cand3
done · peak unseen_mismatch 97.2
Referenced by
α schedule
α 3.2007 ×25 α 3.1661 ×25 α 3.1038 ×25 α 3.016 ×25 α 2.904 ×25 α 2.7689 ×25 α 2.6113 ×25 α 2.4318 ×25 α 2.2309 ×25 α 2.009 ×25 α 1.7666 ×25 α 1.5038 ×25 α 1.221 ×25 α 0.9185 ×25 α 0.5965 ×25 α 0.2551 ×25
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 0.2551
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 | 50.9 ± 1.6 | 49.9 ± 1.7 | 50.0 ± 1.8 | 49.9 ± 1.6 |
| 50 | 51.7 ± 1.8 | 50.2 ± 1.6 | 50.3 ± 1.7 | 50.0 ± 1.7 |
| 75 | 52.4 ± 2.7 | 50.4 ± 2.1 | 50.8 ± 2.2 | 49.9 ± 1.7 |
| 100 | 58.1 ± 7.7 | 53.7 ± 5.4 | 54.1 ± 6.5 | 51.2 ± 3.9 |
| 125 | 71.0 ± 12.2 | 65.1 ± 10.7 | 63.1 ± 12.1 | 58.0 ± 8.8 |
| 150 | 82.5 ± 10.9 | 77.7 ± 10.2 | 72.8 ± 12.6 | 67.2 ± 10.1 |
| 175 | 88.6 ± 7.7 | 85.1 ± 7.5 | 78.3 ± 10.7 | 73.8 ± 9.8 |
| 200 | 91.5 ± 5.2 | 89.0 ± 5.2 | 81.1 ± 8.7 | 77.0 ± 8.2 |
| 225 | 93.6 ± 3.5 | 91.4 ± 3.6 | 83.1 ± 7.5 | 79.3 ± 6.9 |
| 250 | 95.4 ± 2.5 | 93.5 ± 2.5 | 86.0 ± 6.1 | 82.1 ± 6.0 |
| 275 | 97.1 ± 1.4 | 95.5 ± 1.6 | 89.4 ± 4.3 | 85.4 ± 4.7 |
| 300 | 98.4 ± 0.8 | 97.2 ± 1.2 | 92.7 ± 3.5 | 88.7 ± 4.5 |
| 325 | 99.6 ± 0.3 | 98.8 ± 0.9 | 97.0 ± 2.1 | 93.4 ± 3.7 |
| 350 | 99.9 ± 0.1 | 99.6 ± 0.7 | 99.4 ± 0.7 | 97.0 ± 3.6 |
| 375 | 100.0 ± 0.0 | 99.1 ± 3.4 | 99.9 ± 0.2 | 97.2 ± 5.9 |
| 400 | 100.0 ± 0.0 | 92.8 ± 14.2 | 100.0 ± 0.1 | 89.6 ± 16.4 |