Research question
How should a production content-moderation classifier be built?
Key findings
- Taxonomy design, active learning and data-quality practices behind OpenAI's moderation endpoint - Addresses distribution shift and rare categories
A Holistic Approach to Undesired Content Detection in the Real World
The paper behind OpenAI's moderation API, documenting the practical pipeline for real-world undesired-content detection.
Contribute
Spot an error, or know a record we should add? Members can propose additions and submit corrections.