Kedro is an open-sourced Python framework for creating maintainable and modular data science code.

Kedro

In kedro kubeflow, GPU resource specified in kubeflow.yaml is not be applied, even though the other cpu/memory settings are:
`data-science-candidate-modelling-pipeline-train-model-node:`
   `cpu: 8`
   `memory: 8Gi`
   `<http://nvidia.com/gpu|nvidia.com/gpu>: 1`

Then in the pod, the gpu specification is not there:
``` resources:
        limits:
          cpu: '8'
          memory: 8Gi
        requests:
          cpu: '8'
          memory: 8Gi```
What might cause this?

Um, maybe because the docs suggest it will work, but the code doesn’t know anything about <http://nvidia.com/gpu|nvidia.com/gpu>???
<https://github.com/getindata/kedro-kubeflow/blob/0.7.3/kedro_kubeflow/config.py#L181>

Oh, good catch - that’s a regression on our side. Could you please create an issue on GH and we will fix it?

Here you go:  <https://github.com/getindata/kedro-kubeflow/issues/199>

BTW, I don’t think that’s a regression.  I didn’t check every version but I didn’t see it implemented sampling through all the prior 0.x tags.  I was going to try monkeypatching it but I’m not sure how the to deal with the special characters in `<http://nvidia.com/gpu|nvidia.com/gpu>` that are illegal in python variables…