MODEL_NAME=OpenGVLab/InternVL2-8B SYSTEM_MESSAGE="You are a multimodal large language model with the ability to understand images. Answer questions concisely." MAX_NEW_TOKENS=1024 TOP_P=0.7 TEMPERATURE=0.2 MAX_INPUT_TILES=24 REPETITION_PENALTY=1.1 DEFAULT_PROMPT="Describe this image in detail."