Soumith Chintala, AI research engineer, Facebook, NYC 2017

on YouTube ↗22m

" summary

Covers the shift from static dataset workflows (e.g., ImageNet) toward dynamic, online settings like self-driving cars, robotics, and game environments such as Universe and StarCraft, focusing on reinforcement and online learning problems. Describes dynamic/declarative deep-learning frameworks — specifically PyTorch, DyNet and Chainer — as tools developed to support this paradigm shift.

AI-written summary · about

" chapters (8)

1. 0:00Start
2. 1:01Overview
3. 1:56Today's AI
4. 13:41A next-gen framework for Al
5. 14:44Tools for Al research and deployment
6. 16:59Dynamic graph Frameworks
7. 18:58Py Torch Autograd
8. 20:24PYTORCH

" related talks

• 28mChallenges & future trends of Deep LearningMohammad Shokoohi-Yekta · Global Artificial Intelligence Conference 2020
• 07mQ n A with Drs Tarry SinghSheamus McGovern · ODSC India 2018
• 41mDeep Learning to Solve Challenging ProblemsJeff Dean · Google I/O 2019
• 29mDeep Learning, Neuroscience and the future of AIDarjan Salaj · Berlin Buzzwords 2021
• 39mA Fireside Chat with Turing Award Winner Geoffrey Hinton, Pioneer of Deep LearningGeoffrey Hinton · Google I/O 2019
• 03mRethinking software engineering in the AI eraYangqing Jia · Artificial Intelligence Conference in London 2018
• 12mKeynote: What's New in AIEve Phillips · Cassandra Summit 2023
• 18mOn the History and Future of AIFei Fei Li · Data + AI Summit 2024