search

lucidrains / voicebox-pytorch

Implementation of Voicebox, new SOTA Text-to-speech network from MetaAI, in Pytorch
MIT License
589 stars 49 forks source link

expose attention dropout #21

Closed yzmyyff closed 11 months ago

yzmyyff commented 12 months ago

The default attention dropout in the duration model is 0.1.