Neale Ratzlaff

Bumping this thread. I replicated the important bits from the conditional imagenet config and found one issue I couldn't get past. First you need to extend the `custom` data loader...

+1 I have the same issue in the docker container. When I run on the outside data, or on my own. The optimization component crashes with ``` optimizing... [velodyne_nodelet_manager-1] process...

That works for me. I can run the optimization with the Dockerfile changed to indigo. Here is a [gist](https://gist.github.com/neale/49ba52d9238dcc82e3ef8b5a873791ca) of my modified Dockerfile. I also had to update cmake en...

@saimanoj18 That's the same behavior that I experienced with the modified dockerfile. I think the problem with saving the map should be opened as a separate issue. As long as...

Right, I should make this clear in the readme. I built more of an autoencoder-GAN than a true adversarial autoencoder -- this has better results if you want random samples...

I referenced this in another issue -- lines 142-150 are a GAN style update. This is not strictly the adversarial autoencoder at this point (I'm working on a faithful version)....

Hi, thanks for pointing this out. I'm not sure how that error crept in there. `data/mnist.py` was returning a function `get_epoch` -- which returns a generator. The inf_gen was missing...

@gunshi * You can compute the kernel among any number of particles. But as particles are like random samples, it might make sense to try to approximate the posterior with...