What is the use for mask_inverse

NVIDIA / flowtron

Flowtron is an auto-regressive flow-based generative network for text to speech synthesis with control over speech variation and style transfer

https://nv-adlr.github.io/Flowtron

Apache License 2.0

889 stars 177 forks source link

Closed dariocazzani closed 4 years ago

dariocazzani commented 4 years ago

In FlowtronLoss the inverse of the mask is computed but never used. Is it a legacy or am I missing something?

Thanks

rafaelvalle commented 4 years ago

Thank you for pointing this out. It was probably left there from some previous implementation. It has been removed.