Machine Learning in Elixir

Wrapping Up

In this chapter, you used pre-trained models to significantly improve your model performance over one trained from scratch. You made use of pre-trained models in Axon for feature extraction and fine-tuning, and you learned how to take most models from the Python ecosystem and use them in Elixir. Moving forward, you should always ask yourself if training a model from scratch is worth it, or if you can benefit from using a pre-trained model. More often than not, using a pre-trained model will save you time and improve the performance of your models.

In the previous few chapters, you learned quite a bit about deep learning and Axon. But you’ve been primarily focused on image data. Fortunately, many of the lessons you’ve learned in these ...