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.
Purpose and fit
Author description: 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.
Basic listing based on official public information; functionality has not been independently evaluated.
Sources: official source 1 · official source 2. Based on public basic information; not an in-depth evaluation.
- Browse by purpose
- Data analysis
- Platform
- Not verified
- Pricing
- Pricing not verifiedCheck current official software, hosting, model and infrastructure pricing.
Insufficient history
Repository info
Repository stars, not unique users or individual-skill installs
- Subcategory
- Data science & machine learning
- Platform
- Not verified
- First collected
- 2026-11-03
- Language
- Python
- License
- Apache-2.0 · Permissive
- Repository created
- 2022-09-20
- Last push
- 2026-11-02
- Measured at
- 2026-11-03 09:41 UTC
- Pricing
- Pricing not verified
- Cost details
- Check current official software, hosting, model and infrastructure pricing.
- Maintenance notes
- Not verified
License tiers are derived from metadata. Check obligations for your use; unknown does not mean prohibited.
Tags
Historical observations
Raw observations below are not daily gains.
| Measured at | stars | Source |
|---|---|---|
| 2026-11-03 09:41:07 UTC | 3,599 | Source |