Character-level Classification on DBpedia character-level (test)
1.29Test Error Rate29-layer CNN
Evaluation results
| Method | Links | |
|---|---|---|
| 29-layer CNN2018.03 | 1.29 | |
| r-LSTMn (sampling frequency)=20, l (subsequence length)=15, Number of layers=2, Units=5122018.03 | 1.4 | |
| CNN+RNN2018.03 | 1.43 | |
| LM-LSTM2018.03 | 1.5 | |
| char-CNN2018.03 | 1.66 | |
| SA-LSTM2018.03 | 2.34 | |
| LSTM Full Backprop2018.03 | 13.64 |