A free, code-first course by Jeremy Howard that has you training state-of-the-art models from the first lesson.
fast.ai teaches top-down: you build working models for vision, text and tabular data straight away, then open them up to understand how they work. It uses PyTorch and the fastai library, and is known for getting people with coding experience productive in deep learning quickly.
glitchdata links to this course; its content, schedule and pricing are set by fast.ai.
This course is run by fast.ai. glitchdata describes it and links to it, but doesn't host its lessons — you take it on their site, under their terms.
course.fast.ai ↗2 min read
2 min read
2 min read
2 min read
Build, compare and evaluate a real classifier end to end, predicting penguin species from their measurements.
How neural networks turn inputs into predictions, how they learn with gradient descent, and how to train one in PyTorch.
Google's free, fast-paced introduction to machine learning, with videos, visualisations and exercises.