Summary
In this chapter, we outlined three major components of the ML development life cycle – training dataset creation, model development, and model deployment. We focused on the latter two, starting with development. First, we discussed the popularity of the foundational NN frameworks. Then, we focused on several model development topics – the ONNX universal model representation format, the TB monitoring platform, the TF Lite mobile development library, and mixed precision PyTorch training. Next, we discussed two basic scenarios for model deployment – a REST service as a Flask app and an interactive web app with Gradio.
This concludes this chapter and this book. I hope you’ve enjoyed the journey!
