An LMM that can adaptively select the appropriate visual granularity based on the input image and instruction. conda create -n avg-llava python=3.10 -y conda activate avg-llava pip install --upgrade ...