Verdi March
Verdi March
Hi, I want to share an experimental / stop-gap work called [FrameworkProcessor](https://github.com/aws-samples/smallmatter-package/blob/main/src/smallmatter/sm.py#L111-L315), to simplify submitting a Python processing job with `requirements.txt`, `source_dir`, `dependencies`, and `git_config`, using SageMaker framework training containers...
hi, if this one is accepted, then #2564 is not needed.
I'd like to close and withdraw this PR, as it potentially changes the existing behavior of the pipeline API. In particular, the pipeline risks of using stale version of source...
I'm thinking to default to `data/` (that is, under current directory). Need to see how this will work under SageMaker training job.
In fact, why does a default value needed at all? What if callers must always specify the data location?
why need a custom container?
How does RLlib (say, on Tensorflow) container differ from Tensorflow container? I was wondering if it can be as simple as telling the estimator to use TF container and passing...
any chance you are using ipython-8?
Rebased to #316. Ready for review.
did you run on HyperPod?