Depth-Aware Sensitivity Analysis of Mixture-of-Experts Models via Magnitude-Based Expert Masking

Excerpt

arXiv:2608.13565v1 Announce Type: new Abstract: Mixture-of-Experts (MoE) architectures scale large language models (LLMs) while preserving computational efficiency through sparse activation. Despite their widespread adoption, the relative importance of individual MoE layers remains insufficiently characterized, particularly for model compression. This paper presents a systematic layer-wise sensitivity analysis of the Qwen3.6-35B-A3B model (40 MoE layers, 256 experts per layer, top-8 routing) usi