s4_e0_blk
done · peak unseen_mismatch 98.2
α schedule
α 4.0 ×240 α 3.0 ×240 α 2.0 ×240 α 1.0 ×240 α 0.0 ×240
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 0.0
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 ± 1.9 | 49.6 ± 1.0 | 49.9 ± 2.1 | 50.5 ± 1.2 |
| 100 | 58.7 ± 7.3 | 56.1 ± 6.1 | 52.6 ± 4.2 | 51.5 ± 2.9 |
| 150 | 76.9 ± 6.3 | 75.5 ± 5.2 | 65.1 ± 6.5 | 61.0 ± 5.8 |
| 200 | 82.0 ± 2.5 | 80.4 ± 1.7 | 64.6 ± 3.8 | 63.1 ± 3.1 |
| 250 | 84.5 ± 1.9 | 83.2 ± 2.2 | 65.7 ± 4.3 | 63.5 ± 3.4 |
| 300 | 89.2 ± 1.6 | 87.7 ± 1.2 | 71.1 ± 5.2 | 69.3 ± 4.4 |
| 350 | 91.3 ± 1.5 | 89.7 ± 2.0 | 75.7 ± 4.3 | 73.4 ± 4.9 |
| 400 | 91.6 ± 1.4 | 90.1 ± 1.6 | 74.8 ± 4.6 | 70.3 ± 3.8 |
| 450 | 91.7 ± 1.1 | 89.8 ± 1.5 | 75.7 ± 4.2 | 70.6 ± 4.3 |
| 500 | 96.8 ± 1.1 | 95.5 ± 1.1 | 87.2 ± 2.7 | 82.4 ± 2.3 |
| 550 | 96.6 ± 0.8 | 95.1 ± 1.4 | 84.9 ± 3.4 | 78.3 ± 5.6 |
| 600 | 97.1 ± 0.8 | 95.6 ± 0.9 | 87.4 ± 2.9 | 81.6 ± 3.3 |
| 650 | 97.6 ± 0.5 | 96.1 ± 1.2 | 89.2 ± 1.7 | 82.7 ± 4.5 |
| 700 | 97.8 ± 1.0 | 96.2 ± 1.5 | 88.7 ± 4.3 | 82.5 ± 6.3 |
| 750 | 100.0 ± 0.0 | 99.8 ± 0.3 | 99.8 ± 0.4 | 98.2 ± 2.8 |
| 800 | 100.0 ± 0.0 | 99.5 ± 0.7 | 99.2 ± 0.6 | 94.5 ± 4.2 |
| 850 | 100.0 ± 0.1 | 99.6 ± 0.6 | 99.0 ± 1.2 | 93.3 ± 5.6 |
| 900 | 100.0 ± 0.0 | 99.6 ± 0.6 | 99.2 ± 0.7 | 93.7 ± 6.0 |
| 950 | 100.0 ± 0.0 | 99.5 ± 0.5 | 98.6 ± 1.4 | 92.6 ± 6.7 |
| 1000 | 100.0 ± 0.0 | 68.0 ± 32.9 | 100.0 ± 0.0 | 65.8 ± 33.2 |
| 1050 | 100.0 ± 0.0 | 47.0 ± 29.1 | 100.0 ± 0.0 | 45.6 ± 29.4 |
| 1100 | 100.0 ± 0.0 | 30.3 ± 34.8 | 100.0 ± 0.0 | 29.5 ± 34.2 |
| 1150 | 100.0 ± 0.0 | 22.6 ± 17.9 | 99.4 ± 1.0 | 27.8 ± 17.4 |
| 1200 | 100.0 ± 0.0 | 15.3 ± 16.9 | 99.8 ± 0.3 | 18.7 ± 16.2 |