a bug about IdentityQuantizer

[Chimou0]

Thanks for your great work. I want to ask some question. I found the return value of its forward might miss one value when I train funcodec with IdentityQuantizer. If I use: return x, None, self.zero, None rather than return x, None, self.zero, my training procedure works well. So am I correct in making this change ?

Also, I want to know how can I reconstruct the audio if I use the IdentityQuantizer since it doesn't have the quantization process.

I would appreciate it if you could answer my questions.