List your company
ResearchFormerly WizWand

6-way Question Classification on TREC 6-class (test)

97.6AccuracyAdam
41.4456.0270.685.18Apr 5, 2015Feb 6, 2017Dec 12, 2018Oct 16, 2020Aug 21, 2022Jun 25, 2024Apr 30, 2026
Updated 6mo ago

Evaluation results

MethodLinks
AdamGeneration=Gen 32026.0497.6
AdamWGeneration=Gen 32026.0497.6
Lookahead+AdamWGeneration=Gen 52026.0497.6
RAdamGeneration=Gen 52026.0496.8
Fixed SGDGeneration=Gen 12026.0496.6
SAMGeneration=Gen 52026.0496.4
SGDRGeneration=Gen 22026.0496.2
LSTM-CNN2017.0896.1
d-TBCNN2015.0496
TBCNN2017.0896
BCN+Char+CoVeembeddings=GloVe, character n-gram, CoVe2017.0895.8
DSCNN-P2017.0895.6
SVMsource=Van-Tu and Anh-Cuong, 20162017.0895.2
LARSGeneration=Gen 42026.0495.2
SVMfeatures=10k features + 60 rules2015.0495
SVMsource=da Silva et al., 20112017.0895
SVMFeatures=unigrams, bigrams, wh-word, head word, POS tags, parser, hypernyms, WordNet synsets, 60 hand-coded rules2015.1195
LionGeneration=Gen 52026.0495
DALS-FastGeneration=SOTA2026.0495
c-TBCNN2015.0494.8
Cosine Decay SGDGeneration=Gen 22026.0494.8
C-LSTMImplementation=our implementation2015.1194.6
DALS-AccGeneration=SOTA2026.0494.2
DALSGeneration=SOTA2026.0494
CNN-non-static2015.0493.6
CNN-non-static2015.1193.6
LSTMImplementation=our implementation2015.1193.2
Deep-CNN2015.0493
DCNN2015.1193
Bi-LSTMImplementation=our implementation2015.1193
AdaBoundGeneration=Gen 32026.0492.8
Ada-CNN2015.0492.4
Ada-CNNModel Type=self-adaptive hierarchical sentence model with gating networks2015.1192.4
CNN-multichannel2015.0492.2
CNN-multichannel2015.1192.2
Paragraph VectorDescription=infers the new paragraph vector for unseen documents2015.1191.8
RNN2015.0490.2
Discriminative LRGeneration=Gen 42026.0484.6
SAM+DiscriminativeGeneration=SOTA2026.0484.6
GrokfastGeneration=Gen 52026.0484.4
STLR+DiscriminativeGeneration=Gen 52026.0443.6