s0_e2_blk
done · peak unseen_mismatch 97.8
α schedule
α 0.0 ×400 α 1.0 ×400 α 2.0 ×400
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 2.0
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 | 50.4 ± 1.6 | 49.9 ± 1.1 | 49.8 ± 1.9 | 49.6 ± 1.9 |
| 100 | 50.3 ± 1.4 | 50.1 ± 2.2 | 49.8 ± 1.5 | 50.0 ± 1.9 |
| 150 | 52.9 ± 4.9 | 48.8 ± 3.2 | 53.3 ± 5.4 | 47.9 ± 2.6 |
| 200 | 57.3 ± 16.0 | 44.2 ± 13.4 | 57.9 ± 15.6 | 45.0 ± 14.2 |
| 250 | 68.8 ± 21.0 | 44.6 ± 11.9 | 68.0 ± 20.9 | 45.0 ± 11.8 |
| 300 | 80.4 ± 24.0 | 39.6 ± 13.0 | 80.6 ± 23.8 | 40.6 ± 12.5 |
| 350 | 92.7 ± 15.7 | 28.9 ± 16.1 | 92.4 ± 16.4 | 29.4 ± 15.0 |
| 400 | 97.6 ± 7.1 | 18.9 ± 18.1 | 97.5 ± 7.6 | 19.4 ± 18.5 |
| 450 | 100.0 ± 0.0 | 28.8 ± 22.0 | 100.0 ± 0.0 | 29.8 ± 22.1 |
| 500 | 100.0 ± 0.0 | 31.4 ± 27.9 | 100.0 ± 0.0 | 31.2 ± 27.4 |
| 550 | 100.0 ± 0.0 | 35.2 ± 24.6 | 100.0 ± 0.0 | 34.7 ± 24.0 |
| 600 | 100.0 ± 0.0 | 37.9 ± 27.2 | 100.0 ± 0.0 | 38.2 ± 27.6 |
| 650 | 100.0 ± 0.0 | 44.1 ± 19.6 | 100.0 ± 0.0 | 44.0 ± 19.6 |
| 700 | 100.0 ± 0.0 | 41.8 ± 33.0 | 100.0 ± 0.0 | 41.4 ± 33.0 |
| 750 | 100.0 ± 0.0 | 46.0 ± 21.1 | 100.0 ± 0.0 | 45.6 ± 21.5 |
| 800 | 100.0 ± 0.0 | 50.6 ± 28.0 | 100.0 ± 0.0 | 50.4 ± 28.0 |
| 850 | 100.0 ± 0.0 | 58.1 ± 32.2 | 100.0 ± 0.0 | 57.8 ± 32.4 |
| 900 | 100.0 ± 0.1 | 73.6 ± 25.3 | 99.9 ± 0.4 | 73.2 ± 24.8 |
| 950 | 100.0 ± 0.0 | 81.8 ± 26.2 | 100.0 ± 0.0 | 82.2 ± 26.2 |
| 1000 | 99.9 ± 0.1 | 89.6 ± 14.5 | 100.0 ± 0.1 | 89.1 ± 14.0 |
| 1050 | 99.7 ± 0.4 | 91.3 ± 13.8 | 99.0 ± 1.1 | 90.0 ± 13.7 |
| 1100 | 99.8 ± 0.2 | 99.0 ± 1.2 | 99.2 ± 1.1 | 97.8 ± 2.1 |
| 1150 | 99.3 ± 0.7 | 97.9 ± 1.9 | 98.3 ± 2.0 | 96.6 ± 2.2 |
| 1200 | 99.0 ± 1.0 | 97.9 ± 1.2 | 97.4 ± 2.4 | 94.1 ± 3.9 |