Fast Structured Decoding for Sequence Models

Method and contribution

Adds approximate first-order CRF inference and dynamic transitions to non-autoregressive sequence models.

A first-order structured-decoding reference; Cascaded Text Generation develops higher-order cascades and shared Markov Transformer potentials.

Read in context

Structured Decoding: Higher-Order CRFs, Diffusion Language Models, and Candidate Lattices →

Citation

Zhiqing Sun, Zhuohan Li, Haoqing Wang, Di He, Zi Lin, Zhi-Hong Deng. Fast Structured Decoding for Sequence Models. NeurIPS 2019.

@inproceedings{sun2019fast,
  title = {{Fast Structured Decoding for Sequence Models}},
  author = {Sun, Zhiqing and Li, Zhuohan and Wang, Haoqing and He, Di and Lin, Zi and Deng, Zhi-Hong},
  year = {2019},
  booktitle = {NeurIPS 2019},
  url = {https://proceedings.neurips.cc/paper_files/paper/2019/hash/74563ba21a90da13dacf2a73e3ddefa7-Abstract.html},
  eprint = {1910.11555},
  archivePrefix = {arXiv}
}