Tag: 模型架构 (3 articles)

NVIDIA trained the Nemotron OCR v2 model on 12 million synthetic images, achieving high accuracy (NED as low as 0.035) and high speed (34.7 pages/second on a single A100 GPU) across six languages, demonstrating that synthetic data is a key solution to the multilingual data bottleneck in OCR.

Holotron-12B optimizes inference efficiency and handles long contexts,

Lilian Weng explores how AI models can enhance reasoning and decision-making