in scripts_mgenre/preprocess_sentencepiece.py [0:0]
def encode(self, line):
global sp, old2new
ids = sp.encode_as_pieces(line)
if old2new:
ids = [x for old in ids for x in old2new[old]]
if self.args.offset > 0:
ids = [x + self.args.offset for x in ids]
return list(map(str, ids))