Love your model but my machine gets so few tokens per second with 31b... Any possibility for the MOE gemma version to be trained?
· Sign up or log in to comment
