peak_40
done · peak unseen_mismatch 97.8
α schedule
α 4.0 ×150 α 1.0 ×150 α 2.8571 ×150 α 1.0 ×150 α 1.7143 ×150 α 1.0 ×150
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 |
|---|---|---|---|---|
| 50 | 51.4 ± 1.3 | 50.0 ± 1.7 | 49.6 ± 1.3 | 50.1 ± 1.2 |
| 100 | 56.2 ± 4.6 | 51.0 ± 2.3 | 53.7 ± 3.8 | 50.2 ± 1.5 |
| 150 | 88.6 ± 7.7 | 78.3 ± 8.7 | 84.0 ± 10.4 | 72.3 ± 9.8 |
| 200 | 100.0 ± 0.0 | 73.0 ± 20.9 | 100.0 ± 0.0 | 71.9 ± 19.6 |
| 250 | 100.0 ± 0.0 | 73.4 ± 24.3 | 100.0 ± 0.0 | 73.0 ± 24.1 |
| 300 | 100.0 ± 0.1 | 80.9 ± 24.0 | 100.0 ± 0.1 | 79.6 ± 24.8 |
| 350 | 99.0 ± 1.6 | 92.2 ± 16.2 | 97.2 ± 4.8 | 88.0 ± 17.8 |
| 400 | 98.2 ± 2.1 | 96.3 ± 3.5 | 95.1 ± 6.2 | 90.5 ± 9.2 |
| 450 | 97.6 ± 2.4 | 95.7 ± 2.5 | 93.5 ± 6.1 | 88.8 ± 7.9 |
| 500 | 100.0 ± 0.0 | 99.5 ± 0.8 | 99.8 ± 0.2 | 97.8 ± 2.5 |
| 550 | 100.0 ± 0.0 | 99.3 ± 0.7 | 99.6 ± 1.1 | 97.0 ± 2.7 |
| 600 | 100.0 ± 0.0 | 97.9 ± 2.8 | 99.9 ± 0.4 | 94.1 ± 6.4 |
| 650 | 99.4 ± 0.6 | 97.2 ± 1.7 | 96.5 ± 3.2 | 89.3 ± 7.8 |
| 700 | 99.1 ± 0.6 | 97.0 ± 1.4 | 94.7 ± 3.2 | 88.0 ± 4.6 |
| 750 | 99.1 ± 0.5 | 96.7 ± 1.6 | 94.8 ± 2.2 | 86.6 ± 5.6 |
| 800 | 100.0 ± 0.0 | 99.0 ± 1.0 | 99.3 ± 0.6 | 93.2 ± 6.4 |
| 850 | 99.9 ± 0.2 | 97.3 ± 4.3 | 98.8 ± 1.9 | 89.4 ± 11.4 |
| 900 | 100.0 ± 0.0 | 97.5 ± 2.8 | 99.1 ± 1.5 | 87.3 ± 14.8 |