free-d56d6dda
done · peak unseen_mismatch 98.0
α schedule
α 4.4 ×10 α 4.9 ×10 α 4.1 ×10 α 5.0 ×10 α 4.5 ×20 α 4.8 ×10 α 4.2 ×10 α 4.7 ×10 α 4.5 ×10 α 5.0 ×10 α 4.9 ×20 α 4.4 ×10 α 4.5 ×10 α 4.3 ×10 α 4.4 ×10 α 1.9 ×10 α 0.2 ×20 α 0.0 ×10 α 0.2 ×10 α 2.7 ×10 α 4.3 ×20 α 4.2 ×10 α 1.6 ×10 α 0.5 ×10 α 0.1 ×10 α 0.7 ×110
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 0.7
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 |
|---|---|---|---|---|
| 100 | 60.1 ± 7.4 | 56.6 ± 6.4 | 53.7 ± 5.4 | 51.5 ± 3.6 |
| 110 | 65.6 ± 8.8 | 61.7 ± 7.7 | 56.5 ± 7.2 | 53.6 ± 5.0 |
| 120 | 70.5 ± 8.9 | 66.9 ± 7.7 | 59.5 ± 8.0 | 56.1 ± 5.4 |
| 130 | 74.6 ± 8.4 | 71.5 ± 7.3 | 61.7 ± 8.7 | 58.5 ± 6.4 |
| 140 | 77.5 ± 6.7 | 74.6 ± 6.0 | 63.2 ± 8.1 | 60.2 ± 6.5 |
| 150 | 79.1 ± 5.4 | 76.6 ± 4.9 | 63.9 ± 7.3 | 61.1 ± 6.2 |
| 160 | 80.2 ± 4.6 | 77.8 ± 4.2 | 64.4 ± 7.0 | 61.5 ± 5.9 |
| 170 | 80.8 ± 3.8 | 78.6 ± 3.5 | 64.4 ± 6.8 | 61.8 ± 5.6 |
| 180 | 83.8 ± 3.9 | 81.6 ± 3.6 | 67.8 ± 7.3 | 64.9 ± 6.4 |
| 190 | 88.9 ± 5.5 | 86.5 ± 5.5 | 78.1 ± 9.9 | 74.5 ± 9.5 |
| 200 | 95.0 ± 5.5 | 92.7 ± 6.4 | 90.8 ± 8.8 | 87.2 ± 9.4 |
| 210 | 98.2 ± 3.0 | 94.3 ± 8.2 | 96.9 ± 4.5 | 92.0 ± 9.2 |
| 220 | 99.3 ± 1.6 | 84.5 ± 18.4 | 98.9 ± 2.2 | 82.5 ± 18.4 |
| 230 | 99.9 ± 0.2 | 87.0 ± 11.3 | 99.6 ± 0.4 | 84.0 ± 12.2 |
| 240 | 99.4 ± 0.7 | 95.9 ± 2.9 | 98.7 ± 1.7 | 92.1 ± 4.8 |
| 250 | 95.2 ± 2.1 | 92.4 ± 1.8 | 88.9 ± 4.7 | 83.9 ± 3.6 |
| 260 | 91.8 ± 1.8 | 89.5 ± 1.1 | 82.6 ± 4.1 | 78.0 ± 2.7 |
| 270 | 96.7 ± 1.2 | 94.4 ± 2.0 | 91.7 ± 1.7 | 87.1 ± 2.8 |
| 280 | 99.9 ± 0.2 | 98.9 ± 1.2 | 99.3 ± 0.9 | 98.0 ± 2.4 |