smooth_K24
done · peak unseen_mismatch 90.1
α schedule
α 2.9643 ×38 α 2.8929 ×38 α 2.8214 ×38 α 2.75 ×38 α 2.6786 ×38 α 2.6071 ×38 α 2.5357 ×38 α 2.4643 ×38 α 2.3929 ×38 α 2.3214 ×38 α 2.25 ×38 α 2.1786 ×38 α 2.1071 ×37 α 2.0357 ×37 α 1.9643 ×37 α 1.8928 ×37 α 1.8214 ×37 α 1.75 ×37 α 1.6786 ×37 α 1.6071 ×37 α 1.5357 ×37 α 1.4643 ×37 α 1.3928 ×37 α 1.3214 ×37
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 1.3214
seen_* probed at reference α = 1.4 · unseen_* = held-out pairings, uniform (α-independent) · 1000 pairs per condition
| iteration | seen_match | seen_mismatch | unseen_match | unseen_mismatch |
|---|---|---|---|---|
| 25 | 50.5 ± 2.0 | 49.5 ± 1.6 | 50.0 ± 1.7 | 49.8 ± 1.9 |
| 50 | 51.4 ± 1.8 | 50.0 ± 1.0 | 49.7 ± 2.2 | 49.8 ± 1.7 |
| 75 | 52.4 ± 2.2 | 49.3 ± 1.6 | 51.0 ± 1.9 | 49.0 ± 1.7 |
| 100 | 55.7 ± 5.8 | 49.6 ± 2.6 | 53.9 ± 5.4 | 48.7 ± 2.2 |
| 125 | 68.0 ± 16.6 | 55.9 ± 7.2 | 64.8 ± 17.7 | 52.5 ± 5.9 |
| 150 | 84.2 ± 16.3 | 70.6 ± 13.6 | 81.0 ± 18.2 | 66.6 ± 13.8 |
| 175 | 92.8 ± 10.1 | 84.0 ± 15.1 | 89.8 ± 14.1 | 80.3 ± 16.4 |
| 200 | 97.4 ± 4.5 | 91.2 ± 12.5 | 95.3 ± 7.4 | 87.7 ± 13.4 |
| 225 | 97.8 ± 3.2 | 94.3 ± 8.1 | 95.2 ± 6.4 | 90.1 ± 9.1 |
| 250 | 97.9 ± 2.3 | 93.7 ± 9.9 | 94.5 ± 5.6 | 88.5 ± 10.9 |
| 275 | 97.4 ± 2.6 | 95.5 ± 3.3 | 93.0 ± 6.7 | 89.2 ± 8.6 |
| 300 | 97.1 ± 2.5 | 94.8 ± 3.6 | 91.9 ± 5.5 | 87.3 ± 8.1 |
| 325 | 96.9 ± 2.0 | 94.8 ± 3.2 | 90.1 ± 6.6 | 85.2 ± 9.2 |
| 350 | 96.5 ± 2.3 | 94.1 ± 2.7 | 89.7 ± 7.0 | 84.6 ± 8.5 |
| 375 | 96.4 ± 1.9 | 94.0 ± 2.6 | 88.5 ± 6.0 | 82.9 ± 8.0 |
| 400 | 96.6 ± 1.8 | 94.6 ± 2.6 | 88.5 ± 6.6 | 82.8 ± 8.5 |
| 450 | 96.4 ± 1.4 | 93.7 ± 2.2 | 86.8 ± 5.2 | 79.9 ± 6.4 |
| 500 | 96.8 ± 1.0 | 94.8 ± 1.7 | 87.8 ± 4.4 | 80.6 ± 6.4 |
| 550 | 97.5 ± 0.8 | 95.1 ± 1.4 | 89.1 ± 4.1 | 81.3 ± 5.6 |
| 600 | 97.8 ± 0.6 | 95.8 ± 1.5 | 90.2 ± 2.7 | 82.1 ± 6.8 |
| 650 | 98.5 ± 0.5 | 96.3 ± 1.5 | 91.3 ± 2.6 | 82.4 ± 7.5 |
| 700 | 98.9 ± 0.4 | 96.8 ± 2.2 | 93.7 ± 2.6 | 83.5 ± 9.1 |
| 750 | 99.2 ± 0.3 | 96.9 ± 2.2 | 94.6 ± 2.6 | 84.0 ± 9.6 |
| 800 | 99.6 ± 0.2 | 97.2 ± 2.4 | 95.8 ± 2.1 | 84.6 ± 10.8 |
| 850 | 99.8 ± 0.2 | 97.3 ± 3.0 | 97.3 ± 1.5 | 83.5 ± 13.4 |
| 900 | 99.7 ± 0.3 | 96.8 ± 3.9 | 97.4 ± 1.9 | 82.3 ± 16.1 |