Search papers, labs, and topics across Lattice.
Inria, École Normale Supérieure, PSL Research University, University of California, Berkeley
Berkeley AI Research (BAIR)2
0
3
Achieve 13-15% more efficient LLM watermark detection by using e-values for anytime-valid inference, enabling early stopping without sacrificing statistical guarantees.
Forget global worst-case guarantees – this new multi-objective learning method actually adapts to distribution shifts and delivers unbiased predictions in dynamic environments.
