A brief document detailing the development flow for working on models
- Select one or two models from the "candidates" section of MODELS.md
- Create relevant issues for your models
- Read relevant research, search for any code on the model
- Develop/train model
- Share the model output with the team. See if you can refine it, even if it meets everyones standards
- Create a docker container containing the model. Run the container locally and see if it works
- Open a PR
- Once PR is approved, deploy the model live to Azure Functions and update relevant spec files for the new model
- Go back to step 1 until all candidate models have been exhausted.