-
Notifications
You must be signed in to change notification settings - Fork 26
Description
There are a number of dependabot PRs open on /ModelTraining code.
In Zulip, @pastorep said:
I'm a bit reluctant to bump any of these Python libs now because I don't plan to run this code anytime soon. I've actually moved all model dev to repo "orca-ml".
Let me think a bit more on how we can handle the deprecated code in this repo. I don't want to wipe it yet because it's how we trained our current model. But I don't want to maintain it appropriately (with dependabot changes, etc) because we're unlikely to run it again
Currently the /README.md file and the /ModelTraining/README.md contain no notes about this, so there appears to be a documentation issue.
And dependabot files dependency update PRs against them, so that can be excluded too for any deprecated code.
As for deleting code, since it's in a source code control system, it's never really deleted since one can go back to an older commit. However, we could create a branch like "old-model-training" or something to make it easy to find, and remove it from main so it needn't be built or tested by CI/CD, and would be excluded from things like dependabot, code scanning, etc that we want to run on the main branch.