ACE: Adaptive Calibration-Free Expert Skipping for MoE-based LLMs
Mixture-of-Experts (MoE) architectures provide an efficient paradigm for scaling large language models (LLMs), yet fixed top-k routing activates the same number of expert slots for every token, causing substantial redundant computation. Existing expert-skipping methods often rely on router confidence, calibration data, or additional training, and therefore cannot reliably estimate the actual contribution of routed experts. To this end, we propose ACE, a training-free, calibration-free, and checkpoint-preserving framework for token-adaptive expert skipping in MoE-based LLMs. ACE contains two co
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
Every edge carries a method, confidence, and the source snippet that justified it — so bad links are debuggable.
- PossiblePossibly related (embedding) · 64%Adaptive Mixture of Experts Gate (AMG) [R] →
- FuzzyOverlapping authors or contributors · 62%affaan-m/ECC →
“Shared author/contributor keys: jiang”
- FuzzyOverlapping authors or contributors · 62%BerriAI/litellm →
“Shared author/contributor keys: jiang”
- LinkedLinked via arxiv author · 85%Zukang Xu →
“ACE: Adaptive Calibration-Free Expert Skipping for MoE-based LLMs”
- LinkedLinked via arxiv author · 85%Zhixiong Zhao →
“ACE: Adaptive Calibration-Free Expert Skipping for MoE-based LLMs”
- LinkedLinked via arxiv author · 85%Xing Hu →
“ACE: Adaptive Calibration-Free Expert Skipping for MoE-based LLMs”
- LinkedLinked via arxiv author · 85%Jiangyong Yu →
“ACE: Adaptive Calibration-Free Expert Skipping for MoE-based LLMs”
- LinkedLinked via arxiv author · 85%Houji Wen →
“ACE: Adaptive Calibration-Free Expert Skipping for MoE-based LLMs”