fixed_1.7
done · peak unseen_mismatch 98.3
α schedule
α 1.7 ×400
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 1.7
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 | 51.4 ± 2.4 | 49.7 ± 1.1 | 50.1 ± 2.0 | 50.1 ± 2.0 |
| 110 | 51.5 ± 1.6 | 49.8 ± 1.6 | 50.6 ± 1.4 | 50.0 ± 1.3 |
| 120 | 51.7 ± 2.6 | 49.9 ± 1.7 | 50.8 ± 1.8 | 49.3 ± 1.1 |
| 130 | 51.5 ± 2.2 | 49.9 ± 1.2 | 51.0 ± 1.3 | 49.5 ± 1.5 |
| 140 | 51.7 ± 2.5 | 49.6 ± 1.6 | 51.5 ± 2.6 | 49.4 ± 1.5 |
| 150 | 54.4 ± 6.5 | 49.4 ± 3.0 | 53.9 ± 5.9 | 49.0 ± 2.7 |
| 160 | 57.3 ± 10.9 | 48.2 ± 5.3 | 56.8 ± 11.0 | 48.4 ± 5.3 |
| 170 | 60.9 ± 16.0 | 49.8 ± 6.0 | 59.9 ± 16.6 | 49.3 ± 5.9 |
| 180 | 65.4 ± 17.2 | 51.7 ± 6.3 | 63.9 ± 17.8 | 50.9 ± 5.7 |
| 190 | 69.5 ± 19.0 | 52.9 ± 8.2 | 68.5 ± 19.3 | 51.7 ± 7.7 |
| 200 | 74.5 ± 19.2 | 51.3 ± 12.0 | 73.9 ± 20.0 | 50.9 ± 11.0 |
| 210 | 83.2 ± 17.2 | 57.3 ± 14.5 | 81.8 ± 18.1 | 56.3 ± 13.8 |
| 220 | 88.3 ± 16.2 | 60.3 ± 15.9 | 87.2 ± 16.6 | 59.4 ± 15.6 |
| 230 | 91.2 ± 16.5 | 61.0 ± 13.5 | 90.7 ± 16.4 | 60.5 ± 13.1 |
| 240 | 92.2 ± 16.9 | 62.4 ± 15.9 | 91.9 ± 17.0 | 61.8 ± 15.1 |
| 250 | 93.0 ± 15.5 | 64.1 ± 14.3 | 92.6 ± 16.1 | 62.7 ± 14.2 |
| 260 | 94.0 ± 13.3 | 59.7 ± 17.3 | 93.5 ± 14.1 | 59.3 ± 16.5 |
| 270 | 95.3 ± 11.5 | 65.5 ± 13.3 | 94.9 ± 12.3 | 65.1 ± 13.4 |
| 280 | 96.9 ± 10.2 | 72.3 ± 17.4 | 96.5 ± 11.8 | 70.7 ± 17.5 |
| 290 | 97.2 ± 8.6 | 72.6 ± 17.7 | 96.5 ± 11.1 | 71.0 ± 17.4 |
| 300 | 98.5 ± 4.5 | 69.8 ± 13.9 | 98.1 ± 5.8 | 69.3 ± 13.0 |
| 310 | 99.4 ± 2.0 | 73.6 ± 18.4 | 99.0 ± 3.1 | 72.6 ± 18.0 |
| 320 | 99.5 ± 1.4 | 70.6 ± 18.7 | 99.5 ± 1.5 | 69.5 ± 17.5 |
| 330 | 100.0 ± 0.0 | 78.5 ± 14.7 | 100.0 ± 0.0 | 77.9 ± 14.1 |
| 340 | 100.0 ± 0.0 | 77.8 ± 16.0 | 100.0 ± 0.0 | 77.8 ± 14.3 |
| 350 | 100.0 ± 0.0 | 81.2 ± 16.0 | 100.0 ± 0.0 | 80.8 ± 15.4 |
| 360 | 100.0 ± 0.0 | 79.6 ± 21.1 | 100.0 ± 0.0 | 77.9 ± 19.8 |
| 370 | 99.9 ± 0.1 | 82.0 ± 18.3 | 100.0 ± 0.0 | 80.0 ± 16.3 |
| 380 | 100.0 ± 0.0 | 86.9 ± 7.7 | 100.0 ± 0.0 | 85.1 ± 7.6 |
| 390 | 100.0 ± 0.0 | 88.0 ± 0.0 | 100.0 ± 0.0 | 86.7 ± 0.0 |
| 400 | 100.0 ± 0.0 | 98.8 ± 0.0 | 100.0 ± 0.0 | 98.3 ± 0.0 |