concordia-docker/concordia-aligner/add_corpus.sh
2019-06-26 10:03:51 +02:00

9 lines
315 B
Bash
Executable File

#!/bin/sh
CORPUS_NAME=opensubtitles
CORPUS_PATH=../fast-aligner/corpora/$CORPUS_NAME
SRC_LANG_ID=1
TRG_LANG_ID=2
./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