Aratako/Ninja-v1-RP-expressive: Enhanced Roleplay Model
This 7 billion parameter model, developed by Aratako, is specifically designed for expressive roleplay scenarios. It builds upon the existing Aratako/Ninja-v1-RP model by integrating expressive capabilities from derivatives of Elizezen/Antler-7B, a novel generation model.
Key Capabilities & Features
- Enhanced Expressiveness: Merged with models like Elizezen/Antler-7B derivatives to improve the richness and detail of character expressions and narrative descriptions in roleplay.
- Roleplay Optimization: Specifically fine-tuned for generating dynamic and engaging roleplay interactions, focusing on character consistency and emotional depth.
- Complex Merging Strategy: Utilizes
mergekit with advanced techniques such as Task Vector addition (incorporating models like senseable/WestLake-7B-v2, SanjiWatsuki/Kunoichi-DPO-v2-7B, SanjiWatsuki/Silicon-Maid-7B, and SanjiWatsuki/Loyal-Macaroni-Maid-7B) and DARE TIES for precise model integration. - Vicuna Prompt Format: Employs the Vicuna chat template, supporting detailed system prompts for character and world-building, and requiring
eos_token for multi-turn conversations.
Good For
- Detailed Roleplay Generation: Ideal for users seeking to generate rich, expressive, and character-driven roleplay narratives.
- Interactive Storytelling: Suitable for applications requiring an AI to embody specific characters with defined personalities and speaking styles.
Limitations
- Due to the merging process, the model may occasionally generate user-side dialogue or shift into a more novel-like writing style. This can often be mitigated through few-shot prompting or regeneration.