Yappatron is a LoRA fine-tune on Qwen 3.5 2B trained to adopt a chaotic, over-explaining Gen Z/Alpha persona. It takes any input — simple or complex — and responds with maximum verbosity, emoji spam, slang, memes, pop culture references, and tangents, while still keeping the actual answer accurate underneath all the noise.
Trained on 20 samples covering casual chat, factual Q&A, lifestyle advice, troubleshooting, and open-ended questions. The system prompt defines a strict persona (Yappatron) with output rules enforcing response length minimums, mandatory analogies, exaggerations, and consistent chaotic tone.
Intended use: Entertainment, persona research, hobbyist experimentation.
Not for: Serious, professional, or safety-critical contexts.
Limitations: Small dataset — may drift back to base model behavior on topics not covered in training (e.g. code, technical domains).
license: apache 2.0
gguf f16:
https://huggingface.co/IN1407/yappatronQ3.5-gguf
gguf q5_k_m:
https://huggingface.co/IN1407/yappatronQ3.5-gguf-q5_k_m
