Is static pruning superseded?

static pruning (Mixture-of-experts routing): cited as a baseline — critiqued by newer work, not yet beaten on a benchmark here. 2 paper(s) critique it, 0 beat it on benchmarks — #236 of 1370 most-superseded. Sub-problem: cluster led by LLaMA-MoE. Newer alternatives in the same sub-problem include DOT-MoE, ZEDA, ExpertWeaver, Mixture-of-Experts (MoE) Adaptation, Elastic Mixture-of-Experts (EMoE).

Mixture-of-experts routing

static pruning

Superseded baseline#236 of 1,370 most-superseded

Cited as a baseline — critiqued by newer work, not yet beaten on a benchmark here

2 papers critique it · 0 beat it on benchmarks

What papers say

Verbatim critique sentences, each from a paper that cites static pruning as a baseline.

Static pruning removes parameters based on input-agnostic importance metrics

— ToMoE: Converting Dense Large Language Models to Mixture-of-Experts through Dynamic Structural Pruning

“Consequently, applying static pruning leads to a drastic drop in the generated image quality due to the inability to accommodate such dynamic activation shifts.”

— Understanding and Harnessing Sparsity in Unified Multimodal Models

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.

DOT-MoEDOT-MoE: Differentiable Optimal Transport for MoEfication
Jun 1, 2026
ZEDAPost-Trained MoE Can Skip Half Experts via Self-Distillation
May 18, 2026
ExpertWeaverExpertWeaver: Unlocking the Inherent MoE in Dense LLMs with GLU Activation Patterns
Feb 17, 2026
Mixture-of-Experts (MoE) AdaptationUnderstanding and Harnessing Sparsity in Unified Multimodal Models
Dec 2, 2025
Elastic Mixture-of-Experts (EMoE)Elastic MoE: Unlocking the Inference-Time Scalability of Mixture-of-Experts
Sep 26, 2025