diff options
| -rw-r--r-- | gi/pipeline/clsp.config | 2 | ||||
| -rwxr-xr-x | gi/pipeline/evaluation-pipeline.pl | 3 | ||||
| -rwxr-xr-x | gi/pipeline/local-gi-pipeline.pl | 5 |
3 files changed, 8 insertions, 2 deletions
diff --git a/gi/pipeline/clsp.config b/gi/pipeline/clsp.config index 49feada0..27161fab 100644 --- a/gi/pipeline/clsp.config +++ b/gi/pipeline/clsp.config @@ -4,6 +4,6 @@ btec /export/ws10smt/data/btec/ split.zh-en.al lm/en.3gram.lm.gz devtest/devset1 fbis /export/ws10smt/data/chinese-english.fbis corpus.zh-en.al zhen /export/ws10smt/data/chinese-english corpus.zh-en.al aren /export/ws10smt/data/arabic-english corpus.ar-en.al -uren /export/ws10smt/data/urdu-english corpus.ur-en.al +uren /export/ws10smt/data/urdu-english corpus.ur-en.al lm/u2e.en.lm.gz dev/dev.ur dev/dev.en* devtest/devtest.ur eval-devtest.sh nlfr /export/ws10smt/data/dutch-french corpus.nl-fr.al diff --git a/gi/pipeline/evaluation-pipeline.pl b/gi/pipeline/evaluation-pipeline.pl index e0d13340..161dc170 100755 --- a/gi/pipeline/evaluation-pipeline.pl +++ b/gi/pipeline/evaluation-pipeline.pl @@ -100,7 +100,8 @@ if (GetOptions( print_help(); exit; } - +my @fkeys = keys %$feat_map; +die "You must specify one or more features with -f. Known features: @fkeys\n" unless scalar @features > 0; my @xfeats; for my $feat (@features) { my $rs = $feat_map->{$feat}; diff --git a/gi/pipeline/local-gi-pipeline.pl b/gi/pipeline/local-gi-pipeline.pl index 99b487db..a705af3b 100755 --- a/gi/pipeline/local-gi-pipeline.pl +++ b/gi/pipeline/local-gi-pipeline.pl @@ -43,8 +43,12 @@ my $TOPIC_TRAIN = "$PYPTOOLS/pyp-contexts-train"; assert_exec($SORT_KEYS, $REDUCER, $EXTRACTOR, $PYP_TOPICS_TRAIN, $S2L, $C2D, $TOPIC_TRAIN); +my $BACKOFF_GRAMMAR; +my $TAGGED_CORPUS; + my $OUTPUT = './giwork'; usage() unless &GetOptions('base_phrase_max_size=i' => \$BASE_PHRASE_MAX_SIZE, + 'backoff_grammar' => \$BACKOFF_GRAMMAR, 'output=s' => \$OUTPUT, 'model=s' => \$MODEL, 'topics=i' => \$NUM_TOPICS, @@ -54,6 +58,7 @@ usage() unless &GetOptions('base_phrase_max_size=i' => \$BASE_PHRASE_MAX_SIZE, 'em-iterations=i' => \$NUM_EM_PR_ITERS, 'pr-scale=f' => \$PR_SCALE, 'pr-threads=i' => \$PR_THREADS, + 'tagged_corpus=s' => \$TAGGED_CORPUS, ); usage() unless scalar @ARGV == 1; |