free-8e304386
done · peak unseen_mismatch 97.6
α schedule
α 4.56 ×25 α 4.052 ×15 α 3.387 ×45 α 4.016 ×21 α 4.72 ×13 α 3.618 ×14 α 2.89 ×12 α 1.058 ×33 α 3.946 ×12 α 1.922 ×16 α 2.842 ×11 α 2.129 ×67 α 0.8883 ×16 α 0.8883 ×100
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 0.8883
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 | 58.5 ± 6.0 | 55.3 ± 5.1 | 52.4 ± 3.4 | 50.7 ± 2.3 |
| 110 | 63.1 ± 8.9 | 59.0 ± 7.2 | 55.7 ± 7.1 | 52.1 ± 4.0 |
| 120 | 68.6 ± 10.7 | 64.4 ± 9.1 | 59.1 ± 9.9 | 55.3 ± 6.1 |
| 130 | 72.7 ± 10.5 | 69.2 ± 9.8 | 61.5 ± 10.2 | 58.0 ± 7.9 |
| 140 | 76.4 ± 9.8 | 73.3 ± 9.3 | 64.1 ± 10.5 | 60.6 ± 8.7 |
| 150 | 79.6 ± 9.6 | 76.6 ± 9.4 | 67.6 ± 10.6 | 63.5 ± 9.6 |
| 160 | 85.4 ± 9.9 | 80.6 ± 10.4 | 76.2 ± 12.2 | 70.3 ± 11.2 |
| 170 | 92.0 ± 8.3 | 87.0 ± 9.8 | 86.6 ± 11.6 | 79.4 ± 12.0 |
| 180 | 95.6 ± 5.8 | 88.6 ± 9.1 | 92.1 ± 8.6 | 83.0 ± 10.3 |
| 190 | 94.7 ± 4.1 | 89.9 ± 4.7 | 87.7 ± 9.2 | 80.9 ± 8.0 |
| 200 | 93.1 ± 3.9 | 87.7 ± 10.2 | 83.8 ± 9.7 | 76.9 ± 12.0 |
| 210 | 95.1 ± 2.5 | 91.2 ± 5.5 | 87.1 ± 6.9 | 81.0 ± 6.8 |
| 220 | 93.7 ± 2.1 | 91.4 ± 1.7 | 83.9 ± 6.1 | 79.7 ± 6.0 |
| 230 | 94.7 ± 1.5 | 92.8 ± 2.0 | 85.6 ± 4.8 | 81.8 ± 4.5 |
| 240 | 95.7 ± 1.1 | 94.0 ± 1.2 | 87.8 ± 3.5 | 84.0 ± 3.3 |
| 250 | 95.3 ± 1.0 | 93.6 ± 1.5 | 86.3 ± 3.7 | 82.2 ± 3.4 |
| 260 | 95.3 ± 1.1 | 93.8 ± 1.2 | 85.7 ± 3.9 | 81.4 ± 4.4 |
| 270 | 95.2 ± 1.3 | 93.2 ± 1.3 | 85.2 ± 3.9 | 81.4 ± 4.5 |
| 280 | 95.7 ± 1.1 | 94.0 ± 1.3 | 86.3 ± 4.3 | 82.4 ± 4.7 |
| 290 | 97.8 ± 0.8 | 96.5 ± 1.4 | 92.0 ± 3.3 | 88.2 ± 3.8 |
| 300 | 99.8 ± 0.2 | 99.3 ± 0.5 | 98.9 ± 0.7 | 97.6 ± 1.7 |