Shrey Singhal
Titles as the sources gave them. In 1 entry, with 0 checked quotations.
In the Base
Every entry they appear in.
AERO: Adaptive Ensemble-Disagreement Routing for Oracle Feedback in Sample-Efficient Online RLHF
According to the abstract, Shrey Singhal and colleagues propose AERO, an online RLHF method that asks for human-preference labels only where an ensemble of reward models disagrees and fills in the rest with synthetic labels, to cut labelling cost.