Predicting Transformer Generalization on MNIST-Transformers 20% threshold
0.905Kendall's tauTransformer-NFN Quasi
Evaluation Results
| Method | Links | |
|---|---|---|
| Transformer-NFN QuasiExtra Parameters (%)=4.54% params ++, Number of runs=52026.04 | 0.905 | |
| Transformer-NFN largeExtra Parameters (%)=57.66% params ++, Number of runs=52026.04 | 0.904 | |
| Transformer-NFNNumber of runs=52026.04 | 0.899 | |
| MLPNumber of runs=52026.04 | 0.873 | |
| STATNNNumber of runs=52026.04 | 0.872 | |
| XGBoostNumber of runs=52026.04 | 0.839 | |
| LightGBMNumber of runs=52026.04 | 0.835 | |
| Random ForestNumber of runs=52026.04 | 0.758 |