Open Source · MarkTechPost ·

Baidu Releases Unlimited OCR, a 3B Model That Keeps the KV Cache Flat for Long-Document Parsing

Baidu Releases Unlimited OCR, a 3B Model That Keeps the KV Cache Flat for Long-Document Parsing

Baidu open-sourced Unlimited OCR, a 3B-parameter MoE model for long-document parsing. It uses Reference Sliding Window Attention to keep KV cache usage flat as output grows and reports 93.23 on OmniDocBench v1.5 under an MIT license.

Read the full story at MarkTechPost →