Skip to content

Explore: differential privacy training (DP-SGD) #33

Description

@NotYuSheng

Train so the model provably memorises less — directly attacking the central leakage risk of training on personal chats.

Directions: DP-SGD / DP fine-tuning; privacy/utility trade-off vs. the style eval; combine with redaction. Part of the exploratory roadmap.

Metadata

Metadata

Assignees

No one assigned

    Labels

    explorationExploratory research direction

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions