Generative Query Network (GQN) in PyTorch as described in "Neural Scene Representation and Rendering"
Branch: master
Clone or download

Generative Query Network

This is a PyTorch implementation of the Generative Query Network (GQN) described in the DeepMind paper "Neural scene representation and rendering" by Eslami et al. For an introduction to the model and problem described in the paper look at the article by DeepMind.

The current implementation generalises to any of the datasets described in the paper. However, currently, only the Shepard-Metzler dataset has been implemented. To use this dataset you can use the provided script in

sh scripts/ data-dir batch-size

The model can be trained in full by in accordance to the paper by running the file or by using the provided training script

sh scripts/ data-dir


The implementation shown in this repository consists of all of the representation architectures described in the paper along with the generative model that is similar to the one described in "Towards conceptual compression" by Gregor et al.

Additionally, this repository also contains implementations of the DRAW model and the ConvolutionalDRAW model both described by Gregor et al.