facebookresearch/fairscale
A PyTorch extension library for high-performance, large-scale distributed model training.
Not currently ranked — collecting fresh signals.
star history
FairScale extends PyTorch with state-of-the-art distributed training techniques. It provides composable modules and easy-to-use APIs for scaling neural network training with limited compute resources. The library focuses on performance optimization through techniques like model parallelism, gradient accumulation, and mixed precision training.
Frequently asked
- What is facebookresearch/fairscale?
- A PyTorch extension library for high-performance, large-scale distributed model training.
- Is fairscale open source?
- Yes — facebookresearch/fairscale is an open-source project tracked on heatdrop.
- What language is fairscale written in?
- facebookresearch/fairscale is primarily written in Python.
- How popular is fairscale?
- facebookresearch/fairscale has 3.4k stars on GitHub.
- Where can I find fairscale?
- facebookresearch/fairscale is on GitHub at https://github.com/facebookresearch/fairscale.