Search papers, labs, and topics across Lattice.
University of Massachusetts
1
0
2
Implicit feedback from user interactions can boost LLM alignment accuracy by nearly 20%, revealing a hidden layer of user preference that explicit feedback misses.
