Cite
Notes
Only stored in your browser.
Attribution
MeKi: Memory-based Expert Knowledge Injection for Efficient LLM Scaling
arXiv 2026
Rethinking Optimization and Architecture for Tiny Language Models
arXiv 2024
Kangaroo: Lossless Self-Speculative Decoding via Double Early Exiting
from 3 papers
Yehui Tang
Kai Han
Yunhe Wang
Yunsheng Ni
Hyeonmok Ko
Kyeng-Hun Lee
Kyungrae Kim
Linji Hao
Ning Ding
researcher
Shangling Jui