Stochastic Answer Networks for Natural Language Inference
Xiaodong Liu†{\dagger} Affiliation: †{\dagger} Microsoft Research, Redmond, WA, USA Affiliation: ‡{\ddagger} Johns Hopkins University, Baltimore, MD, USA Affiliation: Kevin Duh‡{\ddagger} Affiliation: †{\dagger} Microsoft Research, Redmond, WA, USA Affiliation: ‡{\ddagger} Johns Hopkins University, Baltimore, MD, USA Affiliation: Jianfeng Gao†{\dagger} Affiliation: †{\dagger} Microsoft Research, Redmond, WA, USA Affiliation: ‡{\ddagger} Johns Hopkins University, Baltimore, MD, USA Affiliation:
Abstract
We utilize a stochastic answer network (SAN) to explore multi-step inference strategies in Natural Language Inference. Rather than directly predicting the results given the inputs, the model maintains a state and iteratively refines its predictions. This can potentially model more complex inferences than the existing single-step inference methods. Our experiments show that SAN achieves state-of-the-art results on four benchmarks: Stanford Natural Language Inference (SNLI), MultiGenre Natural Language Inference (MultiNLI), SciTail, and Quora Question Pairs datasets.
原文 arXiv:1804.07888;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1804.07888v2