models/attention/decoders/attention_layer.py

hirofumi0810 / tensorflow_end2end_speech_recognition

End-to-End speech recognition implementation base on TensorFlow (CTC, Attention, and MTL training)

MIT License

314 stars 121 forks source link

Open cuhkebook opened 6 years ago

cuhkebook commented 6 years ago

In the class AttentionLayer (models/attention/decoders/attention_layer.py), the initial parameter does not include "sigmoid_smoothing".

But, in models/attention/attention_seq2seq.py, it calls AttentionLayer with sigmoid_smoothing=self.sigmoid_smoothing (line 338).

Which file should I modify? Thx!

CyanHillFox commented 6 years ago

Encountered the same problem too. So what did you do? I added an parameter "sigmoid_smoothing", and assigned default value to it, will it be ok?