Hierarchical reasoning on symbolic sequences on Long ListOps (test)
62.75AccuracyLiquid-S4
Evaluation Results
| Method | Links | |
|---|---|---|
| Liquid-S4Model Type=State Space Models2023.05 | 62.75 | |
| MULTIRESNETModel Type=Convolution2023.05 | 62.75 | |
| S5Model Type=State Space Models2023.05 | 62.15 | |
| SGConv*Model Type=Convolution, Note=Built to mimic global convolution interpretation of S42023.05 | 61.45 | |
| S4DModel Type=State Space Models2023.05 | 60.52 | |
| S4Model Type=State Space Models2023.05 | 59.6 | |
| DSSModel Type=State Space Models2023.05 | 57.6 | |
| H-Transformer-1DModel Type=Attention2023.05 | 49.53 | |
| CDILModel Type=Convolution2023.05 | 44.05 | |
| ReformerModel Type=Attention2023.05 | 37.27 | |
| Luna-256Model Type=Attention2023.05 | 37.25 | |
| NyströmformerModel Type=Attention2023.05 | 37.15 | |
| TransformerModel Type=Attention2023.05 | 36.37 | |
| BigBirdModel Type=Attention2023.05 | 36.05 | |
| LongformerModel Type=Attention2023.05 | 35.63 | |
| FNetModel Type=Attention2023.05 | 35.33 | |
| Sinkhorn TransformerModel Type=Attention2023.05 | 33.67 | |
| PerformerModel Type=Attention2023.05 | 18.01 | |
| Sparse TransformerModel Type=Attention2023.05 | 17.07 | |
| Linear Trans.Model Type=Attention2023.05 | 16.13 | |
| LinformerModel Type=Attention2023.05 | 16.13 | |
| Local AttentionModel Type=Attention2023.05 | 15.82 |