teeth_4
done · peak unseen_mismatch 98.0
α schedule
α 3.0 ×19 α 3.0 ×19 α 3.0 ×19 α 3.0 ×19 α 3.0 ×19 α 3.0 ×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 2.4286 ×19 α 2.4286 ×19 α 2.4286 ×19 α 2.4286 ×19 α 2.4286 ×19 α 2.4286 ×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 1.8571 ×19 α 1.8571 ×19 α 1.8571 ×19 α 1.8571 ×19 α 1.8571 ×19 α 1.8571 ×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 ×18 α 1.2857 ×18 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
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 | 51.4 ± 1.5 | 49.5 ± 1.6 | 49.8 ± 1.9 | 49.7 ± 2.1 |
| 50 | 50.7 ± 1.9 | 49.6 ± 1.4 | 50.1 ± 1.5 | 50.8 ± 1.1 |
| 75 | 51.9 ± 1.8 | 50.1 ± 1.9 | 50.5 ± 1.7 | 50.0 ± 1.9 |
| 100 | 53.8 ± 2.9 | 52.1 ± 2.9 | 51.5 ± 1.8 | 50.0 ± 1.5 |
| 125 | 63.7 ± 9.8 | 56.9 ± 6.4 | 59.4 ± 9.4 | 53.0 ± 4.1 |
| 150 | 80.3 ± 15.8 | 63.2 ± 16.9 | 77.9 ± 17.1 | 61.7 ± 16.0 |
| 175 | 93.6 ± 13.0 | 63.7 ± 24.2 | 92.9 ± 14.1 | 62.5 ± 23.9 |
| 200 | 97.3 ± 11.3 | 60.0 ± 31.0 | 97.5 ± 10.3 | 58.8 ± 30.4 |
| 225 | 97.4 ± 11.4 | 64.6 ± 30.9 | 97.5 ± 11.0 | 64.5 ± 30.7 |
| 250 | 97.4 ± 11.1 | 74.0 ± 32.8 | 97.5 ± 10.4 | 73.3 ± 32.0 |
| 275 | 97.4 ± 10.3 | 82.9 ± 24.4 | 96.9 ± 10.8 | 81.6 ± 24.1 |
| 300 | 96.9 ± 11.1 | 90.4 ± 15.2 | 96.0 ± 11.1 | 87.9 ± 15.3 |
| 325 | 96.9 ± 9.3 | 93.3 ± 11.1 | 95.5 ± 10.0 | 89.8 ± 12.1 |
| 350 | 98.9 ± 3.8 | 94.8 ± 10.1 | 98.4 ± 4.0 | 93.3 ± 10.6 |
| 375 | 100.0 ± 0.1 | 96.7 ± 6.9 | 99.8 ± 0.4 | 95.2 ± 8.4 |
| 400 | 100.0 ± 0.0 | 94.0 ± 15.8 | 99.8 ± 0.5 | 92.8 ± 15.5 |
| 450 | 100.0 ± 0.0 | 96.0 ± 11.2 | 99.8 ± 0.3 | 93.6 ± 11.6 |
| 500 | 99.2 ± 0.6 | 97.9 ± 1.7 | 96.6 ± 2.8 | 92.5 ± 5.7 |
| 550 | 98.5 ± 0.8 | 97.3 ± 1.4 | 94.6 ± 3.4 | 90.5 ± 5.2 |
| 600 | 100.0 ± 0.1 | 99.7 ± 0.7 | 99.7 ± 0.4 | 98.0 ± 2.9 |
| 650 | 100.0 ± 0.1 | 99.3 ± 1.7 | 99.4 ± 0.7 | 96.7 ± 2.9 |
| 700 | 99.8 ± 0.2 | 98.9 ± 0.8 | 97.2 ± 2.0 | 91.6 ± 6.0 |
| 750 | 99.8 ± 0.2 | 98.8 ± 0.9 | 97.7 ± 1.2 | 91.7 ± 5.2 |
| 800 | 99.9 ± 0.1 | 99.2 ± 1.0 | 98.4 ± 1.4 | 93.2 ± 5.2 |
| 850 | 100.0 ± 0.1 | 99.3 ± 1.2 | 99.5 ± 0.7 | 95.3 ± 5.4 |
| 900 | 100.0 ± 0.1 | 99.5 ± 0.8 | 99.2 ± 0.6 | 94.8 ± 5.6 |