Questions about model structure
0913ktg opened this issue · comments
Dear p0p4k,
I am currently engaged in research on Korean voice synthesis models and have been utilizing your well-crafted vits2_pytorch implementation for training a Korean model. It has been functioning exceptionally well.
While exploring your repository, I came across vits3_pytorch and attempted to discern the differences from vits2_pytorch, but couldn't pinpoint any specific changes. Would it be possible for you to update the readme with the modifications made in vits3_pytorch? If you haven’t made the changes yet, could you possibly share your plans regarding what alterations you intend to implement?
Your response would be greatly appreciated.
Thank you.
Hi, the ideas for vits3 are still not clear, cause I got sidetracked with other stuff. I will archive this repo for now. If you are planning on making some changes to vits2 and need help, let me know!
But my base idea was improve normalizing flow and add Lora.