HiFloat4 Format for Language Model Pre-training on Ascend NPUs

Mehran Taghian, Yunke Peng, Xing Huang +22

Large foundation models have become central to modern machine learning, with performance scaling predictably with model size and data. However, training and deploying such models i…