tikr7
tikr7
Hi there! I installed the sparkmonitor extension into a dockerimage that is based on jupyterhub: ``` FROM jupyterhub/k8s-singleuser-sample:1.2.0 RUN pip install pyspark==3.2.0 RUN pip install delta-spark==1.1.0 USER root RUN apt...
According to this [post](https://github.com/SHI-Labs/OneFormer/issues/30), it took: > For Swin-L OneFormer, the training times are roughly as follows: > ADE20K: ~1.5 days on 8xA6000 GPUs. > COCO: ~8.5 days on 8xA100...
The Knative stack (eventing and serving) is pretty awesome. Overall it makes infra easy to use for our devs and ML engineers. On Kuberntes I don't wanna miss it anymore....
### Long story short In case the kopf-operator has been **restarted**, it seems old CR's are not really under control anymore. I cant't delete the CR's because they still have...
**Problem** [JobSinks](https://knative.dev/docs/eventing/sinks/job-sink/) are used for long-running jobs/tasks and are nicely embedded in the Knative-eventing. It "simply" creates a K8s Job per Event and mounts it into `/etc/jobsink-event/event`. We use it...