Completed:
- Got LongLeaf account
- Set up Dreamscape project on GCP
- Put Dreamscape repo on virtual machine (issued by LongLeaf)
- Partially trained the model (terminated it, wasn’t happy with the configuration settings)
- Tested API calls
To-Do
- Fully train the model using GPUs
- Migrate training filed to GCP storage bucket
- Host the model on GCP
- Set up end points and APIs to the model
- Figure out how to fix storage issue
- Finish Test document
- Fix API calls (still getting 500 Internal Server Error)
Pitfalls/Issues
- Error message when training the model indicates that the VM machine doesnt have enough memory (Could have been consumed due to partially trained model? I’m not sure)