addOpenSubtitles script
This commit is contained in:
parent
46596bc0b8
commit
275deaee18
19
tests/addOpenSubtitles.sh
Executable file
19
tests/addOpenSubtitles.sh
Executable file
@ -0,0 +1,19 @@
|
|||||||
|
#!/bin/sh
|
||||||
|
|
||||||
|
CORPUS_NAME=opensubtitles_plen
|
||||||
|
CORPUS_PATH=/root/opensubtitles_pack/plen
|
||||||
|
SRC_LANG_ID=1
|
||||||
|
TRG_LANG_ID=2
|
||||||
|
|
||||||
|
./addSources.py $CORPUS_PATH/ids_sources.txt
|
||||||
|
./addFastAlignedTM.py $CORPUS_NAME $CORPUS_PATH/src_clean.txt $CORPUS_PATH/src_clean.lem $SRC_LANG_ID $CORPUS_PATH/trg_clean.txt $TRG_LANG_ID $CORPUS_PATH/alignments.txt $CORPUS_PATH/ids_clean.txt
|
||||||
|
|
||||||
|
CORPUS_NAME=opensubtitles_enpl
|
||||||
|
CORPUS_PATH=/root/opensubtitles_pack/enpl
|
||||||
|
SRC_LANG_ID=2
|
||||||
|
TRG_LANG_ID=1
|
||||||
|
|
||||||
|
./addSources.py $CORPUS_PATH/ids_sources.txt
|
||||||
|
./addFastAlignedTM.py $CORPUS_NAME $CORPUS_PATH/src_clean.txt $CORPUS_PATH/src_clean.lem $SRC_LANG_ID $CORPUS_PATH/trg_clean.txt $TRG_LANG_ID $CORPUS_PATH/alignments.txt $CORPUS_PATH/ids_clean.txt
|
||||||
|
|
||||||
|
|
Loading…
Reference in New Issue
Block a user