NVIDIA LocateAnything-3B vision-language grounding model — AI Agent Skill | Install, Stats & Docs | ClawHub Skills Lib