p06
done · peak unseen_mismatch 84.3
α schedule
α 1.0163 ×25 α 1.0253 ×25 α 1.0467 ×25 α 1.0825 ×25 α 1.1341 ×25 α 1.2026 ×25 α 1.289 ×25 α 1.394 ×25 α 1.5186 ×25 α 1.6632 ×25 α 1.8286 ×25 α 2.0154 ×25 α 2.224 ×25 α 2.4549 ×25 α 2.7087 ×25 α 2.9858 ×25
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 2.9858
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 |
|---|---|---|---|---|
| 25 | 50.7 ± 1.5 | 49.8 ± 1.8 | 49.9 ± 1.8 | 49.7 ± 1.5 |
| 50 | 50.5 ± 1.7 | 50.4 ± 1.6 | 50.1 ± 1.3 | 50.1 ± 1.6 |
| 75 | 50.4 ± 1.6 | 49.4 ± 1.5 | 50.2 ± 1.4 | 49.6 ± 1.7 |
| 100 | 50.3 ± 1.7 | 50.0 ± 1.3 | 49.8 ± 1.3 | 50.0 ± 1.6 |
| 125 | 51.1 ± 1.8 | 49.4 ± 1.6 | 50.8 ± 2.0 | 49.3 ± 1.8 |
| 150 | 52.9 ± 5.7 | 48.8 ± 3.1 | 52.2 ± 5.6 | 48.4 ± 3.2 |
| 175 | 58.7 ± 15.7 | 45.5 ± 9.4 | 58.8 ± 15.7 | 45.6 ± 9.0 |
| 200 | 67.4 ± 22.1 | 44.0 ± 14.2 | 67.5 ± 22.0 | 44.7 ± 14.6 |
| 225 | 69.6 ± 22.9 | 44.1 ± 15.9 | 69.7 ± 22.8 | 44.2 ± 16.3 |
| 250 | 76.5 ± 23.0 | 42.8 ± 20.1 | 76.5 ± 22.8 | 42.5 ± 19.8 |
| 275 | 84.3 ± 20.6 | 50.5 ± 23.1 | 83.5 ± 21.4 | 50.1 ± 22.9 |
| 300 | 89.9 ± 18.4 | 50.1 ± 28.5 | 89.6 ± 18.8 | 50.2 ± 28.3 |
| 325 | 92.3 ± 16.1 | 62.1 ± 26.3 | 91.8 ± 16.9 | 61.5 ± 25.9 |
| 350 | 94.5 ± 14.0 | 69.6 ± 23.3 | 94.3 ± 14.2 | 69.3 ± 23.1 |
| 375 | 96.6 ± 9.9 | 78.5 ± 24.8 | 95.9 ± 11.6 | 77.9 ± 24.2 |
| 400 | 98.5 ± 5.4 | 85.9 ± 25.1 | 97.1 ± 8.7 | 84.3 ± 25.0 |