Summary

In this chapter, we focused on deep learning. In particular, we covered the key theoretical concepts behind deep learning, before moving on to discuss how to build and train your own neural networks. We walked through examples of using off-the-shelf models for inference and then adapting them to your specific use cases through fine-tuning and transfer learning. All of the examples shown were based on heavy use of the PyTorch deep learning framework and the Hugging Face APIs.

We then moved on to the topical question of the largest models ever built, LLMs, and what they mean for ML engineering. We explored a little of their important design principles and behaviors before showing how to interact with them in pipelines using the popular LangChain package and OpenAI APIs. We also explored the potential for using LLMs to help with improving software development productivity, and what this will mean for you as an ML engineer.

We finished the chapter with an exploration of...