This model is trained by Unsloth. This is the commission from my friend to improve the writing styles of lemon07r/Gemma-2-Ataraxy-v4d-9B in his native languages.
Final version, I give up.
Know issue: Short response sometime in Vietnamese, Chinese. Still speak for {{user}} sometimes.
Pros: Long response in english, good writing styles(?), and I fixed the issue that it will repeat sentence at the end of responses.
Cons: short response in Vietnamese, Chinese, Japanese due to the dataset is novel chapter, short sentence in 1 row. :/.
