summaryrefslogtreecommitdiff
path: root/cdec/train.sh
diff options
context:
space:
mode:
Diffstat (limited to 'cdec/train.sh')
-rwxr-xr-xcdec/train.sh27
1 files changed, 27 insertions, 0 deletions
diff --git a/cdec/train.sh b/cdec/train.sh
new file mode 100755
index 0000000..02c6c84
--- /dev/null
+++ b/cdec/train.sh
@@ -0,0 +1,27 @@
+SCRIPTS=~/src/scripts/
+CDEC=~/src/cdec-dtrain/
+DATA=../data/
+NAME=news-commentary
+
+#$SCRIPTS/preprocess de < $DATA/$NAME.de.raw > $NAME.de.nof.nocs
+$SCRIPTS/preprocess de < $DATA/$NAME.de.raw > $NAME.de.nof
+$SCRIPTS/preprocess en < $DATA/$NAME.en.raw > $NAME.en.nof
+
+#$MOSES/scripts/generic/compound-splitter.perl -train -corpus $NAME.de.nof.nocs -model cs_model.de 2>compound-splitter-train.de.err
+#$MOSES/scripts/generic/compound-splitter.perl -model cs_model.de < $NAME.de.nof.nocs > $NAME.de.nof
+
+$SCRIPTS/no_empty $NAME.de.nof $NAME.en.nof $NAME.de $NAME.en
+
+$CDEC/corpus/paste-files.pl $NAME.de $NAME.en > $NAME
+
+mkdir lm && cd lm
+$CDEC/klm/lm/builder/lmplz -S 80% -T /tmp -o 4 < ../$NAME.en > $NAME.arpa.4 2>lmplz.err
+$CDEC/klm/lm/build_binary $NAME.arpa.4 $NAME.ken.4 2>build_binary.err
+
+cd .. && mkdir a && cd a
+$CDEC/word-aligner/fast_align -d -v -o -i ../$NAME > forward 2>forward.err
+$CDEC/word-aligner/fast_align -d -v -o -r -i ../$NAME > backward 2>backward.err
+$CDEC/utils/atools -i forward -j backward -c grow-diag-final-and > gdfa
+
+$CDEC/extractor/run_extractor -b news-commentary -a a/gdfa --leave_one_out --grammars g/ < news-commentary.de > news-commentary.de.sgm
+