1 paper · 2 filters
Linzhi Peng, Hanting Chen, Heng Chang +3
Large language model search agents are often trained with synthetic questions whose difficulty is increased through larger evidence graphs, additional hops, and longer trajectories…