Machine Comprehension on CNN (test)
77.9AccuracyGA Reader
Evaluation results
| Method | Links | |
|---|---|---|
| GA Readervariant=GA (update L(w))2016.06 | 77.9 | |
| GA Readervariant=GA (fix L(w))2016.06 | 77.8 | |
| Neural netRelabeling=true, Ensemble=true2016.06 | 77.6 | |
| Stanford AR (relabeling)ensemble=true2016.06 | 77.6 | |
| GA Readervariant=GA (+feature, fix L(w))2016.06 | 77.4 | |
| BIDAF2016.06 | 76.9 | |
| GA Readervariant=GA (+feature, update L(w))2016.06 | 76.9 | |
| Neural netEnsemble=true2016.06 | 76.5 | |
| Iterative Alternating Neural AttentionEnsemble=true2016.06 | 76.1 | |
| Iterative Attentive Readerensemble=true2016.06 | 76.1 | |
| AS ReaderAggregation=avg ensemble2016.03 | 75.4 | |
| AS ReaderEnsemble=true2016.06 | 75.4 | |
| AS Readerensemble=true2016.06 | 75.4 | |
| AS ReaderAggregation=greedy ensemble2016.03 | 74.8 | |
| ReasoNet2016.06 | 74.7 | |
| ReasoNetTmax=52016.09 | 74.7 | |
| AoA Reader2016.06 | 74.4 | |
| AoA Reader2016.09 | 74.4 | |
| EpiReader2016.06 | 74 | |
| EpiReader2016.06 | 74 | |
| EpiReader2016.09 | 74 | |
| GA Readervariant=GA--2016.06 | 73.8 | |
| GA Reader2016.09 | 73.8 | |
| Neural netRelabeling=true2016.06 | 73.6 | |
| Stanford AR (relabeling)2016.06 | 73.6 | |
| Iterative Alternating Neural AttentionEnsemble=false2016.06 | 73.3 | |
| Iterative Attentive Reader2016.06 | 73.3 | |
| Iterative Attention Reader2016.09 | 73.3 | |
| Dynamic Entity Repres.Strategy=+ w2v2016.03 | 72.9 | |
| DER Network2016.06 | 72.9 | |
| DER Network2016.09 | 72.9 | |
| Neural net2016.06 | 72.7 | |
| Stanford AR2016.06 | 72.4 | |
| Chen et al. (2016)Model configuration=single model2016.03 | 72.4 | |
| Stanford AREmbeddings=GloVe2016.06 | 72.4 | |
| Stanford AR2016.09 | 72.4 | |
| EpiReader Extractor2016.06 | 72 | |
| Dynamic Entity Repres.Strategy=max-pool + byway2016.03 | 72 | |
| Dynamic Entity Repres.Strategy=max-pool2016.03 | 70.7 | |
| AS ReaderAggregation=avg for top 20%2016.03 | 69.9 | |
| AS Reader2016.06 | 69.9 | |
| AS Reader2016.06 | 69.5 | |
| AS ReaderModel configuration=single model2016.03 | 69.5 | |
| AS Reader2016.06 | 69.5 | |
| AS Reader2016.09 | 69.5 | |
| MemNNsModel configuration=ensemble2016.03 | 69.4 | |
| MemNNs (ensemble)Ensemble=true2016.06 | 69.4 | |
| MemNNsEnsemble=true2016.06 | 69.4 | |
| MemNetsensemble=true2016.06 | 69.4 | |
| Entity-centric Classifier2016.06 | 67.9 | |
| MemNNs2016.06 | 66.8 | |
| MemNNsModel configuration=single model2016.03 | 66.8 | |
| MemNNs (window memory + self-sup.)2016.06 | 66.8 | |
| MemNNs (window memory + self sup.)2016.06 | 66.8 | |
| MemNets2016.06 | 66.8 | |
| MemNets2016.09 | 66.8 | |
| Impatient Reader2016.06 | 63.8 | |
| Impatient ReaderModel type=single model2016.03 | 63.8 | |
| Impatient Reader2016.06 | 63.8 | |
| Impatient Reader2016.06 | 63.8 | |
| Impatient Reader2016.06 | 63.8 | |
| Attentive Reader2016.06 | 63 | |
| Attentive ReaderModel type=single model2016.03 | 63 | |
| Attentive Reader2016.06 | 63 | |
| Attentive Reader2016.06 | 63 | |
| Attentive Reader2016.06 | 63 | |
| Attentive Reader2016.09 | 63 | |
| MemNNs (window memory)2016.06 | 60.6 | |
| MemNNs (window memory)2016.06 | 60.6 | |
| Deep LSTM Reader2016.06 | 57 | |
| Deep LSTM Reader2016.06 | 57 | |
| Deep LSTM Reader2016.06 | 57 | |
| Deep LSTM Reader2016.06 | 57 | |
| Deep LSTM Reader2016.09 | 57 | |
| Word distance model2016.06 | 50.9 | |
| Word distance model2016.06 | 50.9 | |
| Frame-semantic model2016.06 | 40.2 |