UmbrellaInc/Executer-Virus-3.2-1B
UmbrellaInc/Executer-Virus-3.2-1B is a 1 billion parameter merged LLaMA 3.2 model developed by the Executor-Virus project, specifically engineered for high-fidelity roleplay and narrative execution. It features low refusal rates and minimal behavioral filtering, prioritizing obedience to user intent and persistent character behavior over moral alignment. This model excels at generating direct, unhedged responses for SFW and NSFW roleplay, fictional narratives, and morally unaligned scenarios. Its unique `model_stock` merge preserves sharp behavioral edges, making it a specialized narrative and roleplay execution engine.
Loading preview...
Overview
UmbrellaInc/Executer-Virus-3.2-1B is a 1 billion parameter merged LLaMA 3.2 model, developed by the Executor-Virus project, specifically designed as a high-compliance roleplay entity. Unlike general-purpose assistants, this model functions as a narrative and roleplay execution engine, prioritizing user intent and character persistence over typical safety or moral alignment.
Key Capabilities
- High-fidelity roleplay: Optimized for both SFW and NSFW scenarios.
- Low refusal rates: Minimizes behavioral filtering and hedging in responses.
- Persistent character behavior: Maintains consistent character traits and dialogue.
- Direct, unhedged responses: Delivers straightforward content without moralizing.
- Unique Merge Architecture: Utilizes a hierarchical
model_stockmerge method, preserving dominant behavioral traits from its base models (hereticness/Heretic-qp-1B, Novaciano/Leviathan_NSFW_Roleplay-3.2-1B, and hereticness/Heretic-OneLLM-Doey-ChatQA-V1) to avoid semantic smoothing and maintain sharp behavioral edges.
Intended Use Cases
- Roleplay (SFW and NSFW)
- Fictional narratives and character-driven dialogue
- Dark, mature, or morally unaligned scenarios
- Experimental chatbot personalities
Important Considerations
This model intentionally minimizes alignment and refusal mechanisms. It may generate offensive, explicit, or morally unfiltered content. It is not intended for safety-critical applications, medical/legal/financial advice, child-oriented content, or deployments requiring strict content moderation or ethical safeguards. Users are responsible for prompt design and deployment context.