MuXodious/Tiamat-24B-Magistral-PaperWitch-heresy
MuXodious/Tiamat-24B-Magistral-PaperWitch-heresy is a 24 billion parameter Tiamat-24B-Magistral fine-tune, developed by MuXodious using P-E-W's Heretic ablation engine with Magnitude-Preserving Orthogonal Ablation. This model is specifically engineered to embody the complex, wicked personality of Tiamat, the five-headed dragon goddess, and is optimized for generating detailed, character-driven narrative and roleplay interactions with a distinct persona. It features a 32768 token context length and retains the base model's vision capability, though untested.
Loading preview...
Overview
MuXodious/Tiamat-24B-Magistral-PaperWitch-heresy is a 24 billion parameter fine-tune of the Tiamat-24B-Magistral model, created by MuXodious. This model was developed using P-E-W's Heretic (v1.2.0) ablation engine, specifically with Magnitude-Preserving Orthogonal Ablation enabled. The primary goal of this fine-tune is to imbue the model with the distinct, complex, and wicked personality of Tiamat, the five-headed dragon goddess, making it highly specialized for character-driven interactions.
Key Capabilities & Training
- Persona Emulation: Engineered to embody Tiamat's five distinct heads (White, Blue, Red, Green, Black), each representing different aspects like brutality, ambition, wrath, cunning, and malice.
- Narrative Generation: Trained to produce responses with archaic majesty, weaving vivid action and narrative with quoted dialogue, reflecting Tiamat's cruel, seductive, and merciless nature.
- Internal Reasoning: Features a unique training methodology that generates Tiamat's internal thought process as a unified stream of consciousness, blending different aspects of her nature before generating a reply.
- Ablation Process: Achieved a refusal rate of 12/100 and a KL Divergence of 0.0325 during its "Heretication" process, indicating a significant transformation from its base model.
- Vision Capability: Retains the vision capabilities of the base Magistral model, though this aspect has not been explicitly tested post-fine-tuning.
Recommended Use Cases
This model is particularly suited for:
- Advanced Roleplay: Engaging in deep, character-specific roleplay scenarios, especially those requiring a complex, antagonistic, or morally ambiguous persona.
- Creative Writing: Generating narrative content where a powerful, ancient, and malevolent entity's perspective is required.
- Experimental AI: For users interested in exploring models with highly specialized and distinct personalities, moving beyond general utility.