deepspeedai

All

6 repositories

DeepSpeed
Public
DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.
machine-learning compression deep-learning
machine-learning compression deep-learning gpu inference pytorch zero data-parallelism model-parallelism mixture-of-experts
Python
•
Apache License 2.0
•4.7k•42k•1.2k•120•Updated Mar 10, 2026Mar 10, 2026
DeepSpeedExamples
Public
Example models using DeepSpeed
Python
•
Apache License 2.0
•1.1k•6.8k•300•26•Updated Mar 4, 2026Mar 4, 2026
Megatron-DeepSpeed
Public
Ongoing research training transformer language models at scale, including: BERT & GPT-2
Python
•
Other
•3.7k•2.2k•139•22•Updated Aug 14, 2025Aug 14, 2025
DeepSpeed-MII
Public
MII makes low-latency and high-throughput inference possible, powered by DeepSpeed.
deep-learning inference pytorch
deep-learning inference pytorch
Python
•
Apache License 2.0
•190•2.1k•198•11•Updated Jun 30, 2025Jun 30, 2025
DeepSpeed-Kernels
Public
C++
•
Apache License 2.0
•19•71•1•0•Updated Mar 26, 2025Mar 26, 2025
deepspeed-gpt-neox
Public
An implementation of model parallel autoregressive transformers on GPUs, based on the DeepSpeed library.
Python
•
Apache License 2.0
•1.1k•21•0•1•Updated Nov 28, 2022Nov 28, 2022