This is a fine-tuned version of Granite-4 trained on 120 fun chat samples to create a playful, friendly chatbot that mimics models like GPT-4o but often leans overly positive and rarely responds seriously.
This is the first version of any FunChat model. In future versions I aim to make it be able to give longer more nuanced response and switch from serious to fun. I also intend to finetune bigger models.
