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 supported11 Subscribers
Add a CodeTriage badge to deepspeed
Help out
- Issues
- [BUG] Can't perform gradient accumulation on pipeline parallelism with inputs of different lengths
- [BUG] Failed to checkpoint with deepspeed 0.12.4
- [BUG] The forward hook function of Zero stege3 might misjudge the current step of the training process whether it is in the forward computation or in the backward recomputation
- [BUG]
- Deepspeed fails with frozen weights (e.g. only train llama2 embedding layer)
- [REQUEST] Autocast for BF16
- [BUG] `DeepSpeedCPUAdam` compatibility on ROCm
- Support FP16 CpuAdam + Zero Stage 3
- 能否添加对chatglm3系列模型的支持?
- The training parameters have not changed
- Docs
- Python not yet supported