r7_c0
done · peak unseen_mismatch 99.2
α schedule
α 5.0 ×25 α 5.0 ×25 α 5.0 ×25 α 0.3692 ×25 α 0.3718 ×25 α 5.0 ×25 α 5.0 ×25 α 5.0 ×25 α 0.3808 ×25 α 0.3829 ×25 α 0.3849 ×25 α 5.0 ×25 α 5.0 ×25 α 0.3907 ×25
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 0.3907
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 | 54.3 ± 7.8 | 51.3 ± 2.5 | 52.7 ± 6.9 | 49.9 ± 1.8 |
| 125 | 61.2 ± 15.1 | 49.0 ± 4.1 | 58.9 ± 15.8 | 48.2 ± 4.4 |
| 150 | 75.8 ± 18.6 | 61.9 ± 12.6 | 73.8 ± 19.0 | 59.2 ± 11.7 |
| 175 | 87.1 ± 12.1 | 79.3 ± 12.2 | 83.6 ± 14.1 | 75.2 ± 13.1 |
| 200 | 91.8 ± 6.2 | 88.3 ± 6.7 | 87.1 ± 9.1 | 82.6 ± 9.3 |
| 225 | 99.5 ± 1.0 | 93.0 ± 6.6 | 99.4 ± 1.1 | 92.0 ± 6.9 |
| 250 | 99.9 ± 0.3 | 90.1 ± 13.4 | 99.9 ± 0.4 | 89.2 ± 14.2 |
| 275 | 100.0 ± 0.0 | 84.8 ± 20.5 | 100.0 ± 0.0 | 85.0 ± 19.8 |
| 300 | 99.5 ± 2.0 | 95.2 ± 9.2 | 99.1 ± 3.3 | 94.1 ± 10.1 |
| 325 | 97.9 ± 2.5 | 94.1 ± 9.1 | 95.4 ± 4.5 | 90.3 ± 10.3 |
| 350 | 100.0 ± 0.0 | 99.3 ± 2.4 | 100.0 ± 0.0 | 99.2 ± 2.5 |