TransformerEngine
✅ Recently maintainedA 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.
3,599Stars
885Forks
+57-day stars
Python · Apache-2.0 · Last push 2026-10-31
Official repository & installation ↗Star history
2025-09-292026-11-02
View historical data
| Date | Stars |
|---|---|
| 2026-10-04 | 3,565 |
| 2026-10-05 | 3,566 |
| 2026-10-06 | 3,567 |
| 2026-10-07 | 3,569 |
| 2026-10-08 | 3,570 |
| 2026-10-09 | 3,571 |
| 2026-10-10 | 3,572 |
| 2026-10-11 | 3,573 |
| 2026-10-12 | 3,575 |
| 2026-10-13 | 3,576 |
| 2026-10-14 | 3,578 |
| 2026-10-15 | 3,579 |
| 2026-10-16 | 3,582 |
| 2026-10-17 | 3,583 |
| 2026-10-18 | 3,585 |
| 2026-10-19 | 3,585 |
| 2026-10-20 | 3,587 |
| 2026-10-21 | 3,587 |
| 2026-10-22 | 3,589 |
| 2026-10-23 | 3,593 |
| 2026-10-24 | 3,593 |
| 2026-10-25 | 3,593 |
| 2026-10-26 | 3,596 |
| 2026-10-27 | 3,597 |
| 2026-10-28 | 3,597 |
| 2026-10-29 | 3,597 |
| 2026-10-30 | 3,597 |
| 2026-10-31 | 3,597 |
| 2026-11-01 | 3,598 |
| 2026-11-02 | 3,599 |
Get the project
git clone https://github.com/NVIDIA/TransformerEngine.git
Related projects & alternative candidates
Related domains do not imply feature parity. Review official documentation and licenses.
LLMs-from-scratch
Step-by-step guide to building a GPT-like LLM from scratch in PyTorch.
★ 107,671rayRay is an AI compute engine. Ray consists of a core distributed runtime and a set of AI Libraries for accelerating ML workloads.
★ 44,174transformersCore framework for loading and training state-of-the-art text, vision, audio, and multimodal models, for inference and training.
★ 168,432