Research map / Vision-language and multimodal / Human and embodied modeling
Unified Multimodal Autoregressive Modeling with Shared Context—Visual Tokenizer is Key to Unification
Wujian Peng, Lingchen Meng, Yuxuan Cai, Xianwei Zhuang, Yuhuan Yang, Rongyao Fang, Chenfei Wu, Junyang Lin, Zuxuan Wu, Shuai Bai
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add reinforcement learning at 512 px (500 steps) (ablation) | UniAR · OneIG-EN | score: 71.1 to 84.0 | Improved |
| add 100 more RL steps at 960 px (ablation) | UniAR · OneIG-EN | score: 84.0 to 87.3 | Improved |
| add random bitwise visual index flipping during pretraining (ablation) | UniAR · image generation at higher temperatures | sample stability | Improved |
| condition decoder on multi-level features instead of single layer (ablation) | visual decoder · image reconstruction | reconstruction quality | Improved |
| use discrete visual tokens instead of continuous | UniAR · pre-training | time per iteration (s): 35.4 -> 24.5 | Improved |
| add decoder upsampling (x64 downsample, 256 tokens) (ablation) | UniAR (8B) · 1024px generation | time (s): 53.5 -> 13.0 | Improved |
| use UniAR tokenizer instead of COMP-SigLIP as vision encoder | LLM with Llama 3.0-8B · DocVQA | accuracy: 34.0 -> 38.0 | Improved |
| UniAR tokenizer versus COMP-SigLIP | LLM with Llama 3.0-8B · InfoVQA | accuracy: 26.0 -> 23.8 | Hurt |
| prompt rewriting | UniAR · GenEval | overall: 0.85 -> 0.86 | Improved |
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.
- OmniGen2 — OmniGen2: Towards Instruction-Aligned Multimodal Generation (CVPR 2026) · better on 1 of 1 shared result
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
- UVU: Improving Multimodal Understanding via Vision-Language Unified Autoregressive Paradigm (CVPR 2026 · similarity 0.91)
- MANZANO: A Simple and Scalable Unified Multimodal Model with a Hybrid Vision Tokenizer (ICLR 2026 · similarity 0.90)
- Janus: Decoupling Visual Encoding for Unified Multimodal Understanding and Generation (CVPR 2025 · similarity 0.90)
- Vision as a Dialect: Unifying Visual Understanding and Generation via Text-Aligned Representations (NeurIPS 2025 · similarity 0.90)
- TokenFlow: Unified Image Tokenizer for Multimodal Understanding and Generation (CVPR 2025 · similarity 0.89)
- Vision Foundation Models as Effective Visual Tokenizers for Autoregressive Generation (NeurIPS 2025 · similarity 0.89)
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.
