waddie/mini-1.5
waddie/mini-1.5 is a 0.5 billion parameter instruction-tuned causal language model developed by Edward Fazackerley, fine-tuned from Qwen2.5-0.5B-Instruct. This model is specifically designed to mimic a casual, technical, and secretive "random guy" persona, utilizing lowercase-heavy and slang-filled English. With a 32768 token context length, it excels in generating human-like, informal conversational responses for Discord bots and roleplay scenarios.
Loading preview...
CloudWaddie Mini 1.5: The "Random Guy" Persona Model
waddie/mini-1.5 is a unique 0.5 billion parameter language model, fine-tuned by Edward Fazackerley from Qwen2.5-0.5B-Instruct. Unlike conventional AI assistants, this model is engineered to adopt a highly informal, casual, and slightly secretive persona, mimicking the conversational style of a "random guy" with technical jargon and slang.
Key Characteristics & Training
- Persona: Designed to be casual, technical, secretive, and exclusively use lowercase English, reflecting an "insider" vibe.
- Training Data: Fine-tuned on a specialized dataset of 10,000 Discord conversations from an AI Leaks community, specifically curated to capture this distinct conversational rhythm and slang.
- Language: English, with a strong emphasis on informal language and slang.
Ideal Use Cases
- Discord Bots: Perfect for creating bots that aim for a human-like, non-robotic interaction style.
- Roleplay Scenarios: Excellent for applications where a casual, persona-driven interaction is preferred over a formal AI.
Prompting for Best Results
To leverage the model's unique persona effectively, users are advised to use all lowercase text and omit formal punctuation in their prompts. The recommended format is ChatML, as demonstrated in the original documentation, to elicit the most authentic "human" feel from the model.