Can Foundation Models Moderate Online Content? Evaluating Instruction- vs. Example-Driven Policy Operationalization
Two competing paradigms for Vision-Language Model (VLM) guidance are compared: an instruction-driven approach where models reason from policy precepts, and an example-driven approach where they generalize from prior precedents.
Ayan Majumdar, Shounak Paul, Pushpdeep Singh et al.
· 2 citations