upgrading repo to latest version
diff --git a/DeReKo/spacy_train/conll2spacy.py b/DeReKo/spacy_train/conll2spacy.py
index f0d0d4c..9be14ab 100644
--- a/DeReKo/spacy_train/conll2spacy.py
+++ b/DeReKo/spacy_train/conll2spacy.py
@@ -4,17 +4,32 @@
if __name__ == "__main__":
"""
- --- TIGER New Orthography ---
- python DeReKo/spacy_train/conll2spacy.py --corpus_name TigerNew --gld_token_type CoNLLUP_Token \
- -i /home/daza/datasets/TIGER_conll/data_splits/train/Tiger.NewOrth.train.conll \
- -o DeReKo/spacy_train/Tiger.NewOrth.train.json \
- -t DeReKo/spacy_train/Tiger.NewOrth.train.txt
-
- python DeReKo/spacy_train/conll2spacy.py --corpus_name TigerNew --gld_token_type CoNLLUP_Token \
+
+ --- TIGER NEW Orthography ---
+ python DeReKo/spacy_train/conll2spacy.py --corpus_name TigerNew --gld_token_type CoNLLUP_Token \
+ -i /home/daza/datasets/TIGER_conll/data_splits/train/Tiger.NewOrth.train.conll \
+ -o DeReKo/spacy_train/Tiger.NewOrth.train.json \
+ -t DeReKo/spacy_train/Tiger.NewOrth.train.txt
+
+ python DeReKo/spacy_train/conll2spacy.py --corpus_name TigerNew --gld_token_type CoNLLUP_Token \
-i /home/daza/datasets/TIGER_conll/data_splits/test/Tiger.NewOrth.test.conll \
-o DeReKo/spacy_train/Tiger.NewOrth.test.json \
-t DeReKo/spacy_train/Tiger.NewOrth.test.txt
+ --- TIGER NEW + OLD Orthography ---
+ cat Tiger.OldOrth.train.conll Tiger.NewOrth.train.conll > Tiger.ALL.Orth.train.conll
+ cat Tiger.OldOrth.test.conll Tiger.NewOrth.test.conll > Tiger.ALL.Orth.test.conll
+
+ python DeReKo/spacy_train/conll2spacy.py --corpus_name TigerALL --gld_token_type CoNLLUP_Token \
+ -i /home/daza/datasets/TIGER_conll/data_splits/train/Tiger.ALL.Orth.train.conll \
+ -o /home/daza/datasets/TIGER_conll/data_splits/train/Tiger.ALL.Orth.train.json \
+ -t /home/daza/datasets/TIGER_conll/data_splits/train/Tiger.ALL.Orth.train.txt
+
+ python DeReKo/spacy_train/conll2spacy.py --corpus_name TigerALL --gld_token_type CoNLLUP_Token \
+ -i /home/daza/datasets/TIGER_conll/data_splits/test/Tiger.ALL.Orth.test.conll \
+ -o /home/daza/datasets/TIGER_conll/data_splits/test/Tiger.ALL.Orth.test.json \
+ -t /home/daza/datasets/TIGER_conll/data_splits/test/Tiger.ALL.Orth.test.txt
+
"""
parser = argparse.ArgumentParser()