Jinpeng Wang

Humans see text — but LLMs don’t. I wrote a short blog post exploring how models can perceive text visually rather than tokenize it: 🔗 csu-jpg.github.io/Blog/people_se… From PIXEL, CLIPPO, VisInContext, VIST to DeepSeek-OCR, this is a quick story of how vision-centric modeling is changing how machines read, and a reflection on some of our own small efforts in the past two years.

Adapt the Qwen-Image model to a target domain with one single example.

Thanks AK for sharing our work. We speed up Qwen-image and Flux Inference speed from 50 steps to less than 10steps. 1 sample is enough for speeding up diffusion model in specific domain. Paper Link: arxiv.org/abs/2512.02899 Code Link: github.com/CSU-JPG/Glance

Glance Accelerating Diffusion Models with 1 Sample

@xyz2maureen 👏👏👏

I will join Tsinghua University, College of AI, as an Assistant Professor in the coming month. I am actively looking for 2026 spring interns and future PhDs (ping me if you are in #NeurIPS). It has been an incredible journey of 10 years since I attended an activity organized by Tsinghua University and decided to change my undergraduate major from Economics to Computer Science, inspired by one of the teammates. During the 10 years, I met with appreciation of many wonderful researchers/professors who led me to continued growth. 🐿️ My research focus will continue to be AI & Robotics, with a specific emphasis on Interactive Embodied Intelligence. You can check my homepage to learn more: maureenzou.github.io/lab.html. I am currently local to San Diego and will be attending #NeurIPS. Please ping me over WeChat or Email if any old or new friends are interested in having a coffee chat! (Really looking forward to meeting as many friends as possible at #NeurIPS) [The photo is one of the places that I will miss a lot in the US]

We added Gemini3-Pro #Gemini results on VCode! Check out at csu-jpg.github.io/VCode/ 🏅Gemini3-Pro performs quite well at image2svg. It achieved 52.1 in VCode, surpassing GPT-5's 46.8 (+5.3). Open-sourced at github.com/CSU-JPG/VCode

thanks JIQIZHIXIN for sharing our work. We use visual encoder to compress long text.

@iclr_conf Get three 🥚 all with confidence 5； 😃 unique life experience. 😃 One guy forgot to remove the GPT call logs.

Unique avatar 🤣🤣🤣😂😂😂 Try demo at: huggingface.co/spaces/CSU-JPG…

🤣🤣🤣🤣🤣

Turn any interesting imgs into svg🤣🤣🤣🤣🤣

Thanks @_akhaliq sharing our work! 🚀 Glad to introduce our newest work — VCode! 🎨 VCode: A Multimodal Coding Benchmark with SVG as Symbolic Visual Representation For decades, RGB pixels have been the default medium for representing images. But in the agentic era, how can we move beyond raw pixels toward interpretable, executable, and symbolic visual representations? 🔍 We address this question with VCode, which reframes visual representation as SVG code — aligning with how humans reason over sketches and symbolic abstractions. 👥 Happy to collaborate with: Yuhao Zheng, Hangyu Ran, Dantong Zhu, Dongxing Mao, @LINJIEFUN , @philiptorr and @awinyimgprocess 📄 arXiv: arxiv.org/pdf/2511.02778 🌐 website: csu-jpg.github.io/VCode/ 💻 github: github.com/CSU-JPG/VCode 🤗 Huggingface daily paper: huggingface.co/papers/2511.02…

@KieraFields7 Already reply in my csu email.

@awinyimgprocess Hi Jinpeng, I'm a reporter for The Stack. I'm looking to chat to experts in OCR and vision compression. I can't seem to find your email on your website. Please reach out if you're interested! kiera@thestack.technology

Humans see text — but LLMs don’t. I wrote a short blog post exploring how models can perceive text visually rather than tokenize it: 🔗 csu-jpg.github.io/Blog/people_se… From PIXEL, CLIPPO, VisInContext, VIST to DeepSeek-OCR, this is a quick story of how vision-centric modeling is changing how machines read, and a reflection on some of our own small efforts in the past two years.

Post your paper title here if missing in this blog.

The idea "Compress Text With Visual Token". "already proposed in our NeurIPS2024 work <>, NeurIps 2025 work <> (arxiv 25'2) arxiv.org/abs/2406.02547 arxiv.org/pdf/2502.00791 The idea of processing text as image already proposed in a series of works: 1. LANGUAGE MODELLING WITH PIXELS； ICLR’23. 2. CLIPPO: Image-and-Language Understanding from Pixels Only； CVPR’23 3. Improving Language Und

BOOOOOOOM! CHINA DEEPSEEK DOES IT AGAIN! An entire encyclopedia compressed into a single, high-resolution image! — A mind-blowing breakthrough. DeepSeek-OCR, unleashed an electrifying 3-billion-parameter vision-language model that obliterates the boundaries between text and vision with jaw-dropping optical compression! This isn’t just an OCR upgrade—it’s a seismic paradigm shift, on how machines perceive and conquer data. DeepSeek-OCR crushes long documents into vision tokens with a staggering 97% decoding precision at a 10x compression ratio! That’s thousands of textual tokens distilled into a mere 100 vision tokens per page, outmuscling GOT-OCR2.0 (256 tokens) and MinerU2.0 (6,000 tokens) by up to 60x fewer tokens on the OmniDocBench. It’s like compressing an entire encyclopedia into a single, high-definition snapshot—mind-boggling efficiency at its peak! At the core of this insanity is the DeepEncoder, a turbocharged fusion of the SAM (Segment Anything Model) and CLIP (Contrastive Language–Image Pretraining) backbones, supercharged by a 16x convolutional compressor. This maintains high-resolution perception while slashing activation memory, transforming thousands of image patches into a lean 100-200 vision tokens. Get ready for the multi-resolution "Gundam" mode—scaling from 512x512 to a monstrous 1280x1280 pixels! It blends local tiles with a global view, tackling invoices, blueprints, and newspapers with zero retraining. It’s a shape-shifting computational marvel, mirroring the human eye’s dynamic focus with pixel-perfect precision! The training data? Supplied by the Chinese government for free and not available to any US company. You understand now why I have said the US needs a Manhattan Project for AI training data? Do you hear me now? Oh still no? I’ll continue. Over 30 million PDF pages across 100 languages, spiked with 10 million natural scene OCR samples, 10 million charts, 5 million chemical formulas, and 1 million geometry problems!. This model doesn’t just read—it devours scientific diagrams and equations, turning raw data into a multidimensional knowledge. Throughput? Prepare to be floored—over 200,000 pages per day on a single NVIDIA A100 GPU! This scalability is a game-changer, turning LLM data generation into a firehose of innovation, democratizing access to terabytes of insight for every AI pioneer out there. This optical compression is the holy grail for LLM long-context woes. Imagine a million-token document shrunk into a 100,000-token visual map—DeepSeek-OCR reimagines context as a perceptual playground, paving the way for a GPT-5 that processes documents like a supercharged visual cortex! The two-stage architecture is pure engineering poetry: DeepEncoder generates tokens, while a Mixture-of-Experts decoder spits out structured Markdown with multilingual flair. It’s a universal translator for the visual-textual multiverse, optimized for global domination! Benchmarks? DeepSeek-OCR obliterates GOT-OCR2.0 and MinerU2.0, holding 60% accuracy at 20x compression! This opens a portal to applications once thought impossible—pushing the boundaries of computational physics into uncharted territory! Live document analysis, streaming OCR for accessibility, and real-time translation with visual context are now economically viable, thanks to this compression breakthrough. It’s a real-time revolution, ready to transform our digital ecosystem! This paper is a blueprint for the future—proving text can be visually compressed 10x for long-term memory and reasoning. It’s a clarion call for a new AI era where perception trumps text, and models like GPT-5 see documents in a single, glorious glance. I am experimenting with this now on 1870-1970 offline data that I have digitalized. But be ready for a revolution! More soon. [1] github.com/deepseek-ai/De…

"Huamn mnid deos not raed ervey lteter by istlef, but the wrod as a wlohes (Rawlinson, 1976)." – Graham Rawlinson 🚀 We propose new work: “See the Text: From Tokenization to Visual Reading” ，which renders text as images and uses a vision-centric pipeline instead of traditional subword tokenization. • Up to 4.4× fewer tokens & ~70% less FLOPs • Better cross-lingual transfer & robustness to typos/surface noise • Works on not only long-text compression, but also VQA and classification in NLP. 👉 arXiv: arxiv.org/pdf/2510.18840

@james_im I write a blog csu-jpg.github.io/Blog/people_se… (People See Text, But LLM Not).

I don't know about y'all but I mainly use vision to input text into my brain