Hello. I’m newbie. I use DeepSpeech2 for korean speech recognition.
I have question. When i don’t use “\s” second index 1 instead of the last index, it is different loss.
Does The index on the label affects the model’s predictions?
Hello. I’m newbie. I use DeepSpeech2 for korean speech recognition.
I have question. When i don’t use “\s” second index 1 instead of the last index, it is different loss.
Does The index on the label affects the model’s predictions?