Naphula/Ancient-Awakening-12B
Naphula/Ancient-Awakening-12B is a 12 billion parameter language model with a 32768 token context length, created by Naphula through a complex multi-stage merge of 70 diverse Mistral Nemo finetunes and other merged models using advanced techniques like karcher, flux, arcee_fusion, ramplus_tl, and pdq. This model is specifically optimized for rich, long-form narrative generation and roleplay, leveraging its extensive merge history to produce highly creative and detailed outputs.
Loading preview...
Ancient Awakening 12B: A Deeply Merged LLM for Advanced Roleplay
Ancient Awakening 12B is a 12 billion parameter language model developed by Naphula, distinguished by its exceptionally complex, multi-stage merging process. It integrates 70 different Mistral Nemo finetunes and other specialized merges, utilizing advanced techniques such as karcher, flux, arcee_fusion, ramplus_tl, and pdq via mergekit. This intricate architecture is designed to enhance the model's capabilities in generating rich, detailed, and long-form narratives.
Key Capabilities
- Complex Narrative Generation: Excels at creating intricate storylines and detailed world-building.
- Advanced Roleplay: Optimized for engaging in sophisticated and immersive role-playing scenarios.
- Extensive Context Window: Supports a 32768 token context length, ideal for maintaining coherence over long interactions.
- Diverse Knowledge Integration: Benefits from the fusion of 70 distinct donor models, contributing to a broad and nuanced understanding.
Good for
- Creative Writing & Storytelling: Users seeking a model for generating fiction, scripts, or detailed lore.
- Immersive Role-Playing: Developers and enthusiasts focused on creating highly interactive and character-driven experiences.
- Experimental Merging Research: Those interested in exploring the results of advanced, multi-method model merging.
Note: For users requiring an uncensored experience, the MPOA version is available, which has guardrails removed and can produce graphic content. The model works best with ChatML or Mistral Tekken chat templates.