Leo

i see. objects are indeed not deallocated. i'll need to take a closer look to see why arc isn't cleaning things up.

Hi, I have not been able to reproduce or track down this error. Sometimes these errors are related to a recent change in Caffe2. Would it be possible to build...

Hi, it's not that easy to inspect the pb file directly, but simply writing and reading it back in should give you the exact same weights. `workspace.GetBlob("nameofparameter")).data()` should give you...

I agree, it is a little confusing. FWIW, the idea is plotting y and radius against x, therefore grouping y and radius in the second argument. I am open to...

Hi @rafaelmizrahi, thank you for opening an issue. I went ahead and made paleness extern. I also changed the default to no paleness. Does this resolve the issue?

Hi, this repo contains examples using `googlenet` for training and inference. For example take a look at `imagenet.cc`. There are currently no examples that use Detectron.

That's a good question. This trainable/non_trainable thing is a bit of a hack. It's mainly used to decide for which operators we need a to add a gradient in `AddGradientOps`....