Density-Guided Response Optimization: Community-Grounded Alignment via Implicit Acceptance Signals

[Invited Talk] Density-Guided Response Optimization: Community-Grounded Alignment via Implicit Acceptance Signals


Invited talk for the UMD Digital Twin reading group; discussed recent ACM FAccT paper, which uses a new methodology to recover latent community preferences for downstream alignment use.

It appears you don't have a PDF plugin for this browser. No biggie... you can click here to download the PDF file.