Language Modelling On Penn Treebank Character
Metrics
Bit per Character (BPC)
Number of params
Results
Performance results of various models on this benchmark
Comparison Table
| Model Name | Bit per Character (BPC) | Number of params |
|---|---|---|
| seq-u-net-a-one-dimensional-causal-u-net-for | 1.31 | 5.9M |
| improved-language-modeling-by-decoding-the | 1.169 | 13.8M |
| hypernetworks | 1.219 | 14.4M |
| accessing-higher-level-representations-in | 1.160 | 10.7M |
| mogrifier-lstm | 1.083 | 24M |
| recurrent-highway-networks-with-grouped | 1.147 | 16.0M |
| mogrifier-lstm | 1.120 | 24M |
| seq-u-net-a-one-dimensional-causal-u-net-for | 1.3 | 5.9M |
| trellis-networks-for-sequence-modeling | 1.158 | 13.4M |
| r-transformer-recurrent-neural-network | 1.24 | - |
| an-analysis-of-neural-language-modeling-at | 1.187 | 13.8M |
| independently-recurrent-neural-network-indrnn | 1.19 | - |
| deep-independently-recurrent-neural-network | 1.18 | - |
| an-empirical-evaluation-of-generic | 1.31 | - |
| neural-architecture-search-with-reinforcement | 1.214 | 16.3M |
| fast-slow-recurrent-neural-networks | 1.190 | 27M |
| discrete-flows-invertible-generative-models | 1.38 | - |
| gating-revisited-deep-multi-layer-rnns-that-1 | 1.30 | - |
| an-analysis-of-neural-language-modeling-at | 1.175 | 13.8M |
| fast-slow-recurrent-neural-networks | 1.193 | 27M |