s1_e1
done · peak unseen_mismatch 89.5
α schedule
α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50 α 1.0 ×50
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 | 50.2 ± 1.0 | 49.1 ± 1.5 | 49.8 ± 1.6 | 51.2 ± 1.6 |
| 100 | 50.8 ± 2.6 | 49.7 ± 1.7 | 51.3 ± 1.9 | 48.7 ± 2.4 |
| 150 | 56.6 ± 14.6 | 48.7 ± 1.4 | 55.0 ± 15.0 | 50.6 ± 1.6 |
| 200 | 71.1 ± 19.7 | 39.3 ± 10.7 | 69.6 ± 19.9 | 40.4 ± 11.2 |
| 250 | 90.7 ± 18.5 | 27.3 ± 16.4 | 90.4 ± 19.0 | 26.9 ± 15.4 |
| 300 | 99.6 ± 0.7 | 28.0 ± 20.8 | 99.5 ± 0.8 | 28.2 ± 20.6 |
| 350 | 100.0 ± 0.0 | 17.0 ± 17.0 | 100.0 ± 0.0 | 17.1 ± 17.7 |
| 400 | 100.0 ± 0.0 | 21.9 ± 21.9 | 100.0 ± 0.0 | 22.3 ± 22.6 |
| 450 | 100.0 ± 0.0 | 29.9 ± 17.6 | 100.0 ± 0.0 | 30.5 ± 18.3 |
| 500 | 100.0 ± 0.0 | 27.9 ± 17.2 | 100.0 ± 0.0 | 27.7 ± 17.6 |
| 550 | 100.0 ± 0.0 | 46.4 ± 18.8 | 100.0 ± 0.0 | 47.0 ± 19.0 |
| 600 | 100.0 ± 0.0 | 40.6 ± 27.2 | 100.0 ± 0.0 | 41.1 ± 26.8 |
| 650 | 100.0 ± 0.0 | 57.2 ± 27.0 | 100.0 ± 0.0 | 56.8 ± 26.8 |
| 700 | 100.0 ± 0.0 | 41.5 ± 22.3 | 100.0 ± 0.0 | 41.9 ± 21.7 |
| 750 | 100.0 ± 0.0 | 50.7 ± 28.9 | 100.0 ± 0.0 | 51.0 ± 28.9 |
| 800 | 100.0 ± 0.0 | 57.3 ± 26.0 | 100.0 ± 0.0 | 56.8 ± 26.3 |
| 850 | 100.0 ± 0.0 | 68.6 ± 20.2 | 100.0 ± 0.0 | 69.3 ± 20.5 |
| 900 | 100.0 ± 0.0 | 68.1 ± 25.9 | 100.0 ± 0.0 | 67.9 ± 26.5 |
| 950 | 100.0 ± 0.0 | 70.5 ± 23.2 | 100.0 ± 0.0 | 70.6 ± 23.4 |
| 1000 | 100.0 ± 0.0 | 76.0 ± 19.5 | 100.0 ± 0.0 | 76.0 ± 19.3 |
| 1050 | 100.0 ± 0.0 | 76.3 ± 28.6 | 100.0 ± 0.0 | 76.0 ± 28.9 |
| 1100 | 100.0 ± 0.0 | 84.8 ± 19.5 | 100.0 ± 0.0 | 84.2 ± 19.6 |
| 1150 | 100.0 ± 0.0 | 90.5 ± 15.5 | 100.0 ± 0.0 | 89.5 ± 16.4 |
| 1200 | 100.0 ± 0.0 | 87.8 ± 24.9 | 100.0 ± 0.0 | 86.9 ± 25.0 |