r1_c1
done · peak unseen_mismatch 90.5
α schedule
α 1.5563 ×25 α 1.5564 ×25 α 1.5573 ×25 α 1.5601 ×25 α 1.5664 ×25 α 1.578 ×25 α 1.5975 ×25 α 1.6275 ×25 α 1.6712 ×25 α 1.7321 ×25 α 1.8141 ×25 α 1.9214 ×25 α 2.0584 ×25 α 2.2302 ×25 α 2.442 ×25 α 2.6992 ×25
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 2.6992
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 | 50.5 ± 1.6 | 50.1 ± 1.5 | 49.8 ± 1.7 | 49.5 ± 2.1 |
| 50 | 50.6 ± 1.7 | 50.2 ± 1.6 | 50.2 ± 1.5 | 49.5 ± 1.5 |
| 75 | 50.4 ± 1.5 | 49.6 ± 1.7 | 50.3 ± 1.3 | 49.5 ± 1.6 |
| 100 | 50.6 ± 2.5 | 50.1 ± 1.5 | 50.5 ± 1.7 | 49.8 ± 1.2 |
| 125 | 52.0 ± 3.1 | 49.2 ± 2.0 | 51.0 ± 3.3 | 48.8 ± 2.4 |
| 150 | 56.9 ± 14.6 | 49.0 ± 9.2 | 56.0 ± 14.5 | 48.2 ± 9.2 |
| 175 | 63.3 ± 17.8 | 47.7 ± 8.9 | 62.9 ± 17.8 | 47.4 ± 8.9 |
| 200 | 74.5 ± 22.1 | 51.9 ± 13.6 | 74.3 ± 22.3 | 51.8 ± 13.4 |
| 225 | 79.7 ± 22.8 | 55.7 ± 21.2 | 79.3 ± 23.1 | 55.5 ± 20.7 |
| 250 | 85.1 ± 20.6 | 59.7 ± 25.5 | 84.5 ± 21.0 | 58.9 ± 25.4 |
| 275 | 90.9 ± 16.7 | 72.1 ± 22.0 | 90.3 ± 17.1 | 71.4 ± 21.7 |
| 300 | 95.1 ± 13.2 | 77.9 ± 24.6 | 94.8 ± 13.2 | 77.7 ± 23.9 |
| 325 | 98.1 ± 5.3 | 81.8 ± 22.8 | 97.2 ± 6.3 | 80.4 ± 22.8 |
| 350 | 99.5 ± 0.8 | 87.2 ± 19.0 | 98.9 ± 1.7 | 85.9 ± 19.1 |
| 375 | 99.1 ± 1.2 | 90.3 ± 16.6 | 97.9 ± 2.5 | 88.6 ± 15.4 |
| 400 | 98.8 ± 1.5 | 92.6 ± 13.9 | 97.4 ± 2.8 | 90.5 ± 13.2 |