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 supported18 Subscribers
Add a CodeTriage badge to deepspeed
Help out
- Issues
- Warn when zero.Init silently falls back to a single rank (#8084)
- zero.Init silently does not shard (world_size=1) when the process group is uninitialized before from_pretrained -> per-rank full load -> OOM
- zero.Init partitioning of large fused MoE-expert tensors spikes a single GPU (transient full materialization) -> OOM during load even when the sharded model fits
- fix: use local ev_values and wrap dict.values() in list()
- ZeRO 1/2: wait on all IPG-bucket producer streams in average_tensor (#8061)
- fix: add buffer-length check in shm.cpp
- fix: sanitize subprocess call in ds_aio_job.py
- Avoid CUDA context initialization during op compatibility checks at import
- [BUG]ds_z3_config error
- Add type hints to public API functions
- Docs
- Python not yet supported