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”.
Lines: 186 | Code: 21 | Dir: 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…