Impressive improvement compared with v0.1

#3
by RyanZhu - opened

The improvement of v2.0 is so significant compared with v0.1. Since both models refer to the same paper. Could you share some insights about what crucial changes of v2.0 lead to these impressive results? Could you also share some information about the dataset size for v2.0, and whether you have any insights about the impact of data domain?

Thank!

Sign up or log in to comment