Cite
Notes
Only stored in your browser.
Attribution
LayerSkip: Enabling Early Exit Inference and Self-Speculative Decoding
arXiv 2024
from 1 papers
Ahmed A Aly
Ahmed Roman
Akshat Shrivastava
Basil Hosmer
Beidi Chen
Bilge Acun
Bram Wasti
Carole-Jean Wu
Diana Liskovich
Liangzhen Lai