debug

shixian.shi

2023-06-27 35ea75e9e564624909fbc51300087e503aa7504a

debug

 funasr/datasets/large_datasets/utils/tokenize.py

@@ -58,9 +58,10 @@
            # enable preset hotword detect in sampling
            pre_index = None
            for hw in hw_config['pre_hwlist']:
                _find = data['text'][0].find(hw)
                hw = " ".join(seg_tokenize(hw))
                _find = text.find(hw)
                if _find != -1:
                    _find = data['text'][0][:_find].count(" ")  # bpe sometimes
                    _find = text[:_find].count(" ")  # bpe sometimes
                    pre_index = [_find, _find + max(hw.count(" "), 1)]
                    import pdb; pdb.set_trace()
                    break