free-89bf40a8
done · peak unseen_mismatch 99.3
α schedule
α 2.622 ×83 α 0.721 ×11 α 3.082 ×28 α 1.09 ×22 α 2.609 ×36 α 1.03 ×51 α 2.941 ×22 α 0.6049 ×18 α 2.656 ×29 α 2.656 ×100
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 2.656
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 | 53.3 ± 4.0 | 50.1 ± 2.0 | 51.3 ± 3.4 | 49.4 ± 1.7 |
| 110 | 55.3 ± 6.7 | 50.9 ± 3.0 | 53.0 ± 6.2 | 49.7 ± 1.7 |
| 120 | 58.8 ± 10.3 | 52.1 ± 5.6 | 56.2 ± 9.9 | 50.6 ± 4.3 |
| 130 | 62.4 ± 13.3 | 53.8 ± 7.2 | 59.6 ± 13.2 | 51.6 ± 5.7 |
| 140 | 66.4 ± 16.2 | 55.1 ± 9.0 | 63.8 ± 16.4 | 52.7 ± 7.5 |
| 150 | 71.4 ± 17.6 | 56.4 ± 11.4 | 68.9 ± 18.3 | 54.2 ± 10.6 |
| 160 | 77.0 ± 17.6 | 60.4 ± 13.3 | 74.3 ± 18.7 | 57.3 ± 12.7 |
| 170 | 82.4 ± 16.8 | 65.8 ± 14.3 | 79.0 ± 17.9 | 62.2 ± 13.9 |
| 180 | 87.0 ± 14.9 | 72.3 ± 15.9 | 83.6 ± 16.8 | 68.4 ± 15.1 |
| 190 | 90.4 ± 14.0 | 75.4 ± 15.2 | 88.1 ± 15.2 | 72.2 ± 14.6 |
| 200 | 93.9 ± 12.1 | 75.6 ± 18.8 | 92.8 ± 13.2 | 73.0 ± 18.5 |
| 210 | 95.6 ± 9.0 | 72.7 ± 18.6 | 94.9 ± 10.1 | 70.7 ± 18.2 |
| 220 | 98.3 ± 4.0 | 69.6 ± 16.2 | 97.8 ± 4.9 | 67.6 ± 15.6 |
| 230 | 99.6 ± 1.6 | 65.2 ± 17.3 | 99.5 ± 2.2 | 64.0 ± 16.9 |
| 240 | 99.9 ± 0.1 | 65.6 ± 21.7 | 99.8 ± 0.5 | 64.6 ± 21.4 |
| 250 | 100.0 ± 0.1 | 71.8 ± 22.5 | 99.9 ± 0.4 | 70.9 ± 22.3 |
| 260 | 100.0 ± 0.0 | 75.6 ± 23.1 | 100.0 ± 0.0 | 75.0 ± 22.8 |
| 270 | 100.0 ± 0.0 | 61.8 ± 21.2 | 100.0 ± 0.0 | 61.5 ± 20.1 |
| 280 | 100.0 ± 0.1 | 59.0 ± 21.4 | 100.0 ± 0.1 | 57.8 ± 21.1 |
| 290 | 100.0 ± 0.0 | 59.5 ± 13.5 | 100.0 ± 0.0 | 58.7 ± 13.3 |
| 300 | 100.0 ± 0.1 | 68.8 ± 18.0 | 99.9 ± 0.2 | 68.1 ± 17.3 |
| 310 | 99.9 ± 0.1 | 71.2 ± 16.8 | 99.7 ± 0.6 | 70.6 ± 16.0 |
| 320 | 99.8 ± 0.5 | 81.9 ± 18.4 | 99.5 ± 1.3 | 80.6 ± 18.5 |
| 330 | 100.0 ± 0.0 | 76.2 ± 15.9 | 100.0 ± 0.0 | 75.7 ± 15.9 |
| 340 | 100.0 ± 0.0 | 64.3 ± 13.7 | 100.0 ± 0.0 | 63.3 ± 12.7 |
| 350 | 100.0 ± 0.0 | 72.3 ± 19.3 | 100.0 ± 0.0 | 71.5 ± 19.4 |
| 360 | 100.0 ± 0.0 | 90.7 ± 9.0 | 100.0 ± 0.0 | 88.5 ± 9.6 |
| 370 | 100.0 ± 0.0 | 68.6 ± 0.0 | 100.0 ± 0.0 | 67.9 ± 0.0 |
| 380 | 100.0 ± 0.0 | 88.0 ± 0.0 | 100.0 ± 0.0 | 88.0 ± 0.0 |
| 390 | 100.0 ± 0.0 | 98.8 ± 0.0 | 100.0 ± 0.0 | 99.3 ± 0.0 |