AI Paper Atlas

Research map / Architectures and efficiency / Mixture of experts

REAP the Experts: Why Pruning Prevails for One-Shot MoE compression

Mike Lasby, Ivan Lazarevich, Nish Sinnadurai, Sean Lie, Yani Ioannou, Vithursan Thangarasa

ICLR 2026 · Architectures and efficiency › Mixture of experts · approach: Expert routing and specialization

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
prune experts with REAP instead of merging with HC-SMoESMoE compression · Qwen3-30B-A3B 50% codingcode average: 0.364 -> 0.541Improved
prune with REAP instead of merging with M-SMoESMoE compression · GLM-4.5-Air 50% mathmath average: 0.465 -> 0.857Improved
use REAP saliency instead of frequency-based pruning (ablation)SMoE pruning · Qwen3-Coder-480B 50% codingcode average: 0.010 -> 0.619Improved
use REAP instead of EAN pruning (ablation)SMoE pruning · Qwen3-30B 50% codingcode average: 0.530 -> 0.541Improved
use REAP instead of EAN pruning (ablation)SMoE pruning · Kimi-K2 50% codingcode average: 0.487 -> 0.624Improved
use REAP instead of EAN pruning (ablation)SMoE pruning · Kimi-K2 25% tool use BFCLv3overall accuracy: 0.652 -> 0.635Hurt
use merging instead of pruningSMoE compression · Qwen3-30B 50% multiple-choice averageMC accuracy: 0.503 -> 0.542Improved
calibrate on C4 instead of domain-specific code data (ablation)SMoE compression · code generationaccuracyHurt
restrict maximum merge cluster size to 32 (ablation)HC-SMoE · code generationaccuracyHurt
use REAP instead of EAN on Qwen3-30B math at 50% (ablation)SMoE pruning · Qwen3-30B 50% mathmath average: 0.864 -> 0.857No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy