free-c1f8608e
done · peak unseen_mismatch 89.0
α schedule
α 2.978 ×66 α 2.106 ×115 α 1.495 ×119 α 1.495 ×100
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 1.495
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 |
|---|---|---|---|---|
| 100 | 54.7 ± 7.2 | 51.2 ± 4.0 | 52.5 ± 6.8 | 50.1 ± 2.9 |
| 110 | 56.7 ± 9.7 | 52.4 ± 5.4 | 53.6 ± 9.2 | 50.5 ± 4.1 |
| 120 | 59.7 ± 10.8 | 54.5 ± 7.3 | 56.1 ± 10.5 | 51.7 ± 5.3 |
| 130 | 64.0 ± 13.1 | 58.0 ± 10.3 | 59.8 ± 12.8 | 54.1 ± 8.2 |
| 140 | 68.7 ± 15.3 | 61.1 ± 11.7 | 64.2 ± 15.5 | 56.7 ± 9.8 |
| 150 | 74.7 ± 16.6 | 66.7 ± 13.0 | 69.9 ± 17.4 | 61.2 ± 12.0 |
| 160 | 80.5 ± 16.5 | 72.2 ± 14.9 | 76.0 ± 18.1 | 66.6 ± 14.3 |
| 170 | 84.7 ± 15.6 | 77.1 ± 13.9 | 79.8 ± 17.6 | 70.7 ± 14.1 |
| 180 | 87.3 ± 14.6 | 80.6 ± 13.3 | 82.7 ± 16.6 | 74.6 ± 13.5 |
| 190 | 89.3 ± 14.3 | 83.5 ± 12.9 | 84.9 ± 16.3 | 77.4 ± 13.4 |
| 200 | 91.2 ± 13.7 | 84.1 ± 14.0 | 87.8 ± 15.2 | 78.9 ± 13.7 |
| 210 | 93.2 ± 12.2 | 85.4 ± 13.7 | 89.9 ± 14.3 | 80.4 ± 13.9 |
| 220 | 93.5 ± 12.5 | 85.3 ± 15.3 | 90.2 ± 14.3 | 80.1 ± 15.0 |
| 230 | 93.6 ± 12.2 | 84.0 ± 15.0 | 90.4 ± 14.2 | 78.8 ± 13.9 |
| 240 | 94.4 ± 12.3 | 85.4 ± 15.4 | 91.4 ± 13.7 | 80.2 ± 14.4 |
| 250 | 94.1 ± 12.5 | 87.0 ± 15.6 | 91.9 ± 13.6 | 82.9 ± 13.8 |
| 260 | 94.0 ± 11.3 | 84.0 ± 17.1 | 90.7 ± 12.3 | 79.7 ± 14.6 |
| 270 | 94.7 ± 8.2 | 81.6 ± 17.2 | 92.0 ± 10.0 | 78.1 ± 14.2 |
| 280 | 95.8 ± 6.6 | 81.5 ± 16.2 | 93.0 ± 8.4 | 78.9 ± 13.4 |
| 290 | 96.2 ± 6.9 | 84.6 ± 17.0 | 92.8 ± 9.8 | 81.2 ± 16.4 |
| 300 | 94.9 ± 6.7 | 80.5 ± 11.4 | 91.4 ± 7.9 | 76.1 ± 5.5 |
| 310 | 97.2 ± 3.8 | 77.7 ± 18.7 | 95.4 ± 4.2 | 75.5 ± 12.3 |
| 320 | 97.2 ± 3.7 | 94.0 ± 4.0 | 95.3 ± 4.0 | 88.9 ± 4.1 |
| 330 | 98.3 ± 1.1 | 94.9 ± 1.7 | 95.5 ± 1.0 | 88.5 ± 4.8 |
| 340 | 99.6 ± 0.0 | 97.2 ± 0.0 | 94.8 ± 0.0 | 81.3 ± 0.0 |
| 350 | 98.8 ± 0.0 | 93.6 ± 0.0 | 92.4 ± 0.0 | 78.7 ± 0.0 |
| 360 | 99.7 ± 0.0 | 97.2 ± 0.0 | 95.1 ± 0.0 | 84.1 ± 0.0 |
| 370 | 99.5 ± 0.0 | 97.9 ± 0.0 | 96.3 ± 0.0 | 89.0 ± 0.0 |
| 380 | 99.1 ± 0.0 | 97.5 ± 0.0 | 93.1 ± 0.0 | 82.2 ± 0.0 |
| 390 | 99.0 ± 0.0 | 97.5 ± 0.0 | 92.6 ± 0.0 | 80.0 ± 0.0 |
| 400 | 99.1 ± 0.0 | 97.1 ± 0.0 | 95.1 ± 0.0 | 84.0 ± 0.0 |