Best-First Beam Search

TL;DR

This paper introduces a faster, memory-efficient variant of beam search for NLP decoding that leverages monotonic scoring assumptions to prune hypotheses early, improving speed without sacrificing performance.

Contribution

It proposes a novel monotonic approximation approach to accelerate beam search and a memory-reduced variant that maintains search bias benefits.

Findings

Up to 10x faster decoding in practice

Effective pruning based on monotonic scoring functions

Memory reduction with maintained search bias

Abstract

Decoding for many NLP tasks requires an effective heuristic algorithm for approximating exact search since the problem of searching the full output space is often intractable, or impractical in many settings. The default algorithm for this job is beam search -- a pruned version of breadth-first search. Quite surprisingly, beam search often returns better results than exact inference due to beneficial search bias for NLP tasks. In this work, we show that the standard implementation of beam search can be made up to 10x faster in practice. Our method assumes that the scoring function is monotonic in the sequence length, which allows us to safely prune hypotheses that cannot be in the final set of hypotheses early on. We devise effective monotonic approximations to popular nonmonontic scoring functions, including length normalization and mutual information decoding. Lastly, we propose a memory-reduced variant of Best-First Beam Search, which has a similar beneficial search bias in terms of downstream performance, but runs in a fraction of the time.