R Devon Hjelm

icon indicating a website link rdevon.github.io

icon company Apple MLR icon location Redmond, Washington icon location Unsupervised learning, representation learning, self-supervised learning

Results 53 comments of R Devon Hjelm

Include `seed` to argument parser

What is the high-level goal 2-4?

Include `seed` to argument parser

OK, let's see your PR then. Thanks for working on this!

Discrete Celeba Experiment

It's been a while since I've looked at this code, but the disconnected_grad is a trick to specify a loss such that their gradient looks like the importance sampled version.