setup_name	default
training_folder	/training
encoding	detect
word_regex	(?:\p{L}\p{M}*|[\'\-\^~=])+
annotator_id	
