Make proper_bs.py a userbenchmark
Open
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 1k
- Forks
- 346
- PR merge metrics
- No merged PRs in 30d
Description
We would like to deploy it as a new model profiling CI
When ever user submit a new model in PR, run a new model CI that:
- profile it and upload the profiling result to Amazon S3
- post the profiling link to the GH Issue
- give suggestions on the proper batch size
- run latency stableness check, inference result stableness check
- for inference test, output dump flops numbers increase with batch size
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reading proper_bs.py and determine how it would become a userbenchmark entry point for model submissions. Map the requested CI workflow from profiling and Amazon S3 upload through GitHub Issue reporting, batch-size suggestions, and stability checks. Done means all five listed checks and outputs run automatically for a newly submitted model.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- aws, github, python, pytorch
- Domain
- ci-cd, cloud, machine-learning, performance
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 20/100