We use LLaVA-v1.5, a extensively used open-sourced MLLM, as our base mannequin and practice it utilizing…