Search
Has the Age of Lightweight Models Dawned? Baidu's 0.07B-Parameter OCR Mini-Model Surpasses GPT-4o and Qwen2.5-VL
2 day ago /
Read about 0 minute
Author：小编
Recently, Baidu unveiled its newest, ultra-compact optical character recognition model, PP-OCRv5, on the international stage. Despite having a mere 0.07 billion parameters, this model delivers OCR accuracy on par with that of massive models boasting 70 billion parameters, utilizing just a thousandth of the parameter count. In numerous tests, it has even outperformed versatile visual large-scale models like GPT-4o.
-
C114 Communication Network -
Communication Home
7 X 24 Track global technological trends
