fixed_1.1
done · peak unseen_mismatch 50.3
α schedule
α 1.1 ×400
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 1.1
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 | 51.1 ± 2.5 | 49.6 ± 1.3 | 50.3 ± 2.4 | 50.3 ± 1.8 |
| 110 | 51.6 ± 1.6 | 49.5 ± 1.7 | 50.9 ± 1.3 | 50.0 ± 1.7 |
| 120 | 51.3 ± 2.1 | 49.4 ± 1.8 | 50.7 ± 2.0 | 49.8 ± 1.5 |
| 130 | 51.8 ± 4.2 | 49.2 ± 2.8 | 51.9 ± 3.8 | 49.3 ± 2.3 |
| 140 | 52.4 ± 5.8 | 48.6 ± 3.7 | 52.4 ± 5.0 | 48.8 ± 3.7 |
| 150 | 53.1 ± 8.9 | 48.1 ± 5.1 | 53.4 ± 8.4 | 48.1 ± 5.2 |
| 160 | 54.8 ± 10.9 | 47.8 ± 4.9 | 54.2 ± 11.0 | 47.8 ± 4.9 |
| 170 | 56.7 ± 13.5 | 46.4 ± 8.8 | 55.8 ± 13.7 | 46.7 ± 8.5 |
| 180 | 58.6 ± 16.3 | 46.9 ± 7.7 | 58.0 ± 16.3 | 46.9 ± 7.7 |
| 190 | 60.4 ± 17.4 | 45.6 ± 9.5 | 59.9 ± 17.7 | 45.4 ± 8.9 |
| 200 | 65.1 ± 19.5 | 41.1 ± 13.3 | 64.4 ± 19.8 | 42.0 ± 12.8 |
| 210 | 66.8 ± 20.8 | 44.8 ± 10.6 | 66.2 ± 21.3 | 44.9 ± 10.0 |
| 220 | 70.0 ± 22.4 | 40.2 ± 14.8 | 69.3 ± 23.1 | 40.3 ± 14.3 |
| 230 | 70.8 ± 22.9 | 41.1 ± 15.1 | 70.2 ± 23.1 | 40.8 ± 15.2 |
| 240 | 72.2 ± 22.4 | 41.8 ± 14.7 | 71.3 ± 23.2 | 42.1 ± 14.9 |
| 250 | 73.8 ± 21.9 | 38.7 ± 16.6 | 73.0 ± 22.4 | 38.8 ± 16.4 |
| 260 | 76.6 ± 21.4 | 36.0 ± 16.7 | 76.0 ± 22.0 | 35.2 ± 16.5 |
| 270 | 79.3 ± 22.0 | 33.8 ± 16.9 | 79.2 ± 21.9 | 33.9 ± 17.0 |
| 280 | 82.2 ± 22.7 | 34.6 ± 17.3 | 82.3 ± 22.5 | 34.1 ± 17.3 |
| 290 | 83.4 ± 22.7 | 35.2 ± 18.1 | 83.7 ± 22.3 | 35.1 ± 17.9 |
| 300 | 84.4 ± 21.8 | 36.2 ± 15.9 | 84.6 ± 21.4 | 35.6 ± 15.4 |
| 310 | 87.7 ± 20.2 | 34.4 ± 17.0 | 87.7 ± 20.0 | 34.1 ± 16.5 |
| 320 | 88.9 ± 19.9 | 35.4 ± 19.8 | 89.0 ± 19.6 | 35.1 ± 19.7 |
| 330 | 90.5 ± 18.9 | 34.0 ± 19.9 | 90.3 ± 19.3 | 34.0 ± 19.9 |
| 340 | 91.1 ± 17.9 | 32.8 ± 22.3 | 90.6 ± 18.8 | 32.6 ± 22.3 |
| 350 | 91.4 ± 17.4 | 35.2 ± 15.7 | 91.3 ± 17.7 | 34.3 ± 15.7 |
| 360 | 92.7 ± 15.8 | 31.4 ± 15.4 | 92.6 ± 16.3 | 30.9 ± 15.6 |
| 370 | 95.2 ± 13.7 | 35.2 ± 15.4 | 94.8 ± 14.5 | 34.3 ± 15.3 |
| 380 | 95.5 ± 13.4 | 29.7 ± 17.4 | 95.5 ± 13.5 | 29.7 ± 16.6 |
| 390 | 95.5 ± 13.6 | 36.2 ± 19.3 | 95.4 ± 13.9 | 35.5 ± 19.1 |
| 400 | 95.6 ± 13.2 | 32.8 ± 18.7 | 95.6 ± 13.3 | 32.7 ± 18.6 |