r9_c1
done · peak unseen_mismatch 99.2
α schedule
α 5.0 ×25 α 5.0 ×25 α 5.0 ×25 α 0.0264 ×25 α 0.0265 ×25 α 5.0 ×25 α 5.0 ×25 α 5.0 ×25 α 0.027 ×25 α 0.0272 ×25 α 0.0275 ×25 α 5.0 ×25 α 5.0 ×25 α 0.0288 ×25
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 0.0288
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 | 54.1 ± 7.2 | 50.9 ± 1.9 | 52.8 ± 6.4 | 50.0 ± 2.3 |
| 125 | 60.6 ± 15.1 | 48.3 ± 5.9 | 58.6 ± 15.6 | 47.3 ± 7.0 |
| 150 | 74.3 ± 20.2 | 58.3 ± 13.7 | 72.9 ± 20.3 | 56.1 ± 12.0 |
| 175 | 85.7 ± 14.3 | 76.9 ± 15.5 | 83.0 ± 15.3 | 73.4 ± 15.1 |
| 200 | 91.3 ± 7.4 | 87.6 ± 8.4 | 86.7 ± 10.5 | 82.3 ± 10.9 |
| 225 | 99.2 ± 1.7 | 88.6 ± 16.9 | 99.1 ± 1.8 | 87.9 ± 17.1 |
| 250 | 100.0 ± 0.0 | 83.4 ± 22.5 | 100.0 ± 0.0 | 82.5 ± 22.9 |
| 275 | 100.0 ± 0.0 | 72.2 ± 28.1 | 100.0 ± 0.0 | 72.1 ± 28.5 |
| 300 | 100.0 ± 0.1 | 94.7 ± 11.8 | 99.9 ± 0.4 | 93.8 ± 12.6 |
| 325 | 98.6 ± 2.2 | 95.9 ± 7.3 | 97.2 ± 4.1 | 93.2 ± 8.9 |
| 350 | 100.0 ± 0.0 | 99.3 ± 1.9 | 100.0 ± 0.0 | 99.2 ± 1.8 |