Scaling-law extrapolation on MNIST high-D holdout
0.122RMSE (log space)Muennighoff
Evaluation Results
| Method | Links | |
|---|---|---|
| Muennighoff2026.05 | 0.122 | |
| Chinchilla2026.05 | 0.123 | |
| Ours (Eq. (2))2026.05 | 0.137 | |
| BNSL k=22026.05 | 0.169 | |
| M4 (D-axis)2026.05 | 0.192 | |
| Farseer2026.05 | 1.49 |