Open
Description
Background of the issue:
- We have showcase demo scripts for how to use Cerebros and demonstration of SOTA results.
- We want to use the same scripts for benchmarks in our CICD testing, but have a limit on resources for GitHub Actions.
- We don't want to maintain 2 separate scripts, one for showcase use and one for CICD tests.
Proposed solution:
- Parameterize whether the script it is running for CICD only.
- Option 1:
- Create an environment variable in the Github runner like
CICD_TEST
, then make all the Python scripts that demonstrate how to use Cerebros, read in this variable, but default it to False if the variable does not exist. - If the execution environment the script runs in has the environment variable CICD_TEST set to true, then a small subset of the data is run in the training jobs. If the variable is set absent or set to false, then the full data set runs.
- Create an environment variable in the Github runner like
- Option 2:
- Use fire.Fire or argparse to run the job, adding a shell arg:
--CICD_TEST
, which if set will trigger the small subset run.
- Use fire.Fire or argparse to run the job, adding a shell arg:
Metadata
Metadata
Assignees
Labels
No labels