s3_e0_blk
done · peak unseen_mismatch 93.9
α schedule
α 3.0 ×300 α 2.0 ×300 α 1.0 ×300 α 0.0 ×300
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 | 51.3 ± 2.2 | 49.6 ± 0.8 | 50.0 ± 1.7 | 50.3 ± 1.6 |
| 100 | 55.9 ± 7.4 | 53.6 ± 4.6 | 52.7 ± 5.7 | 50.5 ± 3.7 |
| 150 | 76.3 ± 12.5 | 72.0 ± 11.7 | 69.1 ± 13.0 | 63.1 ± 11.1 |
| 200 | 88.2 ± 4.3 | 86.4 ± 4.6 | 78.2 ± 8.2 | 75.1 ± 6.9 |
| 250 | 90.7 ± 1.6 | 89.4 ± 1.6 | 79.0 ± 4.9 | 76.2 ± 3.6 |
| 300 | 91.8 ± 1.5 | 90.3 ± 1.5 | 78.3 ± 4.3 | 75.5 ± 4.1 |
| 350 | 96.2 ± 0.8 | 94.8 ± 1.0 | 86.2 ± 3.1 | 81.8 ± 3.3 |
| 400 | 96.4 ± 1.2 | 94.6 ± 1.4 | 87.2 ± 3.3 | 81.0 ± 4.4 |
| 450 | 97.0 ± 0.8 | 95.8 ± 1.2 | 86.4 ± 4.9 | 81.1 ± 6.2 |
| 500 | 97.3 ± 0.8 | 95.8 ± 1.0 | 88.7 ± 2.8 | 82.5 ± 5.1 |
| 550 | 97.6 ± 0.8 | 96.2 ± 0.7 | 89.4 ± 4.0 | 82.2 ± 5.1 |
| 600 | 98.0 ± 0.5 | 96.5 ± 1.0 | 90.7 ± 2.9 | 83.7 ± 5.0 |
| 650 | 99.9 ± 0.1 | 99.5 ± 0.5 | 98.8 ± 0.7 | 93.9 ± 4.2 |
| 700 | 100.0 ± 0.1 | 99.0 ± 1.5 | 99.2 ± 0.7 | 92.5 ± 5.2 |
| 750 | 100.0 ± 0.0 | 99.3 ± 1.0 | 99.3 ± 0.9 | 92.5 ± 7.0 |
| 800 | 100.0 ± 0.0 | 98.9 ± 1.1 | 99.5 ± 0.3 | 90.2 ± 8.0 |
| 850 | 100.0 ± 0.0 | 99.0 ± 1.2 | 99.2 ± 0.9 | 89.8 ± 8.4 |
| 900 | 100.0 ± 0.0 | 98.3 ± 1.7 | 99.2 ± 1.1 | 85.0 ± 10.2 |
| 950 | 100.0 ± 0.0 | 55.9 ± 33.9 | 100.0 ± 0.0 | 51.3 ± 33.3 |
| 1000 | 100.0 ± 0.0 | 48.6 ± 33.9 | 100.0 ± 0.0 | 46.9 ± 32.9 |
| 1050 | 100.0 ± 0.0 | 18.9 ± 13.6 | 100.0 ± 0.1 | 21.5 ± 13.5 |
| 1100 | 100.0 ± 0.0 | 26.1 ± 17.8 | 100.0 ± 0.1 | 28.8 ± 15.9 |
| 1150 | 100.0 ± 0.0 | 10.9 ± 4.7 | 99.7 ± 0.2 | 17.5 ± 8.2 |
| 1200 | 100.0 ± 0.0 | 11.4 ± 7.2 | 99.7 ± 0.3 | 15.7 ± 10.6 |