r4_c2
done · peak unseen_mismatch 98.1
α schedule
α 4.2358 ×25 α 4.0116 ×25 α 3.7763 ×25 α 3.5344 ×25 α 3.2878 ×25 α 3.0374 ×25 α 2.7839 ×25 α 2.5278 ×25 α 2.2694 ×25 α 2.009 ×25 α 1.7467 ×25 α 1.4827 ×25 α 1.2172 ×25 α 0.9502 ×25 α 0.682 ×25 α 0.4124 ×25
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 0.4124
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 | 50.8 ± 1.3 | 50.5 ± 1.6 | 49.8 ± 1.5 | 49.5 ± 1.9 |
| 50 | 51.6 ± 1.6 | 50.3 ± 1.7 | 50.1 ± 1.1 | 49.9 ± 1.6 |
| 75 | 52.5 ± 2.3 | 50.3 ± 1.8 | 50.7 ± 1.6 | 49.4 ± 1.5 |
| 100 | 58.7 ± 7.5 | 56.4 ± 5.2 | 53.3 ± 4.6 | 51.4 ± 2.9 |
| 125 | 71.3 ± 11.8 | 67.8 ± 10.6 | 60.0 ± 10.5 | 57.1 ± 8.8 |
| 150 | 79.9 ± 7.4 | 77.2 ± 7.4 | 66.6 ± 8.6 | 63.4 ± 8.1 |
| 175 | 86.3 ± 5.4 | 83.7 ± 5.0 | 72.8 ± 10.1 | 69.3 ± 9.4 |
| 200 | 89.1 ± 3.5 | 86.8 ± 3.4 | 75.2 ± 7.2 | 71.8 ± 7.1 |
| 225 | 92.5 ± 2.5 | 90.4 ± 2.6 | 79.6 ± 6.1 | 76.5 ± 6.4 |
| 250 | 94.6 ± 1.5 | 92.9 ± 2.0 | 82.7 ± 5.7 | 79.7 ± 5.7 |
| 275 | 97.2 ± 1.2 | 95.9 ± 1.3 | 88.9 ± 4.8 | 86.1 ± 4.8 |
| 300 | 98.4 ± 0.8 | 97.4 ± 1.3 | 93.2 ± 3.7 | 89.8 ± 5.0 |
| 325 | 99.3 ± 0.5 | 98.8 ± 0.6 | 96.4 ± 2.6 | 93.5 ± 3.3 |
| 350 | 99.9 ± 0.1 | 99.7 ± 0.3 | 99.2 ± 0.8 | 97.3 ± 1.9 |
| 375 | 100.0 ± 0.0 | 99.7 ± 0.5 | 99.9 ± 0.3 | 98.1 ± 2.3 |
| 400 | 100.0 ± 0.0 | 95.1 ± 9.7 | 99.9 ± 0.1 | 91.0 ± 13.8 |