diff options
Diffstat (limited to 'gi/pipeline')
| -rwxr-xr-x | gi/pipeline/local-gi-pipeline.pl | 13 |
1 files changed, 3 insertions, 10 deletions
diff --git a/gi/pipeline/local-gi-pipeline.pl b/gi/pipeline/local-gi-pipeline.pl index 7c31be3b..a9107907 100755 --- a/gi/pipeline/local-gi-pipeline.pl +++ b/gi/pipeline/local-gi-pipeline.pl @@ -13,9 +13,7 @@ my $NUM_TOPICS = 50; my $NUM_SAMPLES = 1000; my $CONTEXT_SIZE = 1; my $BIDIR = 1; - -my $HIERARCHICAL_TOPICS = 0; -my $FILTER_SINGLETONS = 0; +my $TOPICS_CONFIG = "pyp-topics.conf"; my $EXTOOLS = "$SCRIPT_DIR/../../extools"; die "Can't find extools: $EXTOOLS" unless -e $EXTOOLS && -d $EXTOOLS; @@ -42,8 +40,7 @@ usage() unless &GetOptions('base_phrase_max_size=i' => \$BASE_PHRASE_MAX_SIZE, 'topics=i' => \$NUM_TOPICS, 'trg_context=i' => \$CONTEXT_SIZE, 'samples=i' => \$NUM_SAMPLES, - 'hierarchical-topics' => \$HIERARCHICAL_TOPICS, - 'filter-singletons' => \$FILTER_SINGLETONS, + 'topics-config=s' => \$TOPICS_CONFIG, ); usage() unless scalar @ARGV == 1; @@ -135,11 +132,7 @@ sub topic_train { if (-e $OUT_CLUSTERS) { print STDERR "$OUT_CLUSTERS exists, reusing...\n"; } else { - my $FILTER_SINGLETONS_ARG = ""; - $FILTER_SINGLETONS_ARG = "--filter-singleton-contexts" if $FILTER_SINGLETONS; - my $HIERARCHICAL_TOPICS_ARG = ""; - $HIERARCHICAL_TOPICS_ARG = "--hierarchical-topics" if $HIERARCHICAL_TOPICS; - safesystem("$TOPIC_TRAIN --data $IN_CONTEXTS --backoff-type simple -t $NUM_TOPICS -s $NUM_SAMPLES -o $OUT_CLUSTERS $HIERARCHICAL_TOPICS_ARG $FILTER_SINGLETONS_ARG -w /dev/null") or die "Topic training failed.\n"; + safesystem("$TOPIC_TRAIN --data $IN_CONTEXTS --backoff-type simple -t $NUM_TOPICS -s $NUM_SAMPLES -o $OUT_CLUSTERS -c $TOPICS_CONFIG -w /dev/null") or die "Topic training failed.\n"; } } |