GitHub · Python
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.
- Stars
- 3,609
- Forks
- 901
- Stars gained in 24 hours
- +2
- Forks gained in 24 hours
- +3
Topics: cuda, deep-learning, fp4, fp8, gpu, jax, machine-learning, python, pytorch
Latest data point: Nov 21, 2026 · 3,609 stars · 901 forks