r6_c7
done · peak unseen_mismatch 96.5
α schedule
α 5.0 ×25 α 5.0 ×25 α 5.0 ×25 α 0.2387 ×25 α 0.2334 ×25 α 5.0 ×25 α 5.0 ×25 α 5.0 ×25 α 0.2168 ×25 α 0.2134 ×25 α 0.2102 ×25 α 5.0 ×25 α 5.0 ×25 α 0.2015 ×25
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 0.2015
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 | 54.6 ± 7.3 | 51.1 ± 2.1 | 52.7 ± 6.3 | 49.8 ± 1.9 |
| 125 | 61.1 ± 15.2 | 49.1 ± 7.6 | 59.1 ± 15.7 | 48.0 ± 7.9 |
| 150 | 75.2 ± 18.9 | 60.6 ± 13.5 | 73.1 ± 19.3 | 58.4 ± 12.4 |
| 175 | 86.3 ± 13.5 | 77.1 ± 12.9 | 83.0 ± 14.6 | 73.3 ± 13.1 |
| 200 | 90.8 ± 9.5 | 87.2 ± 10.0 | 86.4 ± 10.8 | 81.9 ± 11.0 |
| 225 | 98.3 ± 6.7 | 89.8 ± 13.7 | 98.2 ± 7.1 | 89.7 ± 13.0 |
| 250 | 100.0 ± 0.1 | 80.4 ± 24.3 | 100.0 ± 0.1 | 80.0 ± 24.6 |
| 275 | 100.0 ± 0.0 | 72.2 ± 32.2 | 100.0 ± 0.0 | 72.0 ± 32.2 |
| 300 | 99.6 ± 1.7 | 93.3 ± 11.2 | 99.4 ± 2.9 | 92.5 ± 11.5 |
| 325 | 98.1 ± 2.7 | 94.1 ± 6.2 | 96.0 ± 4.5 | 90.3 ± 8.3 |
| 350 | 100.0 ± 0.0 | 96.7 ± 8.6 | 100.0 ± 0.0 | 96.5 ± 8.7 |