Long-range Sequence Modeling on Permuted MNIST (test)
98.13AccuracyTrellisNet
Evaluation Results
| Method | Links | |
|---|---|---|
| TrellisNetnumber of parameters=8M2018.10 | 98.13 | |
| Transformer (self-attention)2018.10 | 97.9 | |
| Generic TCN2018.10 | 97.2 | |
| IndRNN2018.10 | 96 | |
| r-LSTM w/ Aux. Loss2018.10 | 95.2 | |
| Dilated GRU2018.10 | 94.6 |