Luni/StarDust-12b-v2
Luni/StarDust-12b-v2 is a 12 billion parameter language model, merged using the DARE TIES method with Sao10K/MN-12B-Lyra-v3 as its base. This model is specifically optimized for creative story writing and long-form role-playing, producing vibrant and less generic prose. It excels in scenarios requiring imaginative narrative generation and character impersonation, offering a distinct alternative to models with KTO training.
5.0 based on 1 review
Hugging Face Quickly became my favorite 12B model, surpassing many 72B. Very good coherence in my replies, good enough memory and a surprisingly low amount of slop.
Roleplay