Stochastic Beams and Where to Find Them: The Gumbel-Top-k Trick for Sampling Sequences Without Replacement

03/14/2019
by   Wouter Kool, et al.
10

The well-known Gumbel-Max trick for sampling from a categorical distribution can be extended to sample k elements without replacement. We show how to implicitly apply this 'Gumbel-Top-k' trick on a factorized distribution over sequences, allowing to draw exact samples without replacement using a Stochastic Beam Search. Even for exponentially large domains, the number of model evaluations grows only linear in k and the maximum sampled sequence length. The algorithm creates a theoretical connection between sampling and (deterministic) beam search and can be used as a principled intermediate alternative. In a translation task, the proposed method compares favourably against alternatives to obtain diverse yet good quality translations. We show that sequences sampled without replacement can be used to construct low-variance estimators for expected sentence-level BLEU score and model entropy.

READ FULL TEXT

Please sign up or login with your details

Forgot password? Click here to reset