#Fp4

NVIDIA

TransformerEngine

A library for accelerating Transformer models on NVIDIA GPUs, including using 8-bit and 4-bit floating point (FP8 and FP4) precision on Hopper, Ada and Blackwell GPUs, to provide better performance with lower memory utilization in both training and inference.

Tencent

AngelSlim

Model compression toolkit engineered for enhanced usability, comprehensiveness, and efficiency.