r4_c7
done · peak unseen_mismatch 99.1
α schedule
α 3.1128 ×25 α 3.0863 ×25 α 3.0348 ×25 α 2.9587 ×25 α 2.8585 ×25 α 2.7343 ×25 α 2.5864 ×25 α 2.4148 ×25 α 2.2199 ×25 α 2.0016 ×25 α 1.7601 ×25 α 1.4955 ×25 α 1.2078 ×25 α 0.8972 ×25 α 0.5636 ×25 α 0.2073 ×25
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 0.2073
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 ± 1.2 | 50.3 ± 1.7 | 50.0 ± 1.5 | 49.5 ± 1.9 |
| 50 | 51.3 ± 1.5 | 50.4 ± 1.7 | 49.9 ± 1.3 | 49.9 ± 1.8 |
| 75 | 52.1 ± 1.9 | 50.1 ± 1.7 | 50.4 ± 1.4 | 49.6 ± 1.6 |
| 100 | 55.3 ± 6.7 | 53.0 ± 3.2 | 52.1 ± 3.6 | 50.0 ± 1.5 |
| 125 | 65.9 ± 12.7 | 61.0 ± 10.4 | 58.6 ± 11.4 | 54.6 ± 7.4 |
| 150 | 76.8 ± 13.0 | 73.0 ± 12.4 | 68.9 ± 13.7 | 64.7 ± 12.5 |
| 175 | 85.6 ± 9.7 | 82.5 ± 9.3 | 75.2 ± 13.4 | 70.9 ± 12.2 |
| 200 | 90.9 ± 6.8 | 88.4 ± 6.6 | 81.2 ± 10.0 | 77.1 ± 10.3 |
| 225 | 94.1 ± 3.1 | 91.8 ± 3.5 | 84.5 ± 6.9 | 80.7 ± 7.6 |
| 250 | 95.3 ± 1.8 | 93.4 ± 2.2 | 86.5 ± 5.1 | 83.0 ± 6.1 |
| 275 | 97.4 ± 1.1 | 95.8 ± 1.3 | 90.3 ± 4.9 | 86.7 ± 5.1 |
| 300 | 98.5 ± 0.9 | 97.4 ± 1.1 | 94.0 ± 3.6 | 90.5 ± 4.1 |
| 325 | 99.6 ± 0.3 | 98.8 ± 0.7 | 97.0 ± 2.3 | 94.1 ± 3.1 |
| 350 | 100.0 ± 0.1 | 99.6 ± 0.4 | 99.5 ± 0.7 | 98.0 ± 1.7 |
| 375 | 100.0 ± 0.0 | 99.8 ± 0.4 | 100.0 ± 0.1 | 99.1 ± 1.4 |
| 400 | 100.0 ± 0.0 | 94.3 ± 14.1 | 100.0 ± 0.0 | 92.0 ± 16.1 |