r9_c2
done · peak unseen_mismatch 79.0
α schedule
α 5.0 ×25 α 4.4678 ×25 α 3.7406 ×25 α 0.664 ×25 α 0.5674 ×25 α 2.3192 ×25 α 1.9626 ×25 α 1.6396 ×25 α 0.2802 ×25 α 0.2228 ×25 α 0.1692 ×25 α 0.5693 ×25 α 0.3406 ×25 α 0.0257 ×25
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 0.0257
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 ± 1.8 | 50.6 ± 1.5 | 50.0 ± 1.3 | 50.0 ± 1.6 |
| 75 | 52.9 ± 2.4 | 50.5 ± 1.7 | 50.5 ± 1.4 | 49.5 ± 1.6 |
| 100 | 54.5 ± 7.3 | 51.3 ± 2.0 | 52.7 ± 6.1 | 50.0 ± 1.9 |
| 125 | 60.7 ± 14.8 | 49.7 ± 3.2 | 58.0 ± 15.6 | 49.2 ± 3.6 |
| 150 | 74.4 ± 19.1 | 57.6 ± 14.2 | 72.4 ± 19.7 | 54.9 ± 12.8 |
| 175 | 87.1 ± 16.8 | 71.3 ± 19.0 | 85.7 ± 17.7 | 68.8 ± 17.9 |
| 200 | 96.3 ± 9.0 | 80.4 ± 16.4 | 95.9 ± 9.0 | 79.0 ± 16.3 |
| 225 | 99.9 ± 0.2 | 74.0 ± 21.6 | 100.0 ± 0.1 | 73.5 ± 21.0 |
| 250 | 100.0 ± 0.1 | 57.3 ± 28.7 | 100.0 ± 0.1 | 57.3 ± 28.2 |
| 275 | 100.0 ± 0.0 | 54.4 ± 29.6 | 100.0 ± 0.0 | 54.2 ± 29.5 |
| 300 | 100.0 ± 0.0 | 48.8 ± 25.8 | 100.0 ± 0.0 | 48.9 ± 25.4 |
| 325 | 100.0 ± 0.0 | 50.5 ± 30.3 | 100.0 ± 0.0 | 50.7 ± 29.5 |
| 350 | 100.0 ± 0.0 | 44.3 ± 30.4 | 100.0 ± 0.0 | 44.1 ± 29.3 |