1 paper · 1 filter
Keenan Samway, Daniel D'souza, Omar Darwiche Domingues +3
Multi-agent systems promise a new axis for scaling inference: rather than having one agent think longer or sample more attempts, a task is split across many agents that work in par…