| funasr/datasets/large_datasets/utils/tokenize.py | ●●●●● 补丁 | 查看 | 原始文档 | blame | 历史 |
funasr/datasets/large_datasets/utils/tokenize.py
@@ -56,10 +56,9 @@ if 'hw_tag' in data: if hw_config['pre_hwlist'] is not None and hw_config['pre_prob'] > 0: # enable preset hotword detect in sampling import pdb; pdb.set_trace() pre_index = None for hw in hw_config['pre_hwlist']: _find = text.find(hw) _find = data['text'].find(hw) if _find != -1: _find = text[:_find].count(" ") # bpe sometimes pre_index = [_find, _find + max(hw.count(" "), 1)]