The recent explosion in the power and popularity of machine learning techniques has been fueled in part by the ecosystem of open source Python libraries. One of those was PyTorch, a successor to Torch7, which is rapidly becoming one of the most essential tools in every ML researcher’s toolbox.
However, research is not the end of the story. Machine learning is transforming entire fields, meaning that efficient inference is now becoming more important than ever. Even though PyTorch has strong roots in research applications, inference functionality is one of our recent focuses and has driven us to develop some innovative solutions in that space.
In this talk I will present the ideas underlying the library and how it can be utilized in the variety of scenarios in which machine learning appears, which range from research all the way to production.
Adam Paszke is an author and maintainer of PyTorch. He has already worked with large organizations like Facebook AI Research, Google and NVIDIA, despite the fact that he has only recently graduated from the Master’s program in Computer Science at the University of Warsaw. Currently, he is also finishing his second major in Mathematics. His general interests include graph theory, programming languages, numerical computing, and machine learning.
