ZH_style_base_model

Open WinterStraw opened this issue 2 years ago • 2 comments

Thank you very much for your contributions. This is a very useful and clear open-source project.

I noticed that it performed surprisingly well on English tasks. However, there seem to be some potential areas for improvement in the Chinese task, such as the lack of a style model that can adjust emotions.

Do you plan to release the relevant training process in the future, such as data and code? I want to try to fine-tune the Chinese base model and train the style-model. Thanks a lot.

Dec 29 '23 22:12 WinterStraw

We are indeed workinig on a new project that will opensource the data, with diverse emotion/styles, for both Chinese and English. Stay tuned!

Jan 02 '24 03:01 Zengyi-Qin

how to use?

Jan 04 '24 10:01 minguilong