Colab link: https://colab.research.google.com/github/fastai/fastbook/blob/master/02_production.ipynb
Drivetrain approach
Production Classifier
Steps:
- Data
- Gather data
- Verify
- Create data block (variable types, path to data, transforms, etc...) & loader
- Tweak
- Create learner (loader + model + metric)
- Quick train
- Create interpreter & plot (e.g. confusion matrix, top/bottom losses)
- Clean data
- Full train
- Export & deploy (steps depend on application)