Character-level Language Modeling on Penn Treebank char-level (test)
1.16BPCdense-IndRNN
Evaluation results
| Method | Links | ||
|---|---|---|---|
| dense-IndRNNSteps=1002019.10 | 1.16 | — | |
| res-IndRNNLayers=21, Steps=1002019.10 | 1.17 | — | |
| Fast-slow LSTM2019.10 | 1.19 | — | |
| dense-IndRNNSteps=502019.10 | 1.19 | — | |
| Neural Architecture Search2019.10 | 1.21 | — | |
| IndRNNLayers=6, Steps=1002019.10 | 1.21 | — | |
| res-IndRNNLayers=21, Steps=502019.10 | 1.21 | — | |
| Hierarchical Multiscale LSTM + LN2019.10 | 1.24 | — | |
| R-Transformer2019.10 | 1.24 | — | |
| HyperLSTM + LN2019.10 | 1.25 | — | |
| IndRNNLayers=6, Steps=502019.10 | 1.26 | — | |
| LSTM+Zoneout2019.10 | 1.27 | — | |
| LSTM+Recurrent dropout2019.10 | 1.32 | — | |
| LSTM+Recurrent batchnorm2019.10 | 1.32 | — | |
| Sparse PC2022.11 | 1.35 | — | |
| LSTM2019.10 | 1.36 | — | |
| Bipartite flow2022.11 | 1.38 | — | |
| HF-MRNN2019.10 | 1.42 | — | |
| Transformerreported in=[26]2019.10 | 1.45 | — | |
| AF/SCF2022.11 | 1.46 | — | |
| RNN-path2019.10 | 1.47 | — | |
| RNN-TRec2019.10 | 1.48 | — | |
| RNN-tanh2019.10 | 1.55 | — | |
| RNN-ReLU2019.10 | 1.55 | — | |
| IAF/SCF2022.11 | 1.63 | — |