r7_c5
done · peak unseen_mismatch 98.3
α schedule
α 5.0 ×25 α 5.0 ×25 α 5.0 ×25 α 0.0695 ×25 α 0.0701 ×25 α 5.0 ×25 α 5.0 ×25 α 5.0 ×25 α 0.0741 ×25 α 0.0755 ×25 α 0.077 ×25 α 5.0 ×25 α 5.0 ×25 α 0.0823 ×25
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 0.0823
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 ± 1.3 | 50.5 ± 1.6 | 49.9 ± 1.4 | 49.7 ± 2.0 |
| 50 | 51.5 ± 2.0 | 50.5 ± 1.5 | 50.0 ± 1.3 | 50.0 ± 1.6 |
| 75 | 53.4 ± 2.6 | 50.7 ± 1.7 | 50.4 ± 1.3 | 49.6 ± 1.5 |
| 100 | 54.5 ± 7.9 | 51.1 ± 2.0 | 53.0 ± 7.0 | 50.1 ± 2.4 |
| 125 | 61.2 ± 14.9 | 47.9 ± 7.8 | 59.0 ± 15.6 | 46.9 ± 8.2 |
| 150 | 75.2 ± 19.3 | 55.9 ± 14.0 | 73.6 ± 19.9 | 53.4 ± 13.3 |
| 175 | 86.5 ± 13.5 | 77.3 ± 13.1 | 84.1 ± 14.4 | 74.1 ± 13.1 |
| 200 | 91.8 ± 8.1 | 88.1 ± 8.7 | 87.7 ± 9.9 | 83.5 ± 10.5 |
| 225 | 99.4 ± 1.1 | 89.9 ± 15.0 | 99.3 ± 1.3 | 89.2 ± 14.9 |
| 250 | 100.0 ± 0.1 | 86.7 ± 18.1 | 100.0 ± 0.1 | 86.3 ± 18.2 |
| 275 | 100.0 ± 0.0 | 79.4 ± 23.1 | 100.0 ± 0.0 | 79.3 ± 23.2 |
| 300 | 99.8 ± 0.8 | 94.4 ± 13.7 | 99.6 ± 2.0 | 93.7 ± 14.2 |
| 325 | 98.6 ± 2.1 | 95.0 ± 8.7 | 97.1 ± 4.2 | 92.2 ± 10.1 |
| 350 | 100.0 ± 0.0 | 98.5 ± 4.9 | 100.0 ± 0.0 | 98.3 ± 5.5 |