peak_20
done · peak unseen_mismatch 55.5
α schedule
α 2.0 ×150 α 1.0 ×150 α 1.4286 ×150 α 1.0 ×150 α 0.8571 ×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 | 51.7 ± 1.4 | 50.0 ± 1.5 | 50.3 ± 1.8 | 50.1 ± 1.2 |
| 100 | 52.0 ± 2.2 | 50.0 ± 2.1 | 51.9 ± 2.9 | 49.9 ± 2.2 |
| 150 | 68.9 ± 20.0 | 52.2 ± 7.8 | 67.7 ± 20.5 | 50.3 ± 7.0 |
| 200 | 89.0 ± 18.1 | 35.6 ± 18.1 | 88.4 ± 19.2 | 36.2 ± 17.7 |
| 250 | 95.4 ± 13.9 | 36.0 ± 23.3 | 95.5 ± 13.3 | 36.0 ± 22.7 |
| 300 | 100.0 ± 0.0 | 30.6 ± 28.9 | 100.0 ± 0.0 | 30.8 ± 28.3 |
| 350 | 100.0 ± 0.0 | 27.7 ± 23.2 | 100.0 ± 0.0 | 28.5 ± 23.6 |
| 400 | 100.0 ± 0.0 | 37.2 ± 28.0 | 100.0 ± 0.0 | 37.9 ± 27.8 |
| 450 | 100.0 ± 0.0 | 47.2 ± 28.5 | 100.0 ± 0.0 | 48.0 ± 28.5 |
| 500 | 100.0 ± 0.0 | 53.1 ± 30.8 | 100.0 ± 0.0 | 53.9 ± 30.6 |
| 550 | 100.0 ± 0.0 | 47.1 ± 29.4 | 100.0 ± 0.0 | 47.3 ± 28.9 |
| 600 | 100.0 ± 0.0 | 53.2 ± 35.1 | 100.0 ± 0.0 | 53.6 ± 35.1 |
| 650 | 100.0 ± 0.0 | 50.3 ± 32.1 | 100.0 ± 0.0 | 50.5 ± 32.0 |
| 700 | 100.0 ± 0.0 | 44.3 ± 32.6 | 100.0 ± 0.0 | 45.1 ± 32.3 |
| 750 | 100.0 ± 0.0 | 51.5 ± 29.1 | 100.0 ± 0.0 | 51.9 ± 29.1 |
| 800 | 100.0 ± 0.0 | 55.3 ± 31.7 | 100.0 ± 0.0 | 55.5 ± 31.4 |
| 850 | 100.0 ± 0.0 | 54.4 ± 28.2 | 100.0 ± 0.0 | 54.1 ± 27.9 |
| 900 | 100.0 ± 0.0 | 54.1 ± 28.3 | 100.0 ± 0.0 | 53.5 ± 27.6 |