r6_c6
done · peak unseen_mismatch 76.8
α schedule
α 4.4834 ×25 α 3.9114 ×25 α 3.4876 ×25 α 0.0554 ×25 α 0.0498 ×25 α 2.519 ×25 α 2.2486 ×25 α 1.9945 ×25 α 0.0311 ×25 α 0.027 ×25 α 0.0231 ×25 α 1.0926 ×25 α 0.8882 ×25 α 0.0122 ×25
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 0.0122
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 | 50.7 ± 1.3 | 50.5 ± 1.6 | 49.9 ± 1.5 | 49.5 ± 1.9 |
| 50 | 51.5 ± 1.7 | 50.4 ± 1.5 | 50.1 ± 1.3 | 49.9 ± 1.7 |
| 75 | 52.2 ± 2.2 | 50.2 ± 1.7 | 50.7 ± 1.3 | 49.5 ± 1.6 |
| 100 | 53.1 ± 5.0 | 50.7 ± 1.7 | 51.9 ± 3.8 | 49.6 ± 1.6 |
| 125 | 59.5 ± 15.2 | 47.8 ± 7.7 | 57.5 ± 15.6 | 47.4 ± 7.9 |
| 150 | 70.5 ± 18.6 | 50.7 ± 11.4 | 68.6 ± 19.0 | 49.1 ± 10.3 |
| 175 | 83.1 ± 18.9 | 64.2 ± 16.5 | 82.6 ± 19.3 | 62.9 ± 15.6 |
| 200 | 92.1 ± 15.1 | 78.4 ± 19.0 | 91.5 ± 15.3 | 76.8 ± 18.7 |
| 225 | 98.4 ± 6.7 | 66.6 ± 20.3 | 98.4 ± 6.7 | 67.0 ± 19.3 |
| 250 | 99.8 ± 1.2 | 53.0 ± 26.7 | 99.7 ± 1.3 | 53.3 ± 26.2 |
| 275 | 100.0 ± 0.0 | 47.8 ± 33.5 | 100.0 ± 0.0 | 47.8 ± 33.5 |
| 300 | 100.0 ± 0.0 | 49.5 ± 31.1 | 100.0 ± 0.0 | 49.8 ± 31.0 |
| 325 | 100.0 ± 0.0 | 55.4 ± 31.9 | 100.0 ± 0.0 | 56.0 ± 31.7 |
| 350 | 100.0 ± 0.0 | 48.0 ± 31.0 | 100.0 ± 0.0 | 48.5 ± 30.4 |