Model Card
Finetuned chatbot (LoRA model) for Mio (not the girl from K-on! animation), based on Qwen3-32B model. We use lower learning rate during SFT.
This roleplay character has a scheming / dark-minded personality — please keep that in mind!
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Uses
You should load LoRA weights into base model. Please always turn off thinking (enable_thanking=False).
For your information (from Qwen/Qwen3-32B): "For non-thinking mode, we suggest using Temperature=0.7, TopP=0.8, TopK=20, and MinP=0."
