DDI training compared to not DDI training

jaywalnut310 / glow-tts

A Generative Flow for Text-to-Speech via Monotonic Alignment Search

MIT License

651 stars 150 forks source link

DDI training compared to not DDI training #73

Open cantabile-kwok opened 1 year ago

cantabile-kwok commented 1 year ago

Hi! I am curious about why you use DDI (data-dependent initialization) here, as not doing DDI won't cause a bug in the program. So how is the performance of not using DDI at the beginning? Does it have a specific use?

cantabile-kwok commented 1 year ago

Also, where is the source of this method? I found a paper (https://arxiv.org/pdf/1511.06856.pdf) but it does not seem to be the implementation used in this repo. Appreciate any discussions!