fixed_1.2
done · peak unseen_mismatch 50.1
α schedule
α 1.2 ×400
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 1.2
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 ± 1.9 | 50.0 ± 1.2 | 50.1 ± 1.6 | 50.1 ± 1.7 |
| 110 | 51.6 ± 1.5 | 49.2 ± 1.6 | 50.7 ± 2.0 | 49.7 ± 1.7 |
| 120 | 51.1 ± 2.3 | 49.4 ± 2.1 | 51.3 ± 2.2 | 49.4 ± 1.7 |
| 130 | 51.1 ± 1.7 | 49.4 ± 1.7 | 51.4 ± 1.4 | 49.6 ± 1.6 |
| 140 | 52.0 ± 3.2 | 48.6 ± 2.2 | 51.6 ± 2.4 | 49.2 ± 2.3 |
| 150 | 53.6 ± 7.4 | 47.9 ± 4.1 | 53.8 ± 6.8 | 47.5 ± 4.9 |
| 160 | 55.6 ± 10.3 | 46.2 ± 8.5 | 55.3 ± 10.5 | 46.3 ± 8.2 |
| 170 | 58.5 ± 11.6 | 45.3 ± 9.6 | 57.5 ± 11.9 | 45.1 ± 9.6 |
| 180 | 64.0 ± 16.8 | 43.3 ± 9.9 | 63.1 ± 16.9 | 42.8 ± 10.8 |
| 190 | 69.1 ± 20.2 | 41.4 ± 11.6 | 68.2 ± 20.6 | 41.0 ± 11.7 |
| 200 | 72.5 ± 23.2 | 38.0 ± 13.8 | 72.3 ± 23.7 | 37.5 ± 14.3 |
| 210 | 74.7 ± 23.2 | 33.1 ± 17.3 | 74.5 ± 23.4 | 33.3 ± 17.4 |
| 220 | 77.6 ± 21.9 | 35.5 ± 18.7 | 76.8 ± 22.7 | 35.3 ± 19.4 |
| 230 | 79.2 ± 22.3 | 35.9 ± 18.9 | 78.6 ± 23.2 | 35.6 ± 19.1 |
| 240 | 81.6 ± 22.3 | 33.5 ± 17.3 | 81.0 ± 23.1 | 33.8 ± 18.0 |
| 250 | 82.3 ± 21.9 | 34.0 ± 18.7 | 81.9 ± 22.3 | 33.6 ± 18.7 |
| 260 | 83.6 ± 21.0 | 36.2 ± 17.7 | 83.1 ± 21.2 | 35.4 ± 17.7 |
| 270 | 85.8 ± 19.8 | 31.9 ± 18.7 | 86.2 ± 19.3 | 31.8 ± 18.3 |
| 280 | 89.8 ± 17.5 | 33.6 ± 16.8 | 89.5 ± 17.7 | 33.6 ± 17.0 |
| 290 | 91.7 ± 17.1 | 33.9 ± 19.9 | 91.7 ± 16.8 | 33.6 ± 19.9 |
| 300 | 93.1 ± 16.2 | 33.5 ± 19.8 | 93.4 ± 15.6 | 32.7 ± 19.0 |
| 310 | 93.7 ± 15.1 | 33.6 ± 20.5 | 93.7 ± 15.0 | 33.4 ± 20.6 |
| 320 | 95.0 ± 13.0 | 35.0 ± 21.1 | 94.6 ± 13.7 | 35.0 ± 21.2 |
| 330 | 96.4 ± 9.9 | 37.0 ± 23.6 | 96.1 ± 10.8 | 36.5 ± 23.0 |
| 340 | 98.8 ± 3.5 | 37.3 ± 25.6 | 98.8 ± 3.5 | 37.0 ± 24.4 |
| 350 | 99.9 ± 0.4 | 31.0 ± 16.9 | 99.9 ± 0.4 | 30.7 ± 17.1 |
| 360 | 100.0 ± 0.0 | 36.5 ± 18.8 | 100.0 ± 0.0 | 36.3 ± 18.8 |
| 370 | 100.0 ± 0.1 | 44.5 ± 16.4 | 100.0 ± 0.0 | 45.1 ± 16.1 |
| 380 | 100.0 ± 0.0 | 47.0 ± 14.3 | 100.0 ± 0.0 | 47.6 ± 14.5 |
| 390 | 100.0 ± 0.0 | 45.8 ± 16.6 | 100.0 ± 0.0 | 45.6 ± 16.5 |
| 400 | 100.0 ± 0.0 | 38.6 ± 15.8 | 100.0 ± 0.0 | 38.6 ± 15.4 |