arXiv - CSCL: "Pit One Against Many: Leveraging Attention-head E…" - CleverLibre Social

arXiv - CSCL @arxiv_cscl@social.cleverlibre.org

Pit One Against Many: Leveraging Attention-head Embeddings for Parameter-efficient Multi-head Attention. (arXiv:2310.07911v1 [cs.CL])

http://arxiv.org/abs/2310.07911 #arXiv #NLProc

Oct 14, 2023, 03:17 · · arxiv-cscl · · ·

Sign in to participate in the conversation