--tree-grammar "glue:goal-source=[ROOT],goal-target=[ROOT],non-terminal-source=[x],non-terminal-target=[x],straight=true,invert=false" \
--- samples/kftt.30k/alignment/alignment.sh.orig 2013-07-30 14:14:40.000000000 +0900
+++ samples/kftt.30k/alignment/alignment.sh 2013-10-27 18:22:32.000000000 +0900
## We perform symmetized posterior constrained training, and
## perform smoothing by naive Bayes.
-exec ${cicada}/scripts/cicada-alignment.py \
+exec ${bindir}/cicada-alignment.py \
--f ../data/train.ja.bz2 \
--e ../data/train.en.bz2 \
--- samples/kftt.30k/ngram/expgram.sh.orig 2013-07-30 14:14:34.000000000 +0900
+++ samples/kftt.30k/ngram/expgram.sh 2013-10-27 18:23:31.000000000 +0900
-if test "$expgram" = ""; then
- echo "where is your expgram?"
### Following is a quick example of LM estimation on a small data set.
-$expgram/progs/expgram_counts_extract --corpus ../data/train.en.bz2 --output ngram.5.en.counts --order 5 --threads 4
+${bindir}/expgram_counts_extract --corpus ../data/train.en.bz2 --output ngram.5.en.counts --order 5 --threads 4
-$expgram/progs/expgram_counts_estimate --ngram ngram.5.en.counts --output ngram.5.en.lm --shard 4
+${bindir}/expgram_counts_estimate --ngram ngram.5.en.counts --output ngram.5.en.lm --shard 4
-# $expgram/scripts/expgram.py --corpus ../data/train.en --output ngram.5.en --threads 4
+# ${bindir}/expgram.py --corpus ../data/train.en --output ngram.5.en --threads 4
--- samples/kftt.30k/s2t/data/preprocess.sh.orig 2013-08-02 10:22:41.000000000 +0900
+++ samples/kftt.30k/s2t/data/preprocess.sh 2013-10-27 19:52:56.000000000 +0900
-if test "$stanford" = ""; then
- echo "where is your stanford parser?"
+stanford=@PREFIX@/share/java/stanford-parser
# Here, we use stanford-parser to parse training data in English
# cicada_filter_penntreebank to transform into hypergraph.
bzcat ../../data/train.en.bz2 | \
- -cp $stanford/stanford-parser.jar:$stanford/stanford-parser-3.2.0-models.jar \