aws / aws/sagemaker-training-toolkit
consider using Horovod for distributed training
Open
priority: medium
related: Horovod
type: enhancement
- Dominant language
- Python
- Stars
- 530
- Forks
- 140
- Avg merge
- 1h 12m
- Merged PRs (30d)
- 2
Description
- How does [Horovod](https://github.com/horovod/horovod) differ from our existing implementation?
- What other implementations of MPI can we use to improve performance?
Contributor guide
Research direction
The issue names no files, tests, or entry points. Start by locating the existing distributed-training implementation and compare its behavior and performance with Horovod and other MPI implementations. Done would require a decided implementation or documented recommendation, but the issue does not define that scope.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- aws, python
- Domain
- distributed-systems, machine-learning, performance
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100