DeepMind Flamingo Explained — 32 Images are Enough To Learn!

Author(s): Boris Meinardus

Originally published on Towards AI.

This AI can solve tasks it has never seen before, looking at only a few examples!
Source: Author

We will be looking at DeepMind’s famous Flamingo paper, what impressive results it achieves, where it falls short, how the model works, and how it was trained.

We, as humans, have a good sense of pattern matching. As the authors themselves say:

One key aspect of intelligence is the ability to quickly learn to perform a new task given a short instruction.

When looking at a few examples, we understand what the task wants us to do. But so does the Flamingo model. It can take in a sequence of interleaved images and text and complete a specific task it has never… Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI