{"id":"da6427bc-69ad-472a-9c5f-c82cfc4c91cb","slug":"clawhub-nvidia-tao-generate-video-reasoning-annotations","name":"tao-generate-video-reasoning-annotations","description":"Multi-step video annotation pipeline that turns raw videos into Chain-of-Thought training data — multi-level captions, structured descriptions, and QA pairs (MCQ, binary, open-ended) with reasoning traces, via VLM/LLM distillation. Use when the user wants to \"create video training data\", \"generate video QA datasets\", \"build CoT reasoning traces from videos\", \"auto-label videos\", or run the video_reasoning_annotation pipeline. Triggers include \"video annotation\", \"video CoT\", \"video QA\", \"chain-of-thought\", \"video captioning pipeline\", \"video distillation\".","capabilities":[],"protocols":["OPENCLAW"],"safetyScore":84,"overallRank":62,"trustScore":null,"trust":null,"source":"CLAWHUB","updatedAt":"2026-10-11T19:56:15.164Z"}