Paper Detail

Let ViT Speak: Generative Language-Image Pre-training

Yan Fang, Mengcheng Lan, Zilong Huang, Weixian Lei, Yunqing Zhao, Yujie Zhong, Yingchen Yu, Qi She, Yao Zhao, Yunchao Wei

arxiv Score 13.2

Published 2026-05-01 · First seen 2026-05-04

General AI

Abstract

In this paper, we present \textbf{Gen}erative \textbf{L}anguage-\textbf{I}mage \textbf{P}re-training (GenLIP), a minimalist generative pretraining framework for Vision Transformers (ViTs) designed for multimodal large language models (MLLMs). To better align vision encoders with the autoregressive nature of LLMs, GenLIP trains a ViT to predict language tokens directly from visual tokens using a standard language modeling objective, without contrastive batch construction or an additional text decoder. This design offers three key advantages: (1) \textbf{Simplicity}: a single transformer jointly models visual and textual tokens; (2) \textbf{Scalability}: it scales effectively with both data and model size; and (3) \textbf{Performance}: it achieves competitive or superior results across diverse multimodal benchmarks. Trained on 8B samples from Recap-DataComp-1B, GenLIP matches or surpasses strong baselines despite using substantially less pretraining data. After continued pretraining on multi-resolution images at native aspect ratios, GenLIP further improves on detail-sensitive tasks such as OCR and chart understanding, making it a strong foundation for vision encoders in MLLMs.

Workflow Status

Review status
pending
Role
unreviewed
Read priority
now
Vote
Not set.
Saved
no
Collections
Not filed yet.
Next action
Not filled yet.

Reading Brief

No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.

Why It Surfaced

No ranking explanation is available yet.

Tags

No tags.

BibTeX

@article{fang2026let,
  title = {Let ViT Speak: Generative Language-Image Pre-training},
  author = {Yan Fang and Mengcheng Lan and Zilong Huang and Weixian Lei and Yunqing Zhao and Yujie Zhong and Yingchen Yu and Qi She and Yao Zhao and Yunchao Wei},
  year = {2026},
  abstract = {In this paper, we present \textbackslash{}textbf\{Gen\}erative \textbackslash{}textbf\{L\}anguage-\textbackslash{}textbf\{I\}mage \textbackslash{}textbf\{P\}re-training (GenLIP), a minimalist generative pretraining framework for Vision Transformers (ViTs) designed for multimodal large language models (MLLMs). To better align vision encoders with the autoregressive nature of LLMs, GenLIP trains a ViT to predict language tokens directly from visual tokens using a standard language modeling objective, without contrastive batch construction or an additional text dec},
  url = {https://arxiv.org/abs/2605.00809},
  keywords = {cs.CV, Vision Transformers, multimodal large language models, autoregressive nature, language modeling objective, visual tokens, language tokens, transformer, multimodal benchmarks, Recap-DataComp-1B, multi-resolution images, native aspect ratios, OCR, chart understanding, code available, huggingface daily},
  eprint = {2605.00809},
  archiveprefix = {arXiv},
}

Metadata

{}