fixed_1.4
done · peak unseen_mismatch 61.1
α schedule
α 1.4 ×400
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 1.4
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 | 51.3 ± 2.0 | 50.0 ± 1.3 | 50.1 ± 1.8 | 50.0 ± 1.7 |
| 110 | 51.7 ± 2.1 | 49.7 ± 1.9 | 50.9 ± 2.1 | 49.8 ± 1.6 |
| 120 | 51.0 ± 1.7 | 49.7 ± 1.6 | 51.3 ± 2.1 | 49.4 ± 1.2 |
| 130 | 52.0 ± 3.5 | 49.5 ± 1.4 | 51.8 ± 3.8 | 49.4 ± 1.9 |
| 140 | 54.0 ± 9.4 | 48.5 ± 3.5 | 53.9 ± 9.0 | 48.9 ± 2.8 |
| 150 | 55.4 ± 10.8 | 48.9 ± 2.4 | 55.2 ± 10.7 | 48.4 ± 2.8 |
| 160 | 57.3 ± 12.6 | 48.2 ± 4.5 | 56.8 ± 13.0 | 48.1 ± 4.5 |
| 170 | 59.9 ± 15.5 | 47.0 ± 5.7 | 58.9 ± 15.9 | 46.9 ± 5.8 |
| 180 | 63.4 ± 18.4 | 45.9 ± 8.2 | 62.6 ± 18.5 | 45.7 ± 9.1 |
| 190 | 67.5 ± 19.8 | 47.2 ± 9.9 | 66.4 ± 20.1 | 46.6 ± 9.5 |
| 200 | 70.4 ± 20.9 | 47.5 ± 11.2 | 69.6 ± 21.1 | 47.4 ± 11.5 |
| 210 | 75.8 ± 21.1 | 44.3 ± 13.6 | 75.1 ± 21.6 | 43.8 ± 13.2 |
| 220 | 81.2 ± 20.4 | 44.8 ± 15.6 | 80.2 ± 21.3 | 44.3 ± 16.1 |
| 230 | 86.0 ± 18.3 | 48.8 ± 19.3 | 85.4 ± 18.8 | 48.0 ± 19.0 |
| 240 | 89.6 ± 17.4 | 49.0 ± 18.2 | 89.3 ± 17.2 | 49.6 ± 18.1 |
| 250 | 92.6 ± 16.1 | 49.9 ± 22.5 | 92.9 ± 15.6 | 50.1 ± 23.2 |
| 260 | 93.9 ± 14.8 | 50.8 ± 24.0 | 93.7 ± 15.2 | 51.1 ± 24.1 |
| 270 | 94.0 ± 15.3 | 46.7 ± 24.4 | 94.1 ± 15.0 | 46.7 ± 23.9 |
| 280 | 94.8 ± 14.2 | 51.9 ± 17.8 | 94.5 ± 14.7 | 51.6 ± 17.4 |
| 290 | 94.6 ± 14.7 | 56.1 ± 16.7 | 94.7 ± 14.3 | 55.8 ± 17.0 |
| 300 | 95.1 ± 13.1 | 54.9 ± 21.8 | 95.1 ± 13.1 | 53.7 ± 21.2 |
| 310 | 95.8 ± 11.5 | 53.2 ± 23.1 | 95.7 ± 11.9 | 52.7 ± 22.4 |
| 320 | 97.5 ± 7.6 | 54.8 ± 23.3 | 97.2 ± 8.0 | 54.5 ± 22.7 |
| 330 | 98.9 ± 2.7 | 47.6 ± 19.8 | 98.9 ± 2.8 | 47.3 ± 19.4 |
| 340 | 99.9 ± 0.3 | 52.1 ± 14.9 | 99.8 ± 0.5 | 51.7 ± 15.0 |
| 350 | 100.0 ± 0.2 | 55.0 ± 20.6 | 99.9 ± 0.2 | 55.0 ± 19.8 |
| 360 | 100.0 ± 0.0 | 60.8 ± 21.5 | 100.0 ± 0.0 | 61.1 ± 20.9 |
| 370 | 100.0 ± 0.0 | 53.5 ± 18.4 | 100.0 ± 0.0 | 53.4 ± 18.5 |
| 380 | 100.0 ± 0.0 | 51.6 ± 23.4 | 100.0 ± 0.0 | 52.1 ± 22.9 |
| 390 | 100.0 ± 0.0 | 55.3 ± 18.3 | 100.0 ± 0.0 | 54.9 ± 18.6 |
| 400 | 100.0 ± 0.0 | 59.2 ± 24.7 | 100.0 ± 0.0 | 59.3 ± 23.2 |