burst_16
done · peak unseen_mismatch 94.5
α schedule
α 5.0 ×25 α 5.0 ×25 α 5.0 ×25 α 0.7 ×25 α 0.7 ×25 α 3.1429 ×25 α 3.1429 ×25 α 3.1429 ×25 α 0.7 ×25 α 0.7 ×25 α 0.7 ×25 α 1.2857 ×25 α 1.2857 ×25 α 0.7 ×25 α 0.7 ×25 α 0.7 ×25
Figures
Results
accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 0.7
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 | 51.1 ± 1.2 | 50.5 ± 1.3 | 49.5 ± 1.5 | 50.1 ± 1.8 |
| 50 | 51.5 ± 1.7 | 50.5 ± 1.5 | 50.2 ± 1.8 | 49.3 ± 1.4 |
| 75 | 53.8 ± 2.1 | 51.4 ± 1.8 | 50.7 ± 2.5 | 49.7 ± 1.6 |
| 100 | 56.0 ± 8.3 | 51.7 ± 3.6 | 53.9 ± 8.4 | 50.9 ± 2.9 |
| 125 | 64.3 ± 15.0 | 51.7 ± 4.5 | 62.2 ± 15.6 | 50.4 ± 3.7 |
| 150 | 83.3 ± 15.9 | 64.8 ± 12.7 | 81.6 ± 16.8 | 62.5 ± 11.5 |
| 175 | 93.1 ± 11.3 | 85.1 ± 13.0 | 91.1 ± 11.6 | 81.9 ± 13.1 |
| 200 | 95.1 ± 7.3 | 91.0 ± 10.1 | 92.0 ± 9.5 | 86.5 ± 11.1 |
| 225 | 99.1 ± 3.8 | 93.9 ± 13.0 | 98.9 ± 4.5 | 92.7 ± 13.9 |
| 250 | 100.0 ± 0.0 | 94.0 ± 13.2 | 100.0 ± 0.0 | 93.2 ± 13.6 |
| 275 | 100.0 ± 0.0 | 91.1 ± 16.1 | 100.0 ± 0.0 | 90.9 ± 16.0 |
| 300 | 100.0 ± 0.1 | 94.3 ± 14.2 | 99.9 ± 0.2 | 93.4 ± 14.0 |
| 325 | 100.0 ± 0.0 | 94.5 ± 17.4 | 99.8 ± 0.2 | 93.7 ± 17.4 |
| 350 | 100.0 ± 0.0 | 92.9 ± 18.9 | 100.0 ± 0.0 | 92.7 ± 18.8 |
| 375 | 100.0 ± 0.0 | 94.7 ± 14.2 | 100.0 ± 0.0 | 94.5 ± 14.0 |
| 400 | 100.0 ± 0.0 | 91.8 ± 17.4 | 100.0 ± 0.0 | 91.1 ± 17.6 |