List your company
ResearchFormerly WizWand

Machine Comprehension on CBT-CN (test)

83.7AccuracyAS Reader (+BookTest, ensemble)
54.89262.37169.8577.329Mar 4, 2016Apr 14, 2016May 25, 2016Jul 5, 2016Aug 15, 2016Sep 25, 2016Nov 6, 2016
Updated 8mo ago

Evaluation results

MethodLinks
AS Reader (+BookTest, ensemble)trained_on_large_set=true, ensemble=true2016.0683.7
Humanscontext + query=true2016.0681.6
HumansInput configuration=context+query2016.0381.6
Humansinput=context+query2016.0681.6
Humans (context + query)2016.0681.6
AS Reader (+BookTest)trained_on_large_set=true2016.0680.8
FG fine-grained gateArchitecture=FG2016.1172.04
NSE2016.0671.9
Munkhdalai & Yu (2016)2016.1171.9
Iterative Alternating Neural Attentionensemble=true2016.0671
Iterative Attentive Readerensemble=true2016.0671
Sordoni et al. (2016) ensembleEnsemble=true2016.1171
GA fine-grained gateArchitecture=GA2016.1170.84
GA Readervariant=GA (+feature, fix L(w))2016.0670.7
EpiReaderensemble=true2016.0670.6
EpiReaderensemble=true2016.0670.6
Trischler et al. (2016) ensembleEnsemble=true2016.1170.6
GA Readervariant=GA (+feature, update L(w))2016.0669.8
GA word char concatArchitecture=GA2016.1169.6
AoA Reader2016.0669.4
Cui et al. (2016)2016.1169.4
GA word char feat concatArchitecture=GA2016.1169.28
Iterative Alternating Neural Attention2016.0669.2
Iterative Attentive Reader2016.0669.2
Sordoni et al. (2016)2016.1169.2
GA scalar gateArchitecture=GA2016.1169.08
AS Readerensemble=true2016.0668.9
AS ReaderModel variant=avg ensemble2016.0368.9
AS Readerensemble=true2016.0668.9
AS Readerensemble=true2016.0668.9
Kadlec et al. (2016) ensembleEnsemble=true2016.1168.9
GA Readervariant=GA (fix L(w))2016.0668
AS ReaderModel variant=greedy ensemble2016.0367.5
EpiReader2016.0667.4
EpiReader2016.0667.4
Trischler et al. (2016)2016.1167.4
GA Readervariant=GA (update L(w))2016.0667.3
EpiReader Extractor2016.0666.7
HumansInput configuration=query2016.0364.4
Humansinput=query2016.0664.4
Humans (query)2016.0664.4
Iterative Alternating Neural Attentionattention=fixed query attention2016.0664.3
GA Readervariant=GA--2016.0663.9
AS Reader2016.0663.4
AS ReaderModel variant=single model2016.0363.4
AS Reader2016.0663.4
AS Reader2016.0663.4
AS ReaderModel variant=avg for top 20%2016.0363.2
MemNNs2016.0663
MemNNsModel variant=window memory + self-sup.2016.0363
MemNNsconfig=window memory + self-sup.2016.0663
MemNets2016.0663
LSTMscontext + query=true2016.0656
LSTMsInput configuration=context+query2016.0356
LSTMsinput=context+query2016.0656
LSTMS (context + query)2016.0656