Naphula/Goetia-8B-v1 is an 8 billion parameter uncensored language model with a 32768 token context length, created by Naphula using the Linear DARE merge method. Built upon aifeifei798/DarkIdol-Llama-3.1-8B-Instruct-1.2-Uncensored, it is designed for creative narrative generation and roleplay, offering a distinct writing style compared to Mistral-based models. This model excels in producing imaginative and uncensored content, making it suitable for applications requiring high creative freedom.
Loading preview...
Goetia 8B v1: Uncensored Creative Narrative Model
Goetia 8B v1 is an 8 billion parameter language model developed by Naphula, specifically engineered for highly creative and uncensored narrative generation and roleplay. Utilizing the Linear DARE merge method, it is built upon aifeifei798/DarkIdol-Llama-3.1-8B-Instruct-1.2-Uncensored as its base, integrating weights from numerous other 8B models to achieve its unique characteristics. This model is fully uncensored, requiring no ablation or jailbreaks, and is noted for its distinct "attitude" and creative output, offering a different writing style compared to models based on the Mistral architecture.
Key Capabilities
- Uncensored Content Generation: Produces narratives and roleplay without content restrictions.
- High Creativity: Excels in generating imaginative and detailed stories for its size.
- Distinct Writing Style: Offers a unique output style, differentiating it from Mistral-based models.
- Mergekit Integration: Developed using the Linear DARE merge method, which proved effective for Llama models.
Good for
- Users requiring an uncensored model for creative writing, storytelling, and roleplay.
- Developers seeking an 8B model with a large context length (32768 tokens) for imaginative applications.
- Those who prefer an alternative to the typical writing style of Mistral-derived models.
- Environments where explicit or graphic content may be part of the desired narrative (with appropriate system prompt adjustments).