Associative Recall on Synthetic sequence modeling Associative recall
98.48AccuracyTransformer
Evaluation Results
| Method | Links | |
|---|---|---|
| TransformerAttention=Softmax2023.12 | 98.48 | |
| MIMOFormerN=2, variant=att., Attention=FAVOR+2023.12 | 96.52 | |
| PerformerAttention=FAVOR+2023.12 | 96.32 | |
| MIMOFormerN=2, variant=att., Attention=DPFP2023.12 | 93.64 | |
| DataMUXN=2, Attention=Softmax2023.12 | 20.04 |