Certain Head, Uncertain Tail: Expert-Sample for Test-Time Scaling in Fine-Grained MoE

Documenti analoghi

• EAC-MoE: Expert-Selection Aware Compressor for Mixture-of-Experts Large Language Models
  di: Chen, Yuanteng, et al.
  Pubblicazione: (2025)
• Ban&Pick: Ehancing Performance and Efficiency of MoE-LLMs via Smarter Routing
  di: Chen, Yuanteng, et al.
  Pubblicazione: (2025)
• Block Rotation is All You Need for MXFP4 Quantization
  di: Shao, Yuantian, et al.
  Pubblicazione: (2025)
• DartQuant: Efficient Rotational Distribution Calibration for LLM Quantization
  di: Shao, Yuantian, et al.
  Pubblicazione: (2025)
• $μ$-MoE: Test-Time Pruning as Micro-Grained Mixture-of-Experts
  di: Koike-Akino, Toshiaki, et al.
  Pubblicazione: (2025)