Bumblebee Training: Training your Pre-trained Models

Katelynn Burns


Follow my lead and we’ll become the queen bees if machine learning! Using pre-trained models can be a great way to integrate machine learning without having to start from scratch. Using a model right out of the box though leads to an overwhelmed model and developer. Let’s teach our models what data we’re actually interested in to help smooth the process. We’re going to be using a pre-trained bumblebee model and giving it our own data to push it in the right direction. Training out data doesn’t have to be hard. Let’s make it as easy as one, two, Bumblebee.

Key takeaways:

  • This talk is meant to ease the process of integrating machine learning into a project. Machine learning is a very popular topic right now and there is a lot of potential for combining machine learning and Elixir. Pre-trained models are a great entry point into machine learning, but they can be easily overwhelmed. This talk aims to teach techniques for honing in pre-trained models to focus on data specific to a given project.


  • Software engineers who are interested in learning about machine learning but who do not want to create their own models. This is meant to be an introductory course for people early on in their machine learning journey.

Level: Intermediate

Tags: Bumblebee, Machine learning, Training

Format: Virtual