#fastdecoder 検索結果

"Decoding UTF8 with Parallel Extract - a nerd's dream. Branchless decoder, 29 instructions. It's compliant, sweet, and full source code available. #UTF8 #CodeNerd #FastDecoder" nrk.neocities.org/articles/utf8-…

A more serious thread on the DeepSeek-OCR hype / serious misinterpretation going on. 1. On token reduction via representing text in images, researchers from Cambridge have previously shown that 500x prompt token compression is possible (ACL'25, Li, Su, and Collier). Without…

🚨 DeepSeek just did something wild. They built an OCR system that compresses long text into vision tokens literally turning paragraphs into pixels. Their model, DeepSeek-OCR, achieves 97% decoding precision at 10× compression and still manages 60% accuracy even at 20×. That…

I asked DeepSeek OCR to convert my long context window to an image for compression and this is what it came up with.

Text or Pixels? It Takes Half — On the Token Efficiency of Visual Text Inputs in Multimodal LLMs (🎉 Accepted to EMNLP 2025 Findings!) 🤔We ask a simple question: If we render long text as a single image and feed it to an off-the-shelf multimodal LLM, can we cut decoder tokens…

DeepSeek-OCR looks impressive, but its core idea is not new. Input “Text” as “Image” — already explored by: LANGUAGE MODELING WITH PIXELS (Phillip et al., ICLR 2023) CLIPPO: Image-and-Language Understanding from Pixels Only (Michael et al. CVPR 2023) Pix2Struct: Screenshot…

This is the JPEG moment for AI. Optical compression doesn't just make context cheaper. It makes AI memory architectures viable. Training data bottlenecks? Solved. - 200k pages/day on ONE GPU - 33M pages/day on 20 nodes - Every multimodal model is data-constrained. Not anymore.…

誰や！？テキストエンコーディングフィルターいらんとかいったやつ！！ むっちゃ的確に情報落とせるやんけ！！

if i'm understanding this correctly, you can use a pure text encoder model to find text that lets you reconstruct an image from the text encoding. basically, the latent space of a text model is expressive enough to serve as a compilation target for images

🚀 DeepSeek-OCR — the new frontier of OCR from @deepseek_ai , exploring optical context compression for LLMs, is running blazingly fast on vLLM ⚡ (~2500 tokens/s on A100-40G) — powered by vllm==0.8.5 for day-0 model support. 🧠 Compresses visual contexts up to 20× while keeping…

DeepSeek just dropped a new OCR model! And this isn't about OCR. We've all heard "a picture is worth a thousand words." DeepSeek literally proved it. They've built a breakthrough in AI memory compression that could change how models handle long contexts. The core idea:…

How it works: convert the long context into one image; the multimodal LLM reads the image + short query, so the decoder sees k visual tokens instead of m text tokens → consistently reduces half the decoder tokens with no performance degradation; no fine-tuning of the VLMs.

Face Depixelizer Given a low-resolution input image, model generates high-resolution images that are perceptually realistic and downscale correctly. 😺GitHub: github.com/tg-bomze/Face-… 📙Colab: colab.research.google.com/github/tg-bomz… P.S. Colab is based on the github.com/adamian98/pulse

Precious.... #AIphoto Denoising strength: 0.55, Clip skip: 2, Hires upscale: 1.5, Hires steps: 15, Hires upscaler: Latent (bicubic antialiased)

DeepSeek-OCR is out! 🔥 my take ⤵️ > pretty insane it can parse and re-render charts in HTML > it uses CLIP and SAM features concatenated, so better grounding > very efficient per vision tokens/performance ratio > covers 100 languages

Pleased to announce fast image-to-image translation in the browser. With 3 trained models introduced. Also I released a processed dataset of 1000 images for edges2cats translation. Demo: zaidalyafeai.github.io/pix2pix/cats.h… Code: github.com/zaidalyafeai/z…

DeepSeek finally released a new model and paper. And because this DeepSeek-OCR release is a bit different from what everyone expected, and DeepSeek releases are generally a big deal, I wanted to do a brief explainer of what it is all about. In short, they explore how vision…

CA) 0x780c9f7b97d8521eb62b362f8d749e0d7a864444

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…

GaussianImage: 1000 FPS Image Representation and Compression by 2D Gaussian Splatting arxiv.org/pdf/2403.08551…