this is what open-source looks like
3B total parameters & 500M activated, yet powerful enough to transcribe 40+ pages in one pass while keeping context intact. Meet Unlimited OCR!
The model achieved state-of-the-art performance on OmniDocBench.
this is what open-source looks like
3B total parameters & 500M activated, yet powerful enough to transcribe 40+ pages in one pass while keeping context intact. Meet Unlimited OCR!
Users are excited about Baidu's Unlimited-OCR release because its compact 3B size enables affordable self-hosting instead of costly API calls.
No Digg Deeper questions have been answered for this story yet.
Baidu just released Unlimited-OCR
this is what enterprise-saas-maxxing looks like
Mistral claims SOTA performance on OlmOCRBench, a popular optical character recognition benchmark, but that isn't the case.
We have a public leaderboard on @huggingface, where Mistral OCR 4 currently ranks #3, behind open models like Chandra OCR 2 by @datalabto
https://huggingface.co/baidu/Unlimited-OCR
Baidu just released Unlimited-OCR

@_akhaliq Unlimited OCR'ın dil modeli ile çalışmasının asıl avantajı burada — kütüphaneler ve arşivlerde yığın halinde yanlış tanınan belgeleri toplu düzeltebiliyor, önceki OCR motorlarının %30-40'lık hata oranını çok altına çekiyor.

@_akhaliq Cool

@_akhaliq It's 3B too... wow, my day job pays for expensive API calls, this could be self hosted. I wonder how it works with handwritten text?

@_akhaliq very cool!

@_akhaliq Really unlimited?

@_akhaliq everyone woke up and chose ocr