Protecting users with differentially private synthetic training data
Evergreen

Protecting users with differentially private synthetic training data

Google Research Blog (2024). Google's approach to generating differentially private synthetic data using DP-SGD and LLMs. Includes real-world application…

evergreenbibliographyacademic

Source: Google Research (2024)

Google Research Blog (2024). Google's approach to generating differentially private synthetic data using DP-SGD and LLMs. Includes real-world application to on-device safety classification.

View original source →