LLaVA-Plus maintains a skill repository that contains a wide range of vision and vision-language pre-trained models (tools), and is able to activate relevant to

LLaVA-Plus maintains a skill repository that contains a wide range of vision and vision-language pre-trained models (tools), and is able to activate relevant tools, given users’ multimodal inputs, to compose their execution results on the fly to fulfill many real-world tasks.