free-fecc32d7
done · peak unseen_mismatch 95.0
α schedule
α 4.582 ×174 α 0.04871 ×50 α 4.343 ×40 α 0.4149 ×36 α 0.4149 ×100
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 0.4149
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 | 62.1 ± 8.7 | 57.9 ± 7.5 | 54.6 ± 7.1 | 52.0 ± 4.2 |
| 110 | 67.1 ± 9.9 | 63.2 ± 9.1 | 58.0 ± 8.9 | 54.5 ± 6.6 |
| 120 | 71.8 ± 9.6 | 67.9 ± 8.7 | 60.4 ± 9.4 | 56.7 ± 6.9 |
| 130 | 76.0 ± 8.1 | 72.4 ± 7.3 | 62.3 ± 9.2 | 58.7 ± 7.1 |
| 140 | 78.6 ± 6.9 | 75.3 ± 6.1 | 64.2 ± 9.0 | 61.1 ± 7.5 |
| 150 | 79.5 ± 5.2 | 76.8 ± 4.2 | 63.9 ± 7.2 | 61.0 ± 5.9 |
| 160 | 80.2 ± 4.3 | 77.6 ± 3.6 | 64.1 ± 6.9 | 61.3 ± 5.8 |
| 170 | 81.0 ± 3.8 | 78.4 ± 2.9 | 63.9 ± 6.1 | 61.4 ± 4.8 |
| 180 | 82.2 ± 4.4 | 79.7 ± 3.4 | 66.2 ± 6.6 | 63.6 ± 5.2 |
| 190 | 86.8 ± 7.2 | 83.9 ± 6.9 | 77.2 ± 10.4 | 73.6 ± 9.1 |
| 200 | 94.9 ± 5.0 | 92.6 ± 7.0 | 91.7 ± 8.1 | 88.3 ± 9.4 |
| 210 | 97.9 ± 3.5 | 93.7 ± 10.5 | 96.8 ± 5.0 | 91.4 ± 12.2 |
| 220 | 99.4 ± 1.1 | 79.5 ± 22.5 | 99.0 ± 2.2 | 78.2 ± 23.1 |
| 230 | 100.0 ± 0.0 | 69.4 ± 21.0 | 100.0 ± 0.0 | 67.1 ± 22.7 |
| 240 | 99.8 ± 0.2 | 77.0 ± 21.4 | 99.4 ± 0.6 | 74.0 ± 23.8 |
| 250 | 100.0 ± 0.0 | 73.8 ± 0.0 | 100.0 ± 0.0 | 66.7 ± 0.0 |
| 260 | 100.0 ± 0.0 | 89.5 ± 0.0 | 100.0 ± 0.0 | 84.1 ± 0.0 |
| 270 | 100.0 ± 0.0 | 91.2 ± 0.0 | 99.4 ± 0.0 | 83.9 ± 0.0 |
| 280 | 100.0 ± 0.0 | 94.7 ± 0.0 | 100.0 ± 0.0 | 89.6 ± 0.0 |
| 290 | 100.0 ± 0.0 | 84.7 ± 0.0 | 100.0 ± 0.0 | 80.4 ± 0.0 |
| 300 | 100.0 ± 0.0 | 86.3 ± 0.0 | 100.0 ± 0.0 | 83.1 ± 0.0 |
| 310 | 100.0 ± 0.0 | 97.3 ± 0.0 | 100.0 ± 0.0 | 95.0 ± 0.0 |