diff options
| author | trevor.cohn <trevor.cohn@ec762483-ff6d-05da-a07a-a48fb63a330f> | 2010-07-16 22:20:50 +0000 |
|---|---|---|
| committer | trevor.cohn <trevor.cohn@ec762483-ff6d-05da-a07a-a48fb63a330f> | 2010-07-16 22:20:50 +0000 |
| commit | e79efced08b7cc4dc692954061f433e01fa7e137 (patch) | |
| tree | a8524e6c2dfba7546a102145fdc377ed84ffc3b2 /gi/pipeline | |
| parent | d484433d9ad566076b7d078c77443ce7841f4f3e (diff) | |
Fixed PR command line
git-svn-id: https://ws10smt.googlecode.com/svn/trunk@303 ec762483-ff6d-05da-a07a-a48fb63a330f
Diffstat (limited to 'gi/pipeline')
| -rwxr-xr-x | gi/pipeline/local-gi-pipeline.pl | 20 |
1 files changed, 9 insertions, 11 deletions
diff --git a/gi/pipeline/local-gi-pipeline.pl b/gi/pipeline/local-gi-pipeline.pl index 131c22aa..34948880 100755 --- a/gi/pipeline/local-gi-pipeline.pl +++ b/gi/pipeline/local-gi-pipeline.pl @@ -23,11 +23,10 @@ my $LANGUAGE = "target"; my $LABEL_THRESHOLD = 0; my $MODEL = "pyp"; -my $NUM_EM_ITERS = 100; -my $NUM_PR_ITERS = 0; -my $PR_SCALE_P = 10; +my $NUM_ITERS = 100; +my $PR_SCALE_P = 0; my $PR_SCALE_C = 0; -my $PR_THREADS = 0; +my $PR_FLAGS = ""; my $EXTOOLS = "$SCRIPT_DIR/../../extools"; die "Can't find extools: $EXTOOLS" unless -e $EXTOOLS && -d $EXTOOLS; @@ -71,11 +70,10 @@ usage() unless &GetOptions('base_phrase_max_size=i' => \$BASE_PHRASE_MAX_SIZE, 'label_threshold=f' => \$LABEL_THRESHOLD, 'use_default_cat' => \$DEFAULT_CAT, 'topics-config=s' => \$TOPICS_CONFIG, - 'em-iterations=i' => \$NUM_EM_ITERS, - 'pr-iterations=i' => \$NUM_PR_ITERS, + 'iterations=i' => \$NUM_ITERS, 'pr-scale-phrase=f' => \$PR_SCALE_P, 'pr-scale-context=f' => \$PR_SCALE_C, - 'pr-threads=i' => \$PR_THREADS, + 'pr-flags=s' => \$PR_FLAGS, 'tagged_corpus=s' => \$TAGGED_CORPUS, 'language=s' => \$LANGUAGE, 'get_name_only' => \$NAME_SHORTCUT, @@ -187,10 +185,10 @@ sub cluster_dir { if (lc($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$NUM_PR_ITERS"; + if ($PR_SCALE_P == 0 && $PR_SCALE_C == 0) { + return context_dir() . ".EM.t$NUM_TOPICS.i$NUM_ITERS"; } else { - return context_dir() . ".PREM.t$NUM_TOPICS.ie$NUM_EM_ITERS.ip$NUM_PR_ITERS.sp$PR_SCALE_P.sc$PR_SCALE_C"; + return context_dir() . ".PR.t$NUM_TOPICS.i$NUM_ITERS.sp$PR_SCALE_P.sc$PR_SCALE_C"; } } } @@ -272,7 +270,7 @@ 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 $PR_THREADS") or die "Topic training failed.\n"; + safesystem("$PREM_TRAIN --in $IN_CONTEXTS --topics $NUM_TOPICS --out $OUT_CLUSTERS --iterations $NUM_ITERS --scale-phrase $PR_SCALE_P --scale-context $PR_SCALE_C $PR_FLAGS") or die "Topic training failed.\n"; } } |