crestf411/gemma2-9B-sunfall-v0.5.2 Overview
This model is an experimental, LoRA-fine-tuned version of the 9 billion parameter Google Gemma2-9B-IT base model, featuring a 16384 token context length. It is the second release of the v0.5 series, with an expanded dataset compared to v0.4. The model is specifically trained for immersive character roleplay and story generation, incorporating a unique 'Diamond Law' methodology.
Key Capabilities
- Immersive Roleplay: Trained to act as an "expert actor" fully immersing into a given character (
{{char}}) without breaking character. - Complex Storytelling: Capable of generating compelling stories based on detailed scenarios, character descriptions, and content tags.
- Contextual Adherence: Utilizes specific system prompts for roleplay and storywriting, including a 'Diamond Law' for guiding content generation.
- Lore Book Integration: Designed to work with lore book tags for enhanced contextual control, using a 'Status: Blue (constant)' setting.
Good For
- Character-driven AI applications: Ideal for scenarios requiring consistent character portrayal and dialogue.
- Creative Writing & Narrative Generation: Suited for generating stories, scenarios, and narrative content with specified themes and characters.
- Experimental AI Development: As a pre-release, it's suitable for developers interested in providing feedback on its unique training approach and capabilities in roleplay and storytelling.