Search papers, labs, and topics across Lattice.
Fudan University, Ant International, Ant Group
1
0
2
Next-turn user reactions can boost multi-turn agent performance by over 10 percentage points, revealing the critical role of local feedback in reinforcement learning.
