Logan O'Sullivan Bruns

GQN trained on CLEVR dataset

10

Thanks for the GQN implementation. I thought you might enjoy seeing some pictures of how it does trained on a different dataset. (Albeit with a limited amount of training time....

Self-Attention and other extensions

2

@ogroth, as mentioned in #27 I've found it useful to add self-attention and also some additional training options to limit the amount of data used for training. Also a simple...

Feed-forward convolutional architecture with self-attention. Self-Attention as described in Self-Attention Generative Adversarial Networks by Zhang, Goodfellow, Metaxas, and Odena added to the tower+pool architecture. As discussed in issues #28 and...

I have done some limited experiments with training [Generative Query Networks](https://deepmind.com/blog/neural-scene-representation-and-rendering/) with CLEVR dataset so I can experiment with using them instead of RESNET based embeddings for CLEVR VQA. I...

Note, the defaults are same as previous behavior. This commit just allows for customization of these settings via helm.