Fast Lexically Constrained Decoding with Dynamic Beam Allocation for Neural Machine Translation
This work addresses a specific bottleneck in NMT for researchers and practitioners by providing a faster method for constrained decoding, though it is incremental as it builds on prior approaches.
The paper tackles the computational inefficiency of existing lexically constrained decoding methods in neural machine translation by introducing an algorithm with O(1) complexity in the number of constraints, demonstrating its ability to properly place constraints and explore the relationship between model and BLEU scores.
The end-to-end nature of neural machine translation (NMT) removes many ways of manually guiding the translation process that were available in older paradigms. Recent work, however, has introduced a new capability: lexically constrained or guided decoding, a modification to beam search that forces the inclusion of pre-specified words and phrases in the output. However, while theoretically sound, existing approaches have computational complexities that are either linear (Hokamp and Liu, 2017) or exponential (Anderson et al., 2017) in the number of constraints. We present a algorithm for lexically constrained decoding with a complexity of O(1) in the number of constraints. We demonstrate the algorithms remarkable ability to properly place these constraints, and use it to explore the shaky relationship between model and BLEU scores. Our implementation is available as part of Sockeye.