s0_e3_blk
done · peak unseen_mismatch 93.7
α schedule
α 0.0 ×300 α 1.0 ×300 α 2.0 ×300 α 3.0 ×300
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 3.0
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 | 90.5 ± 18.9 | 30.9 ± 24.4 | 90.4 ± 19.0 | 31.3 ± 25.3 |
| 400 | 97.0 ± 8.9 | 35.2 ± 28.6 | 97.1 ± 8.7 | 35.9 ± 28.6 |
| 450 | 100.0 ± 0.0 | 27.8 ± 31.8 | 100.0 ± 0.0 | 28.1 ± 32.0 |
| 500 | 100.0 ± 0.0 | 29.3 ± 31.0 | 100.0 ± 0.0 | 29.6 ± 31.5 |
| 550 | 100.0 ± 0.0 | 26.2 ± 26.4 | 100.0 ± 0.0 | 26.3 ± 26.7 |
| 600 | 100.0 ± 0.0 | 40.7 ± 23.5 | 100.0 ± 0.0 | 40.6 ± 22.9 |
| 650 | 100.0 ± 0.0 | 49.0 ± 33.1 | 100.0 ± 0.0 | 48.8 ± 33.1 |
| 700 | 100.0 ± 0.0 | 53.5 ± 27.3 | 100.0 ± 0.0 | 53.7 ± 27.1 |
| 750 | 100.0 ± 0.0 | 63.7 ± 24.7 | 100.0 ± 0.0 | 63.7 ± 24.7 |
| 800 | 100.0 ± 0.0 | 81.1 ± 22.4 | 100.0 ± 0.0 | 80.6 ± 22.9 |
| 850 | 99.9 ± 0.1 | 90.7 ± 14.1 | 99.9 ± 0.1 | 89.3 ± 14.4 |
| 900 | 99.8 ± 0.3 | 95.5 ± 7.3 | 99.0 ± 1.9 | 93.7 ± 8.5 |
| 950 | 98.8 ± 1.0 | 93.7 ± 12.1 | 97.4 ± 3.1 | 91.4 ± 12.1 |
| 1000 | 97.7 ± 1.1 | 96.8 ± 1.8 | 94.3 ± 3.1 | 92.4 ± 4.1 |
| 1050 | 96.4 ± 1.4 | 95.0 ± 1.7 | 91.4 ± 3.8 | 88.9 ± 4.4 |
| 1100 | 95.8 ± 1.6 | 94.8 ± 1.9 | 89.8 ± 4.2 | 87.1 ± 5.3 |
| 1150 | 94.6 ± 1.4 | 93.4 ± 1.4 | 86.0 ± 2.3 | 84.1 ± 4.1 |
| 1200 | 93.6 ± 1.3 | 92.2 ± 1.3 | 84.3 ± 3.0 | 81.4 ± 4.6 |