Language Modeling on Copy
1PPLFull Attention
Evaluation results
| Method | Links | |
|---|---|---|
| Full AttentionNumber of parameters=1/8 of full2021.10 | 1 | |
| ScatterbrainNumber of parameters=1/8 of full2021.10 | 2.58 | |
| ReformerNumber of parameters=1/8 of full2021.10 | 6.8 | |
| PerformerNumber of parameters=1/8 of full2021.10 | 49 |