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 supported9 Subscribers
Add a CodeTriage badge to deepspeed
Help out
- Issues
- Question: recommended method for inference from a pipelined model checkpoint?
- [BUG] zero stage > 0 does not support the case when different iterations see different learnable parameters
- [BUG] tensor values change across different stages of pipeline parallel
- CUDA_VISIBLE_DEVICES isn't correctly inherited on a SLURM system
- [BUG] Host Memory Efficiency
- Warmup Schedueler is Not Linear warmup, Increase in Log curve, inscrease Too Fast for big model
- DeepSpeed Inference with FP16 (torch.half) -- RuntimeError: "softmax_lastdim_kernel_impl" not implemented for 'Half'
- Can't access gradients if using ZERO
- removing per node repeated debug/info prints on distributed setup
- Why ZeRO-2 use more CUDA Memory than ZeRO-1?
- Docs
- Python not yet supported