AI Paper Atlas

Research map / Architectures and efficiency / Mixture of experts

MoDES: Accelerating Mixture-of-Experts Multimodal Large Language Models via Dynamic Expert Skipping

Yushi Huang, Zining Wang, Zhihang Yuan, Yifu Ding, Ruihao Gong, Jinyang Guo, Xianglong Liu, Jun Zhang

CVPR 2026 · Architectures and efficiency › Mixture of experts · approach: Efficient MoE language model systems · 18 citations (OpenAlex)

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
add GMLG (global layer importance) to single-threshold skipping (ablation)Thresholding · ChartQA, skip 83% expertsaccuracy: 76.74 -> 79.28Improved
add GMLG to DMT (ablation)DMT · MMBench, skip 83% expertsaccuracy: 79.42 -> 81.44Improved
use separate text/vision thresholds (DMT) instead of single threshold (ablation)Thresholding · MMBench, skip 83% expertsaccuracy: 65.48 -> 79.42Improved
add GMLG and DMT together (ablation)Thresholding · VideoMMMU, skip 83% expertsaccuracy: 40.33 -> 47.11Improved
add GMLG to DMT (ablation)DMT · MME, skip 83% expertsscore: 2081 -> 2162Improved
use frontier search instead of naive search (ablation)MoDES · 13 benchmarks, skip 83%Avg. relative score (%): 96.24 -> 96.25No clear effect
calibrate and search on COCO instead of GQA (ablation)MoDES · ChartQA, skip 83% expertsaccuracy: 84.20 -> 86.56Mixed
use MoDES instead of MC-MoE at 83% skippingKimi-VL-A3B-Instruct · 13 benchmarksAvg. relative score (%): 88.32 -> 96.25Improved
use MoDES instead of MC-MoE at 88% skippingQwen3-VL-MoE-30B-A3B · 13 benchmarksAvg. relative score (%): 86.66 -> 97.33Improved
use MoDES instead of DiEP at 50% skippingKimi-VL-A3B-Instruct · 13 benchmarksAvg. relative score (%): 98.17 -> 99.91Improved
combine MoDES with 1.5-bit quantization instead of MC-MoEKimi-VL-A3B-Instruct · ChartQA, skip 67%accuracy: 69.46 -> 72.28Improved
skip experts with MoDES instead of default top-kKimi-VL-A3B-Instruct · RealWorldQA, skip 50%accuracy: 65.36 -> 65.62Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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