Run training on Google Cloud

[florianschwanz]

Add a possibility to run training in the cloud in order to reduce needed time. Make sure that

• training session can be started via Github actions
• cloud resources are stopped after training
• results are uploaded to Firebase storage or Google Drive or a Google Cloud bucket

Follow these tutorials

• https://medium.com/codex/how-to-use-gsutil-and-python-to-deal-with-files-in-google-cloud-storage-fc4f430b3b28

Solution A - Run on Google Compute Engine

This approach includes starting a VM and running the training session on it

Solution B - Run on Google AI Platform aka Vertex AI

This approach includes using Vertex AI to run workloads.

Follow these tutorials

• https://cloud.google.com/ai-platform/docs
• https://cloud.google.com/ai-platform/training/docs/getting-started-pytorch
• https://cloud.google.com/vertex-ai/docs/start/introduction-unified-platform
• https://towardsdatascience.com/training-a-model-on-google-ai-platform-84ceff87b5f3

[florianschwanz]

• created Google Cloud Platform account for bikepath.quality@gmail.com
• project bike-path-quality with ID bike-path-quality-328520
• added service account with roles
  • Storage Object Admin
  • Storage Admin
  • Compute Instance Admin
  • Service Account User
• used service account key for Python code