SaisExperiments/RightSheep-Llama3.2-3B

This model does not reflect my own political beliefs, this is simply an experiment

This model works best with Llama-3 formatting

---

This model was fine-tuned (lora) on scale-lab/politune-right using KTO in LLaMA-Factory

---

This model is based on TroyDoesAI/BlackSheep-Llama3.2-3B. It uses two loras trained with different settings, I found adding both to the model resulted in an even more unhinged result.