fixed_1.8
done · peak unseen_mismatch 98.1
α schedule
α 1.8 ×400
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 1.8
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 | 51.8 ± 2.4 | 49.7 ± 1.2 | 50.3 ± 2.0 | 49.8 ± 2.0 |
| 110 | 51.6 ± 1.9 | 49.5 ± 1.6 | 50.4 ± 2.0 | 50.0 ± 1.7 |
| 120 | 51.9 ± 2.8 | 49.7 ± 1.5 | 51.4 ± 2.7 | 49.5 ± 1.2 |
| 130 | 52.5 ± 4.0 | 50.2 ± 2.0 | 51.6 ± 4.0 | 49.8 ± 2.0 |
| 140 | 54.5 ± 8.3 | 49.8 ± 2.4 | 54.0 ± 7.8 | 49.9 ± 1.6 |
| 150 | 56.8 ± 10.3 | 49.8 ± 4.2 | 56.1 ± 9.3 | 48.9 ± 3.8 |
| 160 | 60.3 ± 12.8 | 50.8 ± 5.2 | 58.6 ± 12.8 | 50.2 ± 4.4 |
| 170 | 65.4 ± 17.1 | 53.2 ± 6.3 | 63.4 ± 17.1 | 51.7 ± 4.8 |
| 180 | 69.3 ± 19.7 | 57.4 ± 10.2 | 68.4 ± 20.1 | 56.1 ± 9.8 |
| 190 | 72.1 ± 21.4 | 58.1 ± 13.1 | 70.7 ± 22.3 | 56.9 ± 12.6 |
| 200 | 76.7 ± 19.1 | 61.7 ± 17.5 | 75.1 ± 20.7 | 59.9 ± 17.7 |
| 210 | 83.8 ± 16.5 | 61.8 ± 15.0 | 81.4 ± 17.7 | 59.6 ± 14.9 |
| 220 | 87.6 ± 16.7 | 69.9 ± 13.4 | 86.4 ± 17.6 | 68.2 ± 13.9 |
| 230 | 91.5 ± 15.7 | 72.1 ± 13.3 | 90.1 ± 16.5 | 70.2 ± 13.1 |
| 240 | 92.2 ± 15.3 | 73.6 ± 15.3 | 91.6 ± 15.4 | 71.9 ± 15.9 |
| 250 | 91.6 ± 15.8 | 72.3 ± 16.9 | 90.7 ± 16.0 | 70.5 ± 17.0 |
| 260 | 92.6 ± 15.1 | 70.2 ± 19.3 | 91.6 ± 16.2 | 68.1 ± 19.3 |
| 270 | 94.4 ± 11.0 | 70.1 ± 22.9 | 93.4 ± 12.4 | 68.0 ± 22.5 |
| 280 | 97.2 ± 5.0 | 76.0 ± 23.6 | 96.8 ± 5.2 | 74.0 ± 23.5 |
| 290 | 98.8 ± 1.8 | 77.3 ± 18.0 | 98.6 ± 1.5 | 75.8 ± 17.9 |
| 300 | 99.6 ± 0.6 | 61.2 ± 14.1 | 99.1 ± 1.1 | 59.8 ± 13.6 |
| 310 | 99.6 ± 0.3 | 64.8 ± 12.2 | 99.5 ± 0.7 | 62.4 ± 11.3 |
| 320 | 100.0 ± 0.1 | 66.7 ± 11.4 | 99.7 ± 0.6 | 64.9 ± 8.8 |
| 330 | 100.0 ± 0.0 | 85.9 ± 14.4 | 99.8 ± 0.3 | 84.5 ± 13.5 |
| 340 | 100.0 ± 0.0 | 89.9 ± 7.0 | 100.0 ± 0.0 | 89.1 ± 6.1 |
| 350 | 100.0 ± 0.0 | 98.1 ± 1.2 | 100.0 ± 0.0 | 98.1 ± 1.0 |