free-f31c9ee1
done · peak unseen_mismatch 96.6
α schedule
α 4.602 ×27 α 1.389 ×27 α 4.726 ×18 α 1.443 ×20 α 3.924 ×21 α 1.488 ×19 α 4.378 ×15 α 0.6833 ×11 α 4.681 ×26 α 1.047 ×18 α 4.463 ×13 α 1.403 ×46 α 4.329 ×21 α 0.807 ×18 α 0.807 ×100
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 0.807
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 | 53.4 ± 3.9 | 50.3 ± 2.0 | 51.8 ± 3.5 | 49.8 ± 1.6 |
| 110 | 55.9 ± 7.0 | 51.2 ± 3.3 | 53.9 ± 6.7 | 49.9 ± 2.5 |
| 120 | 60.3 ± 11.2 | 53.1 ± 5.2 | 57.7 ± 10.7 | 50.9 ± 4.5 |
| 130 | 65.0 ± 15.6 | 55.3 ± 9.0 | 62.6 ± 15.4 | 52.8 ± 8.0 |
| 140 | 69.9 ± 17.1 | 59.0 ± 11.1 | 66.9 ± 17.3 | 55.7 ± 9.9 |
| 150 | 75.4 ± 18.1 | 64.4 ± 14.2 | 72.4 ± 18.3 | 61.3 ± 12.7 |
| 160 | 80.2 ± 18.8 | 67.0 ± 15.3 | 78.1 ± 19.4 | 64.5 ± 14.9 |
| 170 | 84.8 ± 17.0 | 71.4 ± 15.3 | 82.1 ± 18.7 | 68.5 ± 15.4 |
| 180 | 85.6 ± 16.1 | 75.5 ± 14.2 | 83.1 ± 17.0 | 72.3 ± 13.7 |
| 190 | 87.8 ± 15.1 | 79.7 ± 14.4 | 85.2 ± 16.1 | 76.3 ± 14.1 |
| 200 | 90.9 ± 14.5 | 81.3 ± 15.3 | 89.7 ± 15.4 | 79.5 ± 15.0 |
| 210 | 92.1 ± 12.6 | 78.0 ± 15.6 | 90.7 ± 14.1 | 75.6 ± 15.5 |
| 220 | 93.4 ± 10.9 | 81.6 ± 14.9 | 90.9 ± 13.4 | 78.4 ± 14.9 |
| 230 | 95.3 ± 10.1 | 82.8 ± 15.2 | 93.6 ± 11.6 | 80.0 ± 14.8 |
| 240 | 95.1 ± 11.5 | 79.0 ± 15.9 | 94.2 ± 12.4 | 76.7 ± 14.3 |
| 250 | 96.3 ± 9.7 | 79.0 ± 14.8 | 95.3 ± 11.3 | 78.1 ± 13.4 |
| 260 | 96.3 ± 8.0 | 77.6 ± 17.8 | 95.4 ± 9.6 | 76.0 ± 17.1 |
| 270 | 97.1 ± 5.3 | 75.6 ± 16.4 | 96.5 ± 5.9 | 74.7 ± 15.7 |
| 280 | 98.1 ± 2.7 | 86.1 ± 11.9 | 97.3 ± 3.4 | 84.4 ± 10.9 |
| 290 | 97.7 ± 2.7 | 86.2 ± 8.7 | 97.5 ± 3.0 | 84.3 ± 8.4 |
| 300 | 99.7 ± 0.2 | 84.3 ± 6.1 | 99.6 ± 0.4 | 80.5 ± 6.4 |
| 310 | 100.0 ± 0.0 | 88.9 ± 3.6 | 100.0 ± 0.0 | 86.3 ± 3.6 |
| 320 | 100.0 ± 0.0 | 87.4 ± 4.2 | 100.0 ± 0.0 | 85.2 ± 4.7 |
| 330 | 100.0 ± 0.0 | 82.7 ± 11.3 | 100.0 ± 0.0 | 81.4 ± 11.9 |
| 340 | 100.0 ± 0.0 | 76.3 ± 13.6 | 100.0 ± 0.0 | 73.7 ± 12.2 |
| 350 | 100.0 ± 0.0 | 71.8 ± 13.8 | 100.0 ± 0.0 | 71.0 ± 12.9 |
| 360 | 100.0 ± 0.0 | 64.5 ± 12.3 | 100.0 ± 0.0 | 63.9 ± 12.8 |
| 370 | 100.0 ± 0.0 | 73.0 ± 18.0 | 100.0 ± 0.0 | 72.0 ± 18.5 |
| 380 | 100.0 ± 0.0 | 59.5 ± 0.0 | 99.9 ± 0.0 | 59.6 ± 0.0 |
| 390 | 100.0 ± 0.0 | 80.4 ± 0.0 | 100.0 ± 0.0 | 76.6 ± 0.0 |
| 400 | 100.0 ± 0.0 | 97.0 ± 0.0 | 100.0 ± 0.0 | 96.6 ± 0.0 |