
MIT’s Jonathan Frankle on “The Lottery Hypothesis” - Ep. 115
NVIDIA AI Podcast
00:00
How to Train a Neural Network Smaller
It's an expensive proposition to run a research group doing neural networks. So wouldn't it be exciting if, you know, we could just train smaller networks? Sure. A lot of people see the pruning results and they say, well, doesn't that mean we can just training smaller networks from the beginning? And the answer is no.
Play episode from 03:18
Transcript


