If beam search is the answer, what was the question?
This addresses a fundamental problem in neural language generation for researchers and practitioners, offering insights into decoding strategies to improve text quality, though it is incremental in refining existing methods.
The paper investigates why exact MAP decoding often yields low-quality text despite being theoretically optimal, while beam search performs better despite high search error rates. It finds that beam search implicitly enforces uniform information density, a cognitive science principle, and proposes new decoding objectives that improve translation quality, with strong correlation to BLEU scores.
Quite surprisingly, exact maximum a posteriori (MAP) decoding of neural language generators frequently leads to low-quality results. Rather, most state-of-the-art results on language generation tasks are attained using beam search despite its overwhelmingly high search error rate. This implies that the MAP objective alone does not express the properties we desire in text, which merits the question: if beam search is the answer, what was the question? We frame beam search as the exact solution to a different decoding objective in order to gain insights into why high probability under a model alone may not indicate adequacy. We find that beam search enforces uniform information density in text, a property motivated by cognitive science. We suggest a set of decoding objectives that explicitly enforce this property and find that exact decoding with these objectives alleviates the problems encountered when decoding poorly calibrated language generation models. Additionally, we analyze the text produced using various decoding strategies and see that, in our neural machine translation experiments, the extent to which this property is adhered to strongly correlates with BLEU.