deepspeed
https://github.com/microsoft/deepspeed
Python
DeepSpeed is a deep learning optimization library that makes distributed training easy, efficient, and effective.
Triage Issues!
When you volunteer to triage issues, you'll receive an email each day with a link to an open issue that needs help in this project. You'll also receive instructions on how to triage issues.
Triage Docs!
Receive a documented method or class from your favorite GitHub repos in your inbox every day. If you're really pro, receive undocumented methods or classes and supercharge your commit history.
Python not yet supported8 Subscribers
Add a CodeTriage badge to deepspeed![CodeTriage badge for microsoft/deepspeed](/microsoft/deepspeed/badges/users.svg?count=8)
Help out
- Issues
- [BUG] deepspeed.utils.safe_get_full_grad get all nan value
- [BUG] Gradient accumulation causing training loss differences in Deepspeed vs FSDP
- [NaN check] Add NaN check to support bfloat16.
- [BUG] Training time regression with ZeRO-3 after upgrade to torch 2.3.1 and CUDA 12.1
- [BUG] Universal checkpoint conversion failed
- [BUG] Launcher does not honor CUDA_VISIBLE_DEVICES
- [BUG] Training with RoPE is broken: Can't stop FP32 layers from being cast to FP16/BF16 during training
- Multi node multi GPU sharding for inference / training Llama 405B
- when I finetune the model use deepspeed on 2 4*A800s,log only contain worker1
- Fix circular import in ds_transformer.py
- Docs
- Python not yet supported