PyTorch implementation for ICLR 2018 accepted paper Twin Networks: Matching the Future for Sequence Generation [pdf] for verifying the results for image captioning on MSCOCO dataset and pixel-by-pixel generation on MNIST.

An analysis of the paper and the reproducibility results on both the datasets are provided in Reproducibility_Report.pdf.

The implementation details are provided in ImageCaptioningCOCO and SequentialMNIST.

If you find this code useful, please consider citing the original work by authors:

@inproceedings{Serdyuk2017TwinNM,
  title={Twin Networks: Matching the Future for Sequence Generation},
  author={Dmitriy Serdyuk and Nan Rosemary Ke and Alessandro Sordoni and Adam Trischler and Chris Pal and Yoshua Bengio},
  year={2017}
}