cupdike

Output Files That Exists Beget A FileNotFoundError

10

The training node tries to save a model file to Minio but reports FileNotFound when the file does exist at the intended path. To Reproduce Configure the pipeline to mount...

Make default registry a configuration item

1

I am reporting: - feature request Consider making the default registry a configuration item rather than embedding it in the code. That is, put datbase.org into a config file so...

Readme Does Not Disclose Kale's Dependency on Arrikto EKF

3

I tried running Kale against plain Kubeflow and hit authentication issues. When I asked Arrikto pre-sales about this, I was told that Kale now depends on EKF to make Kubeflow...

Sharded Dataset Has Long Delay Before First Batch (and caching=False)

8

I have a sharded dataset for 1000 large images (675 MB each). The shards were capped at 3 GB, and each shard contains 7 images (143 shards). Using WebDataset with...

I'm not a go dev, but when I asked about a problem I was having with the `go get` command, I was told it should be: `go get github.com/pusher/k8s-auth-example` (without...

Resource nvidia.com/gpu Not Handled by ResourceConfig

3

In kedro-kubeflow 0.7.3, if gpu's are assigned in `kubeflow.yaml` using an `nvidia.com/gpu` element, they are not applied to the node as expected in the generated pipeline since they are not...

In the Spaceflight Tutorial (spaceflights starter, v0.18.3), `kubeflow.yaml` has unqualifed node names like `model_training` which don't align with the actual namespaced node names defined in the pipeline--such as `data-science-active-modelling-pipeline-train-model-node`. Accordingly,...