Beam College 2023 | Part 1: Overview of Beam ML in Python and intro to the problem
Learn about how to incorporate Machine Learning inference into your Beam pipelines. This session gives an overview of the Intermediate Track, and introduces Beam Machine Learning. Our example pipeline for this exercise takes voice input, turns it into text, classifies the text based on use case, sends each class to a different model, and turns the text output back into a voice sample.
We cover the RunInference transform and the ModelHandler class. RunInference makes it easy to call a model for inference, and the ModelHandler helps adapt models from different ML frameworks like PyTorch or TensorFlow, and also allows easy model loading from HuggingFace.