Mitigating these biases relies on exposing and removingnuisance components.

We provide additional qualitative comparative results on both ShapeNetandPix3DinFig8,9.

In the "modern era", such model-parallel training techniques trace their roots back to AlexNet [14] and early influential systems such as DistBelief [6] and Project Adam [3].

In doing so, we are able to find the exact settings for which different methodsprovablybecome identical.