Hi, while I was trying the model I noticed that the model works better on prompts with one language only. When I combine english root prompts + dutch language context, it drops accuracy.
· Sign up or log in to comment
