State-of-the-art table for Question Answering on Children's Book Test

Rank	Method	Accuracy-CN	Accuracy-NE	Paper	Year-Month
1	WebText dataset + Transformer architecture	93.30%	89.05%	Language Models are Unsupervised Multitask Learners	2019-2
2	NSE	71.9%	73.2%	Gated-Attention Readers for Text Comprehension	2016-6
3	GA + feature + fix L(w)	70.7%	74.9%	Gated-Attention Readers for Text Comprehension	2016-6
4	AoA reader	69.4%	72%	Attention-over-Attention Neural Networks for Reading Comprehension	2016-7
5	GA reader	69.4%	71.9%	Gated-Attention Readers for Text Comprehension	2016-6
6	AS reader (avg)	68.9%	70.6%	Text Understanding with the Attention Sum Reader Network	2016-3
7	AS reader (greedy)	67.5%	71%	Text Understanding with the Attention Sum Reader Network	2016-3
8	EpiReader	67.4%	69.7%	Natural Language Comprehension with the EpiReader	2016-6
9	AIA		72%	Iterative Alternating Neural Attention for Machine Reading	2016-6

Question Answering on Children's Book Test Edit Add Remove