DarkArtsForge/Paimon-24B-v2
DarkArtsForge/Paimon-24B-v2 is a 24 billion parameter language model built on the MistralForCausalLM architecture, merged using the DELLA method with mistralai/Mistral-Small-24B-Instruct-2501 as its base. This model is specifically designed for generating creative and uncensored narratives, including violent and graphic erotic content, and excels in roleplay scenarios. It offers a 32768 token context length, making it suitable for extended conversational and story generation tasks.
Loading preview...
Paimon-24B-v2: A Merged Model for Uncensored Narrative Generation
Paimon-24B-v2 is a 24 billion parameter language model developed by DarkArtsForge, created through a sophisticated merge of multiple pre-trained models using the DELLA merge method. Its base model is mistralai/Mistral-Small-24B-Instruct-2501, augmented with contributions from several other 24B models to enhance its specific capabilities.
Key Capabilities
- Uncensored Content Generation: Paimon-24B-v2 is explicitly designed to produce narratives and roleplay content that can include violent and graphic erotic themes. Users are advised to adjust their system prompts accordingly.
- Advanced Merging Technique: Utilizes the DELLA merge method, which allows for fine-grained control over how different models contribute to the final merge, as detailed in the provided audit reports.
- Mistral Architecture: Built upon the MistralForCausalLM architecture, providing a strong foundation for language understanding and generation.
- Extended Context Length: Supports a context length of 32768 tokens, enabling longer and more coherent interactions and story development.
Good For
- Creative Writing & Roleplay: Excels in generating detailed and immersive roleplay scenarios and creative narratives, particularly those requiring a lack of content filters.
- Exploration of Mature Themes: Suitable for applications where the generation of violent or graphic erotic content is intentionally desired and managed by the user.
- Users Requiring Specific Chat Templates: Optimized for use with the Mistral Tekken template for best results, ensuring optimal performance and output quality.