ML@CMU built a year-long video caption pipeline with 100+ professional creators, documenting the development process and outcomes. The project taught the team that scaling supervision, rather than scaling models, was the key lesson for improving caption production and workflows.
Key Points
- 1WHAT: ML@CMU built a year-long video-caption pipeline with 100+ professional creators.
- 2WHY: Experience highlighted scaling supervision over model changes to improve caption data and processes.
- 3SO WHAT: Practitioners should prioritize scalable supervision and creator workflows for video-caption systems.
Scoring Rationale
Practical, hands-on case study with significant relevance for teams building captioning pipelines and data workflows; moderate impact for the broader AI research community.
Sources
Public references used for this report.
Practice with real Hotels & Lodging data
90 SQL & Python problems · 15 industry datasets
Used by DS/ML engineers at top companies
250 free problems · No credit card
See all Hotels & Lodging problems
