Beyond Truncation: Rethinking LLM Decoding as Ensemble Pruning
We introduce Mahalanobis-Ensemble Decoding (ME-Decoding), a novel Large Language Model (LLM) decoding framework that frames candidate token selection as ensemble pruning. Existing selection strategies rely predominantly on scalar probabilities, ignoring geometric semantic relationships and causing candidate redundancy....
Dun-Yao Xue, Cheng-Shuo Du, Zheng-Bo Wang et al.
· 0 citations