Radar ·
Baidu releases Unlimited OCR 2, a 3.3B open source model for long documents
Baidu announced Unlimited OCR 2, an open source 3.3B model for OCR of long documents, listed under the MIT license in the original post.
Baidu announced Unlimited OCR 2, an open source 3.3B model that, according to the authors' claim, processes documents of any size with nearly constant decoding latency and memory. The authors claim state-of-the-art results on the MPDocBench and MarathonDocBench benchmarks. The post indicates an MIT license. The description points to an OCR approach for long documents with limited memory, useful for multi-page PDFs or books.
This is a news item based only on the announcement text, with no additional technical details or independent measurements recorded here. To check it, consult the author's original publication and the repository documentation it links to, verifying the license and the results on the cited benchmarks. Relevance to Rota Nacional: the platform offers document extraction that passes through the same personal data detection barrier, and OCR jobs are in preview. This model is not a Rota feature.
Original source: Baidu lança modelo de OCR open source de 3,3B para documentos longos