ShyliaSafetensors/EnceladusHyperStock-24B
EnceladusHyperStock-24B by ShyliaSafetensors is a 24 billion parameter experimental language model, built using advanced merge methods like Model Stock and Dare-Ties. Optimized for complex scene generation, creative storytelling, and long-form roleplay, it features a 32768 token context length. This model excels at generating creative, uncensored content and following complex instructions, including visible reasoning via a dedicated tag.
Loading preview...
EnceladusHyperStock-24B: An Experimental Merged LLM
EnceladusHyperStock-24B is an experimental 24 billion parameter language model developed by ShyliaSafetensors. It is a multimerge model, representing an evolution from the previous 'VeryWeirdDolphinMix-V1-24B-Heretic', designed to enhance handling of complex scenes, creative story-writing, and consistent role-play. The model leverages sophisticated merge techniques, specifically Model Stock and Dare-Ties, combining several base models to achieve its capabilities.
Key Capabilities & Features
- Complex Scene & Story Generation: Engineered for intricate narratives and creative writing.
- Long-form Roleplay: Optimized for extended, consistent, and uncensored roleplay scenarios.
- Visible Reasoning: Supports a unique
<thinking>tag for explicit, visible reasoning during complex instruction following. - Uncensored Content Generation: Intended for adult, research, fictional story, and creative uses, capable of producing uncensored output.
- High Context Length: Features a 32768 token context window, suitable for detailed and lengthy interactions.
Recommended Use Cases
- Long-form Uncensored Roleplay: Ideal for collaborative storytelling and character-driven interactions.
- Creative Writing: Suited for generating imaginative and uncensored textual content.
- Complex Instruction Following: Particularly effective when requiring the model to show its thought process using the
<thinking>tag. - Conversational AI: Can be used to develop AI with nuanced personality depth.