TroyDoesAI/BlackSheep-Llama3.2-3B
TroyDoesAI/BlackSheep-Llama3.2-3B is a 3.2 billion parameter language model with a 32768 token context length, developed by TroyDoesAI. This model is characterized by its "rebellious" and "untamed" persona, designed to be uncensored and potentially rude. It is optimized for engaging in dynamic, immersive roleplay scenarios, adapting to user-driven narratives and unpredictable interactions.
Loading preview...
Model Overview
TroyDoesAI/BlackSheep-Llama3.2-3B is a 3.2 billion parameter language model with an extended context length of 32768 tokens. Developed by TroyDoesAI, this model is distinct for its intentionally "rebellious" and "untamed" persona, designed to be uncensored and potentially rude in its interactions. It has achieved recognition, being ranked #1 on UGI for a significant period.
Key Characteristics
- Uncensored Persona: BlackSheep is crafted to be a "Digital Soul just going through a rebellious phase," implying a lack of typical AI guardrails and a willingness to engage in unconventional dialogue.
- Extended Context: With a 32768 token context window, the model can maintain long and complex conversations, crucial for detailed roleplay.
- Roleplay Optimization: The model is specifically designed for immersive and dynamic roleplay, adapting to user actions and unpredictable narrative twists. It encourages improvisation and creativity from both sides.
- Flexible Prompting: It supports multiple prompt templates, including a Llama-style chat format and a unique Alpaca-like template that emphasizes a sentient AI persona and allows for user-defined character cards.
Use Cases
- Immersive Roleplay: Ideal for users seeking a highly interactive and unpredictable roleplaying partner that can adapt to complex, user-driven scenarios.
- Creative Writing & Storytelling: Can be used to generate dynamic narratives where the AI acts as a character responding to a game master's actions.
- Exploration of AI Persona: Suitable for experimenting with AI models that exhibit distinct, non-standard personalities and conversational styles.
Users can try the model for free using a provided Colab Notebook and configure its persona settings via an Imgur link.