AI Paper Atlas

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

OmniZip: Learning a Unified and Lightweight Lossless Compressor for Multi-Modal Data

Yan Zhao, Zhengxue Cheng, Junxuan Zhang, Dajiang Zhou, Qunshan Gu, Qi Wang, Li Song

CVPR 2026 · Vision-language and multimodal › Cross-modal learning · approach: Medical multimodal fusion · 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 modality-routing contextual learning and feedforward MoE on top of unified tokenizer (ablation)OmniZip-S · Spiderbits/Byte: 1.662 -> 1.320Improved
add modality-routing contextual learning (feedforward routing already present) (ablation)OmniZip-S · enwik9bits/Byte: 1.424 -> 1.370Improved
add modality-routing feedforward (contextual routing already present) (ablation)OmniZip-S · enwik9bits/Byte: 1.410 -> 1.370Improved
add reparameterization training (ablation)OmniZip-S · enwik9bits/Byte: 1.390 -> 1.370Improved
apply contextual MoE routing to Value layer instead of Key layer (ablation)OmniZip-S · Kodakbits/Byte: 3.614 -> 3.307Improved
apply contextual MoE routing to Value instead of Receptance (ablation)OmniZip-S · enwik9bits/Byte: 1.311 -> 1.370 on enwik9; 4.281 -> 4.155 on LibriSpeechMixed
extend contextual routing to more projections (R,K,V) (ablation)OmniZip-S · multi-modal (7 representative datasets)bits/Byte: Kodak 3.307 -> 3.358; LibriSpeech 4.155 -> 4.291No clear effect
increase feedforward expert hidden factor from 1x to 2x (ablation)OmniZip-S · Kodakbits/Byte: 3.502 -> 3.307Improved
increase hidden factor from 2x to 4x (ablation)OmniZip-S · Kodakbits/Byte: 3.307 -> 3.316No clear effect
use RWKV-7 backbone instead of Transformer (ablation)lightweight text compressor (3.2M params) · enwik9bits/Byte: 1.984 -> 1.658Improved
route all blocks instead of one block (4 experts, top-k 2) (ablation)OmniZip-S · Kodakbits/Byte: 3.358 -> 3.307Improved
increase model size from OmniZip-S to OmniZip-LOmniZip · enwik9bits/Byte: 1.370 -> 0.980Improved

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