r9_c0
done · peak unseen_mismatch 84.7
α schedule
α 5.0 ×25 α 4.6117 ×25 α 4.0038 ×25 α 0.0593 ×25 α 0.0532 ×25 α 2.8857 ×25 α 2.6168 ×25 α 2.3768 ×25 α 0.036 ×25 α 0.0327 ×25 α 0.0296 ×25 α 1.6023 ×25 α 1.4405 ×25 α 0.0215 ×25
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 0.0215
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 ± 1.9 | 50.6 ± 1.5 | 50.0 ± 1.3 | 50.0 ± 1.6 |
| 75 | 53.1 ± 2.5 | 50.5 ± 1.7 | 50.5 ± 1.5 | 49.5 ± 1.5 |
| 100 | 54.3 ± 7.2 | 51.1 ± 1.9 | 52.7 ± 6.2 | 50.1 ± 1.6 |
| 125 | 60.3 ± 15.1 | 48.4 ± 9.3 | 58.4 ± 15.7 | 47.2 ± 9.3 |
| 150 | 73.4 ± 19.2 | 51.4 ± 16.1 | 72.0 ± 19.1 | 49.8 ± 14.9 |
| 175 | 86.4 ± 16.4 | 70.2 ± 16.6 | 85.0 ± 17.4 | 67.8 ± 16.3 |
| 200 | 95.4 ± 8.6 | 85.7 ± 12.7 | 94.1 ± 9.9 | 83.5 ± 13.1 |
| 225 | 99.8 ± 1.0 | 78.7 ± 23.1 | 99.8 ± 0.9 | 78.6 ± 22.7 |
| 250 | 100.0 ± 0.0 | 66.8 ± 25.5 | 100.0 ± 0.1 | 66.4 ± 25.5 |
| 275 | 100.0 ± 0.0 | 58.6 ± 27.0 | 100.0 ± 0.0 | 59.2 ± 26.5 |
| 300 | 100.0 ± 0.0 | 70.6 ± 25.4 | 100.0 ± 0.0 | 70.8 ± 25.1 |
| 325 | 100.0 ± 0.0 | 85.4 ± 19.4 | 100.0 ± 0.1 | 84.7 ± 19.5 |
| 350 | 100.0 ± 0.0 | 74.4 ± 23.1 | 100.0 ± 0.0 | 74.5 ± 22.6 |