fixed_1.3
done · peak unseen_mismatch 50.2
α schedule
α 1.3 ×400
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 1.3
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 | 50.8 ± 1.9 | 49.9 ± 1.5 | 49.9 ± 1.4 | 50.0 ± 1.6 |
| 110 | 51.6 ± 1.8 | 49.4 ± 1.5 | 50.8 ± 1.7 | 49.6 ± 1.5 |
| 120 | 51.6 ± 2.7 | 49.3 ± 1.7 | 51.7 ± 2.8 | 49.3 ± 1.5 |
| 130 | 50.8 ± 1.7 | 49.8 ± 1.5 | 51.1 ± 1.9 | 49.6 ± 1.6 |
| 140 | 52.2 ± 5.6 | 49.1 ± 2.1 | 52.7 ± 5.2 | 49.1 ± 2.3 |
| 150 | 54.2 ± 9.3 | 48.6 ± 2.3 | 54.4 ± 8.9 | 48.6 ± 2.2 |
| 160 | 55.7 ± 11.0 | 48.0 ± 2.5 | 55.1 ± 11.3 | 48.4 ± 2.5 |
| 170 | 57.8 ± 13.7 | 47.5 ± 4.5 | 57.1 ± 14.1 | 47.2 ± 4.2 |
| 180 | 60.4 ± 15.8 | 45.4 ± 8.6 | 59.8 ± 15.7 | 45.1 ± 9.5 |
| 190 | 64.2 ± 17.9 | 45.8 ± 9.0 | 63.2 ± 17.9 | 44.7 ± 9.2 |
| 200 | 69.7 ± 20.0 | 45.8 ± 11.2 | 68.8 ± 20.6 | 45.7 ± 10.9 |
| 210 | 76.2 ± 19.6 | 42.3 ± 13.6 | 75.1 ± 20.2 | 42.5 ± 13.4 |
| 220 | 82.5 ± 19.7 | 38.0 ± 17.5 | 82.1 ± 19.9 | 38.6 ± 17.3 |
| 230 | 85.8 ± 19.9 | 42.5 ± 12.7 | 85.5 ± 20.5 | 42.3 ± 12.7 |
| 240 | 88.9 ± 19.5 | 38.8 ± 13.6 | 88.6 ± 19.9 | 39.0 ± 13.1 |
| 250 | 89.9 ± 19.3 | 39.6 ± 16.0 | 89.8 ± 19.4 | 39.8 ± 15.0 |
| 260 | 90.4 ± 19.2 | 42.0 ± 16.0 | 90.6 ± 18.7 | 42.3 ± 15.8 |
| 270 | 90.7 ± 18.6 | 41.4 ± 22.3 | 90.8 ± 18.5 | 40.9 ± 22.5 |
| 280 | 91.5 ± 17.5 | 48.7 ± 22.9 | 91.5 ± 17.7 | 48.3 ± 22.8 |
| 290 | 92.4 ± 17.1 | 48.2 ± 21.2 | 92.6 ± 16.8 | 47.6 ± 21.5 |
| 300 | 92.9 ± 16.7 | 45.6 ± 25.1 | 92.9 ± 16.8 | 45.3 ± 24.7 |
| 310 | 92.8 ± 16.6 | 46.9 ± 20.9 | 92.4 ± 17.5 | 46.2 ± 20.6 |
| 320 | 92.8 ± 16.3 | 46.9 ± 20.8 | 92.7 ± 16.4 | 46.8 ± 20.6 |
| 330 | 93.8 ± 14.2 | 42.9 ± 24.2 | 93.6 ± 14.7 | 42.7 ± 24.2 |
| 340 | 95.7 ± 11.1 | 43.8 ± 18.2 | 95.4 ± 11.8 | 43.4 ± 18.4 |
| 350 | 97.9 ± 7.7 | 40.5 ± 18.6 | 98.1 ± 6.7 | 40.3 ± 18.5 |
| 360 | 98.9 ± 4.2 | 38.8 ± 23.2 | 99.0 ± 3.6 | 38.9 ± 23.5 |
| 370 | 99.9 ± 0.2 | 50.4 ± 26.3 | 99.9 ± 0.3 | 50.2 ± 26.4 |
| 380 | 100.0 ± 0.0 | 46.7 ± 25.2 | 100.0 ± 0.0 | 46.6 ± 25.6 |
| 390 | 100.0 ± 0.0 | 41.5 ± 19.7 | 100.0 ± 0.0 | 40.5 ± 19.4 |
| 400 | 100.0 ± 0.0 | 41.6 ± 21.9 | 100.0 ± 0.0 | 40.6 ± 22.1 |