free-6d62b591
done · peak unseen_mismatch 94.0
α schedule
α 3.848 ×15 α 3.258 ×56 α 1.882 ×19 α 4.377 ×73 α 0.1827 ×12 α 1.212 ×47 α 1.359 ×48 α 2.489 ×30 α 2.489 ×100
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 2.489
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 |
|---|---|---|---|---|
| 100 | 56.0 ± 6.0 | 52.0 ± 2.9 | 52.6 ± 5.3 | 50.2 ± 2.0 |
| 110 | 59.9 ± 9.0 | 55.4 ± 6.2 | 55.0 ± 8.2 | 51.2 ± 4.1 |
| 120 | 64.7 ± 10.6 | 59.2 ± 8.4 | 58.5 ± 10.3 | 53.2 ± 6.4 |
| 130 | 69.9 ± 11.7 | 65.1 ± 10.0 | 61.5 ± 11.8 | 56.9 ± 8.8 |
| 140 | 74.7 ± 11.0 | 70.1 ± 9.5 | 65.3 ± 11.2 | 60.5 ± 8.5 |
| 150 | 79.1 ± 9.6 | 75.0 ± 9.0 | 68.7 ± 10.4 | 63.7 ± 8.7 |
| 160 | 81.2 ± 8.0 | 77.4 ± 7.6 | 69.4 ± 9.6 | 65.4 ± 7.9 |
| 170 | 83.2 ± 8.6 | 79.7 ± 8.4 | 73.3 ± 10.8 | 68.9 ± 9.1 |
| 180 | 90.0 ± 10.4 | 85.9 ± 10.5 | 85.3 ± 12.1 | 79.9 ± 11.6 |
| 190 | 94.3 ± 9.1 | 90.0 ± 10.0 | 91.8 ± 11.4 | 86.1 ± 11.4 |
| 200 | 95.4 ± 8.0 | 90.7 ± 9.1 | 93.0 ± 9.9 | 86.0 ± 10.8 |
| 210 | 95.6 ± 7.0 | 87.3 ± 9.8 | 93.5 ± 8.0 | 82.8 ± 9.8 |
| 220 | 97.5 ± 4.1 | 87.8 ± 12.3 | 95.9 ± 6.1 | 83.7 ± 11.0 |
| 230 | 98.5 ± 2.1 | 88.4 ± 12.9 | 97.2 ± 3.1 | 84.1 ± 12.5 |
| 240 | 99.2 ± 0.9 | 83.6 ± 15.7 | 98.6 ± 2.1 | 79.5 ± 13.4 |
| 250 | 99.9 ± 0.1 | 86.9 ± 16.6 | 99.9 ± 0.1 | 84.7 ± 15.5 |
| 260 | 99.5 ± 0.6 | 67.5 ± 13.6 | 99.2 ± 0.8 | 67.5 ± 12.9 |
| 270 | 100.0 ± 0.0 | 73.7 ± 1.1 | 100.0 ± 0.0 | 71.7 ± 2.9 |
| 280 | 100.0 ± 0.0 | 74.1 ± 18.1 | 100.0 ± 0.0 | 72.2 ± 19.2 |
| 290 | 100.0 ± 0.0 | 83.9 ± 0.0 | 100.0 ± 0.0 | 82.1 ± 0.0 |
| 300 | 100.0 ± 0.0 | 97.5 ± 0.0 | 100.0 ± 0.0 | 94.0 ± 0.0 |