n-embd_256
done · peak unseen_mismatch 93.1
Continued from
runs/20260704_225157_msize/n-embd_256
Figures
Results
at final waypoint (t = 4800), mean ± sd over seeds
seen_* generated at each grid point's own training α · unseen_* = held-out pairings, uniform (α-independent) · 1000 pairs per condition
| α | seen_match | seen_mismatch | unseen_match | unseen_mismatch |
|---|---|---|---|---|
| 0.9 | 100.0 ± 0.1 | 49.8 ± 16.9 | 92.0 ± 4.6 | 50.4 ± 16.1 |
| 1.0 | 100.0 ± 0.0 | 57.9 ± 18.4 | 92.2 ± 3.8 | 49.1 ± 14.1 |
| 1.1 | 100.0 ± 0.1 | 75.8 ± 9.8 | 87.4 ± 6.9 | 63.8 ± 14.1 |
| 1.2 | 99.9 ± 0.1 | 87.1 ± 5.8 | 87.6 ± 7.6 | 69.1 ± 5.0 |
| 1.3 | 99.9 ± 0.1 | 87.8 ± 10.2 | 87.9 ± 5.8 | 66.8 ± 14.3 |
| 1.4 | 99.8 ± 0.2 | 91.9 ± 3.7 | 84.3 ± 5.5 | 64.8 ± 9.8 |
| 1.5 | 99.8 ± 0.2 | 97.3 ± 1.7 | 85.8 ± 5.6 | 76.8 ± 8.9 |
| 1.6 | 99.9 ± 0.2 | 94.3 ± 6.3 | 86.7 ± 4.5 | 69.4 ± 10.4 |
| 1.7 | 99.8 ± 0.2 | 97.1 ± 1.5 | 85.7 ± 4.5 | 72.0 ± 6.1 |
| 1.8 | 99.7 ± 0.1 | 98.2 ± 0.9 | 85.3 ± 4.4 | 75.0 ± 6.8 |
| 1.9 | 99.8 ± 0.3 | 98.1 ± 0.9 | 84.5 ± 3.6 | 73.9 ± 6.2 |
| 2.0 | 99.7 ± 0.2 | 98.4 ± 0.7 | 83.9 ± 3.1 | 74.5 ± 6.5 |
| 2.1 | 99.8 ± 0.1 | 99.1 ± 0.5 | 83.5 ± 2.5 | 77.7 ± 4.5 |
| 2.5 | 99.8 ± 0.1 | 99.4 ± 0.4 | 80.5 ± 2.9 | 71.2 ± 7.3 |
| 3.0 | 99.9 ± 0.1 | 99.8 ± 0.2 | 80.2 ± 2.7 | 73.6 ± 3.6 |