r3_c3
done · peak unseen_mismatch 90.7
α schedule
α 1.2326 ×25 α 1.3231 ×25 α 1.4284 ×25 α 1.543 ×25 α 1.6647 ×25 α 1.7922 ×25 α 1.9247 ×25 α 2.0616 ×25 α 2.2023 ×25 α 2.3465 ×25 α 2.4939 ×25 α 2.6442 ×25 α 2.7974 ×25 α 2.953 ×25 α 3.1112 ×25 α 3.2716 ×25
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 3.2716
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 | 50.4 ± 1.4 | 50.1 ± 1.5 | 49.8 ± 1.3 | 49.4 ± 1.9 |
| 50 | 50.5 ± 1.7 | 50.1 ± 1.5 | 50.1 ± 1.5 | 49.9 ± 1.7 |
| 75 | 50.8 ± 1.3 | 49.4 ± 1.5 | 50.2 ± 1.5 | 49.5 ± 1.7 |
| 100 | 50.2 ± 2.0 | 50.1 ± 1.4 | 49.9 ± 1.7 | 50.0 ± 1.3 |
| 125 | 51.5 ± 1.8 | 49.2 ± 1.6 | 50.7 ± 1.8 | 49.3 ± 1.9 |
| 150 | 53.9 ± 7.7 | 48.6 ± 4.1 | 52.7 ± 7.2 | 48.5 ± 4.1 |
| 175 | 60.1 ± 16.3 | 48.3 ± 5.0 | 59.7 ± 15.9 | 48.2 ± 5.1 |
| 200 | 69.5 ± 20.0 | 53.6 ± 11.5 | 69.0 ± 20.2 | 53.2 ± 11.1 |
| 225 | 78.2 ± 20.9 | 63.3 ± 17.6 | 77.8 ± 20.7 | 62.4 ± 16.7 |
| 250 | 85.8 ± 19.8 | 70.9 ± 18.6 | 85.0 ± 19.7 | 69.2 ± 17.7 |
| 275 | 91.5 ± 14.3 | 82.6 ± 17.6 | 90.7 ± 14.5 | 81.3 ± 16.9 |
| 300 | 96.0 ± 10.4 | 88.3 ± 14.2 | 94.9 ± 10.2 | 86.2 ± 14.8 |
| 325 | 96.6 ± 9.5 | 91.8 ± 10.9 | 94.8 ± 10.0 | 89.1 ± 10.0 |
| 350 | 96.5 ± 8.5 | 93.9 ± 8.6 | 94.5 ± 9.7 | 90.7 ± 9.8 |
| 375 | 96.2 ± 7.5 | 94.5 ± 6.9 | 93.1 ± 9.5 | 90.1 ± 8.9 |
| 400 | 96.5 ± 3.9 | 94.7 ± 4.1 | 92.4 ± 6.9 | 90.0 ± 7.0 |