($_);
+    print "LMLOAD_TIME ~ $time\n";
+  }
+  if (/^Finished loading phrase tables/) {
+    my $time = RegTestUtils::readTime($_);
+    print "PTLOAD_TIME ~ $time\n";
+  }
+  next unless /^BEST TRANSLATION:/;
+  my $pscore = RegTestUtils::readHypoScore($_);
+  $x++;
+  print "SCORE_$x = $pscore\n";
+}
diff --git a/tests/phrase.multimodel-compactptable/filter-stdout.pl b/tests/phrase.multimodel-compactptable/filter-stdout.pl
new file mode 100755
index 0000000..0715c2d
--- /dev/null
+++ b/tests/phrase.multimodel-compactptable/filter-stdout.pl
@@ -0,0 +1,7 @@
+#!/usr/bin/env perl
+$x=0;
+while (<>) {
+  chomp;
+  $x++;
+  print "TRANSLATION_$x=$_\n";
+}
diff --git a/tests/phrase.multimodel-compactptable/info b/tests/phrase.multimodel-compactptable/info
new file mode 100644
index 0000000..d1f8792
--- /dev/null
+++ b/tests/phrase.multimodel-compactptable/info
@@ -0,0 +1,12 @@
+perplexity minimization (not automated yet);
+optimized on ${MODEL_PATH}/multimodel/extract.sorted.gz
+
+
+Cross-entropy: 2.16812
+Weight vector for feature 0: 0.863753 0.136247 
+Cross-entropy: 1.56377
+Weight vector for feature 1: 0.936675 0.0633248 
+Cross-entropy: 1.74298
+Weight vector for feature 2: 0.825263 0.174737 
+Cross-entropy: 2.28705
+Weight vector for feature 3: 0.936133 0.0638672
\ No newline at end of file
diff --git a/tests/phrase.multimodel-compactptable/moses.ini b/tests/phrase.multimodel-compactptable/moses.ini
new file mode 100644
index 0000000..e497a13
--- /dev/null
+++ b/tests/phrase.multimodel-compactptable/moses.ini
@@ -0,0 +1,33 @@
+# moses.ini for regression test
+
+[distortion-limit]
+4
+
+[beam-threshold]
+0.03
+
+[input-factors]
+0
+
+[mapping]
+T 2
+
+[verbose]
+2
+
+[feature]
+UnknownWordPenalty
+WordPenalty
+Distortion
+PhrasePenalty
+PhraseDictionaryCompact tuneable=false num-features=4 input-factor=0 output-factor=0 path=${MODEL_PATH}/multimodel/model1/phrase-table.minphr table-limit=20
+PhraseDictionaryCompact tuneable=false num-features=4 input-factor=0 output-factor=0 path=${MODEL_PATH}/multimodel/model2/phrase-table.minphr table-limit=20
+PhraseDictionaryMultiModel num-features=4 input-factor=0 output-factor=0 table-limit=20 mode=interpolate lambda=0.2,0.8 components=PhraseDictionaryCompact0,PhraseDictionaryCompact1
+
+[weight]
+WordPenalty0= -0.273416114951401
+Distortion0= 0.141806519223522
+PhrasePenalty0= -0.0426081987467227
+PhraseDictionaryMultiModel0= 0.00402447059454402 0.0685647475075862 0.294089113124688 0.0328320356515851
+PhraseDictionaryCompact0= 0 0 1 0
+PhraseDictionaryCompact1= 0 0 1 0
diff --git a/tests/phrase.multimodel-compactptable/to-translate.txt b/tests/phrase.multimodel-compactptable/to-translate.txt
new file mode 100644
index 0000000..2e1f048
--- /dev/null
+++ b/tests/phrase.multimodel-compactptable/to-translate.txt
@@ -0,0 +1,7 @@
+ich erkläre die am donnerstag , dem 3. februar 2000 , unterbrochene sitzungsperiode des europäischen parlaments für wiederaufgenommen .
+genehmigung des protokolls der vorangegangenen sitzung
+das protokoll der sitzung von donnerstag , dem 3. februar ist verteilt worden .
+gibt es einwände ?
+aus ihrer reaktion schließe ich , daß viele mitglieder des hauses dieses protokoll nicht erhalten haben , und es versteht sich von selbst , daß sie einem protokoll , das sie nicht erhalten haben , nicht zustimmen können .
+ich schlage deshalb vor , das protokoll morgen zu genehmigen , da es ihnen offensichtlich jetzt nicht vorliegt , wofür ich mich entschuldige .
+die genehmigung des protokolls wird somit verschoben .
\ No newline at end of file
diff --git a/tests/phrase.multimodel-compactptable/truth/results.txt b/tests/phrase.multimodel-compactptable/truth/results.txt
new file mode 100644
index 0000000..ac28a55
--- /dev/null
+++ b/tests/phrase.multimodel-compactptable/truth/results.txt
@@ -0,0 +1,13 @@
+TRANSLATION_1=i erkläre which originated thursday , the 3. february 2000 , unterbrochene part-session of the european parliament adjourned on . 
+TRANSLATION_2=adoption of the minutes of the previous sitting 
+TRANSLATION_3=the minutes for ' s sitting tomorrow thursday , the 3. february 's been distributed been . 
+TRANSLATION_4=are there any comments ? 
+TRANSLATION_5=from their reaktion schließe i , that many members of the when this minutes 't obtain have , and it versteht of that kind to what is from even , that you a minutes , it you 't obtain have , 't zustimmen can . 
+TRANSLATION_6=i schlage so a , it minutes tomorrow to , there it the opportunity to propose to you however , now 't vorliegt , wofür i me entschuldige . 
+SCORE_1 = -301.799
+SCORE_2 = 0.036
+SCORE_3 = -102.562
+SCORE_4 = -0.102
+SCORE_5 = -410.920
+SCORE_6 = -403.121
+TOTAL_WALLTIME ~ 3