PyTorch
In this talk, Ph.D. student Andrew Trask reflects on how AI tools like PyTorch have driven the massive wave of AI progress over the last decade, and describes a vision for the next generation of AI tools. Specifically, he makes the case that privacy-enhancing tools will unlock orders of magnitude more data for training and testing AI models, and that this will unlock AI’s true potential for solving society’s most important problems.
Similar Posts
3 thoughts on “The Future of AI Tools | PyTorch Developer Day 2020”
Comments are closed.
Hey Andrew, awesome stuff! I was wondering how one can determine the quality of the data before training if they can't see it? Is the data quality possible to judge and vote upon as a community? How does one protect your model from adversarial datasets out there?
If you're interested in GPT-3 this article is DEFINITELY worth a look
https://medium.com/god-is-a-machine/god-is-a-machine-p1-e9be9c9933d7
awesome stuff! As a researcher, I really appreciate what PyTorch has been doing over the years. Thank you.