Passing the PyTorch: Using Machine Learning Resources

Deep learning is rising in popularity, and rightfully so as it’s astounding technology. It tries to mimic the workings of the human brain and uses artificial neural networks to accomplish this. Deep learning has actually been around for quite some time, dating back to the 1960s, but it wasn’t until the advent of the GPU that deep learning became widely adopted. Deep learning is now used to perform multiple computations simultaneously with supreme accuracy on mass amounts of data.

What often takes a backseat, however, in the rise of deep learning are the frameworks and libraries behind it. PyTorch, a machine learning library developed by Facebook Artificial Intelligence Research (FAIR), has picked up a lot of steam. It’s being used in some of the most innovative projects in the world such as at Disney, which uses it to identify faces in cartoons and animated films. Airbnb uses conversational AI tools in PyTorch to enhance customer experiences and Tesla built its self-driving software on PyTorch. And the list goes on.

On the latest episode of AI & U, Mark McQuade is joined by Randall Hunt, Developer Advocate at Facebook AI, to discuss deep learning and utilizing ML learning libraries such as PyTorch.

Tune in to hear about:

• Being a developer advocate
• Taking AI models into production with frameworks and libraries
• PyTorch as an open-source AI resource
• Comparing Pytorch and Tensorflow
• PyTorch use cases in different sectors including automotive, healthcare and farming

Randall discusses the PyTorch open source community and variety of contributors. “One of the things that I enjoy is I get to focus on the open source tools. And these tools are things that we use internally and that we think are going to be helpful for the community at large. So we end up open sourcing them. And one of the great things about PyTorch is Facebook is just one of many, many contributors. So the there's a huge amount of work that comes from large partners like Nvidia or AWS or Google and all kinds of contributors from large companies. But there are academics as well. So people from Cornell or elsewhere who are coming and adding in new features. Large components of PyTorch get built out and proposed by the community with relatively little involvement from our side. And I think the breadth and depth of contributions and PyTorch is something that I haven't seen in a really long time.”

Randall explains the importance of open source tooling around responsible AI. “There's some very important work in the space of responsible AI. So privacy, respecting AI that is making the right sorts of decisions in terms of under-representing people or kind of classes of the community and making sure you're not doing the wrong thing with AI. And that's a non-trivial problem to solve. And one of the most important parts there is the tooling. So I think some of the open source tooling around responsible AI is very important. And that work should continue.”

Randall also explains why he sees AI as so valuable. “It's an opportunity to makes the world that we see in science fiction, real. So closing the gap between what we see in science fiction, you know, the positive science fiction, and reality is something that's very interesting to me. Progress doesn't happen without people really putting in effort and research. So that's kind of why I see AI as being valuable.”

Join the Conversation: Find Solve on Twitter and LinkedIn, or follow along via RSS.

Stay on top of what's next in technology

Learn about tech trends, innovations and how technologists are working today.