AI Paper Atlas

Research map / Vision-language and multimodal / Cross-modal learning

Symbiotic-MoE: Unlocking the Synergy between Generation and Understanding

Xiangyue Liu, Zijian Zhang, Miles Yang, Zhao Zhong, Liefeng Bo, Ping Tan

ECCV 2026 · Vision-language and multimodal › Cross-modal learning · approach: Multimodal fusion with missing modalities · 1 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 generative training with shared-expert bridgeSymbiotic-MoE vs Only_LM_MMU · MMLUaccuracy: 0.405 -> 0.507Improved
add generative training with shared-expert bridgeSymbiotic-MoE vs Only_LM_MMU · OCRBenchscore: 662 -> 768Improved
add shared expert (ablation)original VLM experts · MMLU at iteration 0accuracy: 0.460 -> 0.697Improved
merge Text and ViT into one understanding group (bimodal) instead of tripartite split (ablation)expert grouping · MMLU at iteration 0accuracy: 0.244 -> 0.601Improved
assign 96 experts to understanding instead of 32 (ablation)bimodal split · MMLU at iteration 0accuracy: 0.285 -> 0.601Improved
assign 96 experts to understanding instead of 64 (ablation)bimodal split · MMLU at iteration 0accuracy: 0.453 -> 0.601Improved
assign 96 experts to understanding instead of 86 (ablation)bimodal split · MMLU at iteration 0accuracy: 0.561 -> 0.601Improved
add warmup gradient shielding on shared experts (ablation)Symbiotic-MoE · MMLU and OCRBenchaccuracyImproved
use 1e-6 instead of 1e-4 learning rate for understanding experts (ablation)Symbiotic-MoE · MMLU and OCRBenchaccuracyImproved
add shared experts and routing disentanglement versus MoTSymbiotic-MoE vs MoT · COCO-30KFID: 19.87 -> 13.65Improved
replace standard MoE with MoT structural isolationMoT vs Standard MoE · MMLUaccuracy: 0.308 -> 0.392Improved
use independent QKV (Bagel) instead of MoTBagel vs MoT · T2I-CompBenchscore: 0.43 -> 0.45Improved

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