FusionNet: Fusing via Fully-Aware Attention with Application to Machine Comprehension

ICLR 2018 Hsin-Yuan HuangChenguang ZhuYelong ShenWeizhu Chen

This paper introduces a new neural structure called FusionNet, which extends existing attention approaches from three perspectives. First, it puts forward a novel concept of "history of word" to characterize attention information from the lowest word-level embedding up to the highest semantic-level representation... (read more)

PDF Abstract

Evaluation Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK COMPARE
Question Answering SQuAD1.1 FusionNet (ensemble) EM 78.978 # 59
Question Answering SQuAD1.1 FusionNet (ensemble) F1 86.016 # 58
Question Answering SQuAD1.1 FusionNet (single model) EM 75.968 # 87
Question Answering SQuAD1.1 FusionNet (single model) F1 83.900 # 85
Question Answering SQuAD1.1 dev FusionNet EM 75.3 # 15
Question Answering SQuAD1.1 dev FusionNet F1 83.6 # 19
Question Answering SQuAD2.0 FusionNet++ (ensemble) EM 70.300 # 114
Question Answering SQuAD2.0 FusionNet++ (ensemble) F1 72.484 # 121