v_1.3
done · peak unseen_mismatch 92.5
Referenced by
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 3.0 ×19 α 3.0 ×19 α 1.3 ×19 α 1.3 ×19 α 1.3 ×19 α 1.3 ×19 α 1.3 ×19 α 1.3 ×19 α 1.3 ×19 α 1.3 ×19 α 2.1429 ×19 α 2.1429 ×19 α 2.1429 ×19 α 2.1429 ×19 α 2.1429 ×19 α 2.1429 ×19 α 2.1429 ×19 α 2.1429 ×19 α 1.3 ×19 α 1.3 ×19 α 1.3 ×19 α 1.3 ×19 α 1.3 ×19 α 1.3 ×19 α 1.3 ×19 α 1.3 ×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 1.3 ×18 α 1.3 ×18 α 1.3 ×18 α 1.3 ×18 α 1.3 ×18 α 1.3 ×18 α 1.3 ×18 α 1.3 ×18
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 1.3
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 | 65.8 ± 10.5 | 60.5 ± 8.1 | 59.9 ± 9.2 | 54.9 ± 5.3 |
| 150 | 79.3 ± 12.4 | 73.6 ± 12.8 | 72.2 ± 14.1 | 66.3 ± 12.7 |
| 175 | 91.2 ± 12.5 | 83.1 ± 16.1 | 87.6 ± 14.1 | 77.9 ± 16.2 |
| 200 | 96.7 ± 10.7 | 87.3 ± 17.8 | 95.8 ± 10.4 | 84.5 ± 17.6 |
| 225 | 97.1 ± 11.1 | 90.3 ± 16.4 | 96.2 ± 10.9 | 87.8 ± 15.9 |
| 250 | 97.2 ± 10.8 | 91.1 ± 20.1 | 96.6 ± 10.1 | 87.8 ± 19.3 |
| 275 | 97.3 ± 10.2 | 93.1 ± 14.3 | 96.6 ± 10.8 | 89.6 ± 13.7 |
| 300 | 98.6 ± 5.2 | 94.4 ± 12.3 | 97.3 ± 5.8 | 90.2 ± 12.7 |
| 325 | 98.4 ± 1.1 | 94.6 ± 8.4 | 94.9 ± 4.3 | 87.7 ± 9.9 |
| 350 | 97.8 ± 1.3 | 96.3 ± 2.2 | 93.2 ± 4.9 | 88.2 ± 7.0 |
| 375 | 97.6 ± 1.2 | 95.9 ± 1.8 | 92.6 ± 3.9 | 87.4 ± 5.8 |
| 400 | 97.4 ± 1.2 | 95.6 ± 1.9 | 91.0 ± 4.6 | 85.3 ± 6.6 |
| 450 | 96.8 ± 1.2 | 95.2 ± 1.6 | 90.0 ± 4.0 | 84.4 ± 5.7 |
| 500 | 99.4 ± 0.5 | 98.5 ± 0.9 | 97.2 ± 1.8 | 92.5 ± 5.0 |
| 550 | 99.7 ± 0.2 | 98.5 ± 1.2 | 97.6 ± 1.6 | 92.2 ± 4.3 |
| 600 | 99.7 ± 0.2 | 98.3 ± 1.6 | 97.6 ± 1.0 | 89.5 ± 8.2 |
| 650 | 99.8 ± 0.2 | 98.5 ± 1.0 | 97.3 ± 1.8 | 90.1 ± 6.5 |
| 700 | 99.8 ± 0.3 | 98.5 ± 1.3 | 97.7 ± 1.7 | 88.2 ± 7.5 |
| 750 | 99.9 ± 0.1 | 98.7 ± 1.1 | 97.9 ± 1.5 | 89.0 ± 7.2 |
| 800 | 99.8 ± 0.2 | 98.4 ± 1.4 | 98.0 ± 1.2 | 88.4 ± 7.6 |
| 850 | 99.8 ± 0.2 | 98.5 ± 1.1 | 97.0 ± 1.7 | 87.7 ± 5.3 |
| 900 | 99.8 ± 0.1 | 98.3 ± 1.3 | 97.4 ± 1.5 | 85.8 ± 8.2 |