C4W3 TFX on Google Cloud Vertex Pipelines - long install time

Course Q&A Machine Learning Engineering for Production(MLOps) Deploying Machine Learning Models in Production
1

I was finally able to access the lab, but the Setup for the vertex_pipelines_simple.ipynb file is taking so long I am running out of lab time (set at 1 hour 30 min). Is there way around it - perhaps with stricter constraints in the pip install line?

2

Please highlight this to qwiklabs via their help channel.

3

Hi Marcus,

I had the exact same problem: first no way to launch the lab then stuck on the install step.

For anyone facing the same issue, I replaced the first cell with:

!pip install --upgrade pip
!pip install tfx==1.2.0
!pip install kfp kfp-server-api --upgrade
!pip install pip install google-cloud-aiplatform --upgrade

And it did the job for me. Note that I cancelled the previous installation so this might not be fully reproducible. pip also warned about broken dependencies, but I was able to finish the lab anyways.

For the record, my experience with these labs was truly awful, not in terms of content but due to the non stop technical issues

Good luck since it seems it’s down to that to pass these labs

1 Like