fixed_2.0
done · peak unseen_mismatch 88.0
α schedule
α 2.0 ×400
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 2.0
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 | 52.3 ± 4.2 | 50.0 ± 1.1 | 51.1 ± 3.7 | 50.0 ± 1.6 |
| 110 | 52.9 ± 3.3 | 50.2 ± 1.9 | 51.5 ± 3.4 | 49.9 ± 1.8 |
| 120 | 54.4 ± 9.5 | 51.1 ± 4.7 | 53.9 ± 9.5 | 50.7 ± 4.4 |
| 130 | 54.8 ± 10.8 | 52.5 ± 8.2 | 54.1 ± 10.8 | 51.6 ± 8.0 |
| 140 | 56.9 ± 11.1 | 52.2 ± 7.0 | 55.6 ± 11.1 | 51.2 ± 6.7 |
| 150 | 60.3 ± 12.4 | 54.0 ± 11.0 | 58.3 ± 12.2 | 51.7 ± 10.5 |
| 160 | 63.7 ± 13.0 | 54.2 ± 9.1 | 61.9 ± 12.3 | 52.0 ± 7.3 |
| 170 | 67.6 ± 15.7 | 56.4 ± 10.4 | 65.7 ± 15.8 | 53.8 ± 10.0 |
| 180 | 73.3 ± 17.5 | 58.5 ± 13.2 | 71.1 ± 17.4 | 55.9 ± 12.8 |
| 190 | 80.0 ± 17.8 | 62.4 ± 15.4 | 77.7 ± 18.3 | 59.5 ± 14.6 |
| 200 | 85.5 ± 15.7 | 67.5 ± 18.2 | 82.8 ± 17.7 | 64.9 ± 17.3 |
| 210 | 89.9 ± 14.4 | 71.1 ± 18.4 | 87.6 ± 16.2 | 68.2 ± 17.3 |
| 220 | 91.7 ± 14.6 | 72.0 ± 20.0 | 89.9 ± 15.1 | 69.1 ± 18.8 |
| 230 | 92.3 ± 13.7 | 69.8 ± 19.9 | 90.0 ± 15.0 | 67.2 ± 18.2 |
| 240 | 93.8 ± 12.6 | 77.1 ± 20.2 | 91.6 ± 14.0 | 73.3 ± 19.2 |
| 250 | 94.7 ± 12.5 | 77.8 ± 20.6 | 93.1 ± 13.2 | 74.6 ± 19.1 |
| 260 | 94.6 ± 12.6 | 77.0 ± 20.1 | 92.8 ± 13.5 | 73.7 ± 18.6 |
| 270 | 93.5 ± 13.1 | 77.8 ± 17.6 | 90.9 ± 13.8 | 73.9 ± 16.3 |
| 280 | 94.4 ± 11.1 | 84.0 ± 16.9 | 92.2 ± 12.3 | 80.6 ± 16.0 |
| 290 | 95.5 ± 7.6 | 86.5 ± 15.5 | 92.6 ± 9.5 | 81.5 ± 15.0 |
| 300 | 96.2 ± 5.0 | 85.2 ± 9.7 | 92.4 ± 7.7 | 78.6 ± 8.4 |
| 310 | 98.1 ± 1.7 | 84.7 ± 17.3 | 96.6 ± 2.8 | 81.6 ± 16.1 |
| 320 | 98.8 ± 1.3 | 87.0 ± 11.7 | 98.1 ± 2.2 | 83.8 ± 11.6 |
| 330 | 98.4 ± 1.6 | 90.0 ± 7.3 | 95.4 ± 4.3 | 85.4 ± 8.2 |
| 340 | 98.3 ± 1.7 | 95.5 ± 3.9 | 94.3 ± 5.7 | 88.0 ± 9.2 |
| 350 | 97.2 ± 0.0 | 93.7 ± 0.0 | 92.1 ± 0.0 | 83.2 ± 0.0 |
| 360 | 97.4 ± 0.0 | 93.1 ± 0.0 | 92.3 ± 0.0 | 87.4 ± 0.0 |
| 370 | 97.2 ± 0.0 | 94.3 ± 0.0 | 90.9 ± 0.0 | 84.0 ± 0.0 |
| 380 | 96.0 ± 0.0 | 93.3 ± 0.0 | 86.9 ± 0.0 | 80.7 ± 0.0 |
| 390 | 97.3 ± 0.0 | 94.2 ± 0.0 | 90.7 ± 0.0 | 84.4 ± 0.0 |
| 400 | 97.4 ± 0.0 | 94.0 ± 0.0 | 90.9 ± 0.0 | 82.9 ± 0.0 |