List your company
ResearchFormerly WizWand

Question Formation on English Question Formation (test)

100Full AccuracyTf
-3.79223.15450.177.046Nov 5, 2025
Updated 4mo ago

Evaluation results

MethodLinks
TfSource=Ahuja et al. (2025), Training objective=language modeling objective2025.11100
TfSource=Qin et al. (2025), Uses ground-truth parse trees=true2025.11100
TfArchitecture=Transformer2025.1199.9
Tf+SupArchitecture=Transformer + Supervision2025.1199.9
Tf+NdArchitecture=Stack Attention Transformer2025.1199.9
Tf+Sup+SupArchitecture=Transformer + Supervision + Supervision2025.1199.8
Tf+Nd+NdArchitecture=Multi-Stack Attention Transformer2025.1199.5
Tree-GRUArchitecture=Tree-structured GRU, Source=McCoy et al. (2020)2025.1196
ON-LSTMArchitecture=Ordered Neurons LSTM, Source=McCoy et al. (2020)2025.1193
GRU+loc. attn.Architecture=GRU + local attention, Source=McCoy et al. (2020)2025.1177
RNNArchitecture=RNN2025.1146.5
LSTM+Nd+RArchitecture=Stack LSTM + Recurrent2025.1111.8
RNN+NdArchitecture=Stack RNN2025.112.7
RNN+SupArchitecture=RNN + Supervision2025.112.2
RNN+Nd+RArchitecture=Stack RNN + Recurrent2025.111.9
RNN+Sup+RArchitecture=RNN + Supervision + Recurrent2025.111.1
LSTM+Sup+RArchitecture=LSTM + Supervision + Recurrent2025.110.5
LSTMArchitecture=LSTM2025.110.3
LSTM+SupArchitecture=LSTM + Supervision2025.110.2
LSTM+NdArchitecture=Stack LSTM2025.110.2