Thea Collection A family of models based on Llama 3.2 including improved reasoning, trained with GaLoRE, rsLoRA and non-Unsloth code. • 4 items • Updated 6 days ago • 1