dhodun

Correct. The main use case where I have used `skaffold delete` is failed deployments and looking to start afresh. Even if there is a slightly different version of the code,...

It looks like if any of the `Cleanup` runs on any of the deployers fails, we quit return the error and exit the loop: https://github.com/GoogleContainerTools/skaffold/blob/0d7d66946fcac997f5d0fcb0fd96afb42d89d821/pkg/skaffold/deploy/deploy_mux.go#L161-L173 So in this case if...

+1 We use FireO with FastAPI.

This appears to happen regardless whether the subclassed Model has a "Meta" class defined, so it's not a matter of it being overwritten.

Running into this issue as well.

Interested in this as well. We recently removed the need for namespaces in our app which gave us a lot of flexibility to point skaffold at per-developer and per-PR namespaces....

I wrote up a blog post and example notebook on how to do this in TF 2.x: https://towardsdatascience.com/how-to-extend-a-keras-model-5effc083265c https://github.com/GoogleCloudPlatform/training-data-analyst/blob/master/blogs/batch_predictions/batch_predictions_keras.ipynb

+1 - this would be helpful when debugging CloudBuild concurrency issues. Currently we have to make dummy edits to files across multiple service to trigger rebuilds.

Thanks @mldangelo ! Might take a stab at it. In general, should I follow the test case generation pattern for simple python read? https://github.com/promptfoo/promptfoo/blob/e5cf9dcd0cbc5f0d315ac9bc446d4252049512d6/src/util/testCaseReader.ts#L106-L117 which uses https://github.com/promptfoo/promptfoo/blob/26e10e9547139d51c41aadeb2642e58b59bfbeec/src/python/pythonUtils.ts#L87 This would probably...

Actually, I see dynamic test cases was just added. Seems like this could be expanded upon more generically? https://github.com/promptfoo/promptfoo/pull/2993/files