OpenBMB releases DSpark draft head to speed up MiniCPM-OCR 1.0 decoding

AIOpenBMB releases the DSpark draft head for MiniCPM-OCR 1.0, a speculative decoding model that proposes tokens for the 1.3B main model to verify, improving end-to-end processing efficiency by 55%. The release reaches about 1.49× the throughput of PaddleOCR-VL-1.6 and 3.62× that of MinerU2.5 Pro.

Read the original huggingface.co

Open source

Source details

OpenBMB (MiniCPM) · new models on Hugging Face · huggingface.coPublished Oct 26, 4:46 AM ET · added here Oct 27, 12:36 PM ET

Original title: openbmb/MiniCPM-OCR-1.0-DSpark

Topics

Open-source ecosystem Deployment & engineering Data & training China AI Model releases

OpenBMB releases DSpark draft head to speed up MiniCPM-OCR 1.0 decoding

Topics

Tags