start_5
done · peak unseen_mismatch 97.4
α schedule
α 5.0 ×19 α 5.0 ×19 α 5.0 ×19 α 5.0 ×19 α 5.0 ×19 α 5.0 ×19 α 5.0 ×19 α 5.0 ×19 α 1.0 ×19 α 1.0 ×19 α 1.0 ×19 α 1.0 ×19 α 1.0 ×19 α 1.0 ×19 α 1.0 ×19 α 1.0 ×19 α 3.1429 ×19 α 3.1429 ×19 α 3.1429 ×19 α 3.1429 ×19 α 3.1429 ×19 α 3.1429 ×19 α 3.1429 ×19 α 3.1429 ×19 α 1.0 ×19 α 1.0 ×19 α 1.0 ×19 α 1.0 ×19 α 1.0 ×19 α 1.0 ×19 α 1.0 ×19 α 1.0 ×19 α 1.2857 ×19 α 1.2857 ×19 α 1.2857 ×19 α 1.2857 ×19 α 1.2857 ×18 α 1.2857 ×18 α 1.2857 ×18 α 1.2857 ×18 α 1.0 ×18 α 1.0 ×18 α 1.0 ×18 α 1.0 ×18 α 1.0 ×18 α 1.0 ×18 α 1.0 ×18 α 1.0 ×18
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 1.0
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 |
|---|---|---|---|---|
| 25 | 51.5 ± 1.5 | 49.9 ± 1.3 | 49.8 ± 1.9 | 49.6 ± 2.2 |
| 50 | 50.9 ± 1.8 | 50.0 ± 1.2 | 50.2 ± 1.6 | 50.4 ± 1.5 |
| 75 | 54.0 ± 1.9 | 51.6 ± 1.9 | 51.1 ± 1.7 | 50.1 ± 2.2 |
| 100 | 62.4 ± 6.2 | 58.5 ± 4.6 | 54.2 ± 4.3 | 51.8 ± 2.1 |
| 125 | 74.4 ± 8.7 | 70.9 ± 8.4 | 62.6 ± 8.9 | 59.0 ± 7.4 |
| 150 | 78.4 ± 4.5 | 76.4 ± 4.0 | 64.7 ± 6.3 | 61.9 ± 5.5 |
| 175 | 93.0 ± 4.7 | 90.8 ± 5.2 | 86.3 ± 9.5 | 82.0 ± 9.6 |
| 200 | 99.4 ± 0.7 | 97.5 ± 3.7 | 98.3 ± 1.5 | 94.9 ± 5.0 |
| 225 | 99.8 ± 0.3 | 97.4 ± 6.3 | 99.3 ± 1.0 | 95.7 ± 6.9 |
| 250 | 99.8 ± 0.3 | 99.1 ± 1.1 | 99.3 ± 1.1 | 96.9 ± 3.0 |
| 275 | 99.9 ± 0.1 | 99.3 ± 0.8 | 99.5 ± 0.5 | 97.1 ± 2.2 |
| 300 | 99.9 ± 0.2 | 99.7 ± 0.4 | 99.7 ± 0.4 | 97.4 ± 2.3 |
| 325 | 96.0 ± 1.5 | 93.3 ± 2.3 | 89.5 ± 4.4 | 83.5 ± 5.9 |
| 350 | 95.3 ± 1.2 | 92.8 ± 2.0 | 87.0 ± 3.4 | 81.8 ± 4.2 |
| 375 | 94.6 ± 1.2 | 92.0 ± 1.6 | 86.0 ± 2.5 | 81.1 ± 3.3 |
| 400 | 94.1 ± 1.1 | 91.8 ± 1.4 | 83.9 ± 2.7 | 78.9 ± 3.3 |
| 450 | 93.6 ± 1.2 | 91.5 ± 1.3 | 83.1 ± 3.1 | 77.7 ± 3.4 |
| 500 | 99.9 ± 0.1 | 99.4 ± 1.0 | 99.3 ± 0.7 | 96.3 ± 5.2 |
| 550 | 100.0 ± 0.0 | 99.8 ± 0.2 | 99.2 ± 0.6 | 97.3 ± 2.2 |
| 600 | 100.0 ± 0.1 | 99.5 ± 0.4 | 99.3 ± 0.8 | 95.2 ± 4.3 |
| 650 | 99.8 ± 0.2 | 98.8 ± 0.7 | 97.0 ± 2.0 | 91.1 ± 5.1 |
| 700 | 99.9 ± 0.1 | 98.6 ± 1.1 | 97.1 ± 1.5 | 89.2 ± 6.0 |
| 750 | 99.9 ± 0.1 | 99.0 ± 0.6 | 97.9 ± 1.6 | 91.7 ± 4.2 |
| 800 | 100.0 ± 0.1 | 99.2 ± 1.0 | 99.0 ± 1.3 | 92.4 ± 5.1 |
| 850 | 100.0 ± 0.1 | 99.2 ± 0.8 | 99.1 ± 0.9 | 92.1 ± 5.3 |
| 900 | 100.0 ± 0.1 | 99.0 ± 1.1 | 98.8 ± 1.1 | 91.1 ± 5.9 |