s470607/mt-summit-corpora
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

add prepare-gonito

Browse Source
This commit is contained in:
jakubknczny 2022-01-15 22:21:56 +01:00
parent 5917c2799c
commit a4b31d2f0c
1 changed files with 34 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

34
prepare-gonito.sh Executable file
View File

@ -0,0 +1,34 @@
#!/bin/bash
tar xvf mt-summit-corpora.tar.gz >> /dev/null 2>&1
cut -f1 mt-summit-corpora/train/train.en-pl > mt-summit-corpora/train/train.en
cut -f2 mt-summit-corpora/train/train.en-pl > mt-summit-corpora/train/train.pl
cut -f1 mt-summit-corpora/dev/dev.en-pl > mt-summit-corpora/dev/dev.en
cut -f2 mt-summit-corpora/dev/dev.en-pl > mt-summit-corpora/dev/dev.pl
cut -f1 mt-summit-corpora/test/constrained-scenario-test.tsv > mt-summit-corpora/test/constrained-scenario-test.en
cut -f2 mt-summit-corpora/test/constrained-scenario-test.tsv > mt-summit-corpora/test/constrained-scenario-test.pl
cut -f2 mt-summit-corpora/test/general-scenario-test.tsv > mt-summit-corpora/test/general-scenario-test.pl
cut -f1 mt-summit-corpora/test/general-scenario-test.tsv > mt-summit-corpora/test/general-scenario-test.en
rm mt-summit-corpora.tar.gz
rm mt-summit-corpora/train/train.en-pl
rm mt-summit-corpora/test/*.tsv
rm mt-summit-corpora/dev/dev.en-pl
mv mt-summit-corpora/dev ./dev-0
mv mt-summit-corpora/train ./train
mv mt-summit-corpora/test ./test-A
rm -r mt-summit-corpora
mv dev-0/dev.en dev-0/in.tsv
mv dev-0/dev.pl dev-0/expected.tsv
mv train/train.en train/in.tsv
mv train/train.pl train/expected.tsv
mv test-A/general-scenario-test.en test-A/in.tsv
mv test-A/general-scenario-test.pl test-A/expected.tsv
mkdir test-B
mv test-A/constrained-scenario-test.en test-B/in.tsv
mv test-A/constrained-scenario-test.pl test-B/expected.tsv
mv kompendium.tsv glossary.tsv