Baidu's CRAZY Unlimited OCR 2 Reads 100 Pages in ONE Shot? I Ran It on an 8GB Laptop

Baidu's Unlimited OCR 2 claims it can parse a hundred-page document in a single request. I explain how it works (SR-SWA), go through the paper's benchmark numbers, and run it myself on an 8GB RTX 4060 laptop. What you will see: the paper's claims (authors' numbers, not mine), how to use the model, the Windows install bug I hit, and my real runs: 1 page, 4 pages in one request, and a 15-page request that did not finish on this laptop. Short test, easy document, so treat it as a first look, not a benchmark. Chapters: 0:00 The claim: 100 pages in one shot 0:30 What Unlimited OCR 2 is 0:57 SR-SWA explained 1:35 The scoreboard, honestly 2:07 Cross-page tables and long documents 2:36 How you use it 3:04 Can an 8GB laptop run it? Install bug 3:28 My test runs: 1 page, 4 pages, 15 pages 4:25 Who it is for 4:38 RunPod and your question My results (RTX 4060 Laptop 8 GB, Transformers path): 1 page: 95.8 s, peak 7.35 GB VRAM 4 pages in one request: 401 s, peak 7.29 GB VRAM 15 pages in one request: stopped by me after more than an hour, no output Word-level match against the PDF text, pages 1-4: 0.948 (rough check) Links: Repo: https://github.com/baidu/Unlimited-OCR Model: https://huggingface.co/baidu/Unlimited-OCR-2 Paper (PDF in the repo): https://github.com/baidu/Unlimited-OCR/blob/Unlimited-OCR-2/Unlimited-OCR-2.pdf If you are looking for cheap and reliable GPUs, I use RunPod: https://get.runpod.io/pe48 Note: the thumbnail image is AI-generated from my reference photo; the video itself has no AI avatar. Get GPUs Runpod: https://get.runpod.io/pe48 Get Hostinger: https://www.hostg.xyz/SHJiu Coupon Code "PROMPT" 10% discount 🔗 Connect with me: Email → [email protected] Ko-fi → https://ko-fi.com/promptengineer Patreon → https://www.patreon.com/PromptEngineer975 Book a Call → https://calendly.com/prompt-engineer48/call GitHub → https://github.com/PromptEngineer48 Twitter/X → https://twitter.com/prompt48 Website → https://promptengineer48.com/