peak30_staged

done · peak unseen_mismatch 94.4

Referenced by

α 3.0 ×150 α 1.0 ×150 α 2.1429 ×150 α 1.0 ×150 α 1.2857 ×150 α 1.0 ×150

accuracy by training iteration, mean ± sd over seeds · schedule ends at α = 1.0

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 ± 1.5	49.8 ± 2.0	50.1 ± 1.8	49.9 ± 1.5
50	51.4 ± 1.7	50.1 ± 1.6	50.7 ± 1.7	49.6 ± 1.3
75	52.2 ± 2.7	49.6 ± 1.6	50.7 ± 2.2	49.7 ± 2.1
100	54.8 ± 7.0	51.8 ± 3.9	52.6 ± 6.4	50.2 ± 2.7
125	63.5 ± 12.7	56.8 ± 10.1	60.4 ± 12.0	54.3 ± 8.6
150	80.1 ± 15.9	70.3 ± 14.2	76.8 ± 16.9	66.9 ± 13.6
175	92.6 ± 12.9	74.2 ± 16.2	91.8 ± 13.8	72.1 ± 16.2
200	98.3 ± 6.7	70.7 ± 22.3	98.3 ± 7.1	69.6 ± 22.4
225	99.9 ± 0.5	67.8 ± 26.0	99.9 ± 0.5	67.6 ± 25.4
250	100.0 ± 0.0	64.4 ± 28.1	100.0 ± 0.0	63.6 ± 27.5
275	100.0 ± 0.0	63.8 ± 31.4	100.0 ± 0.0	63.4 ± 30.8
300	100.0 ± 0.0	61.6 ± 34.9	100.0 ± 0.0	61.4 ± 34.7
325	100.0 ± 0.0	70.0 ± 30.5	99.9 ± 0.3	69.3 ± 30.2
350	99.9 ± 0.2	79.2 ± 25.2	99.6 ± 0.9	77.8 ± 24.7
375	99.8 ± 0.5	89.2 ± 20.2	99.3 ± 1.5	87.3 ± 20.4
400	99.7 ± 0.6	93.4 ± 16.9	98.6 ± 2.7	91.1 ± 17.3
450	99.2 ± 1.1	95.1 ± 10.8	97.2 ± 4.0	91.6 ± 11.7
500	100.0 ± 0.1	95.3 ± 12.0	99.9 ± 0.4	94.4 ± 12.4
550	100.0 ± 0.1	95.6 ± 16.8	99.8 ± 0.5	94.0 ± 16.9
600	100.0 ± 0.1	94.3 ± 12.9	99.8 ± 0.6	90.8 ± 14.5
650	99.9 ± 0.1	95.7 ± 10.5	99.2 ± 1.6	91.5 ± 11.8
700	99.9 ± 0.2	96.3 ± 10.2	99.0 ± 1.5	91.0 ± 12.6
750	99.9 ± 0.1	96.2 ± 8.5	99.1 ± 1.2	91.0 ± 11.2
800	100.0 ± 0.1	96.6 ± 8.3	99.5 ± 1.0	91.2 ± 12.7
850	100.0 ± 0.0	95.0 ± 14.6	99.6 ± 0.9	88.6 ± 17.9
900	100.0 ± 0.0	94.2 ± 12.5	99.5 ± 0.8	87.1 ± 16.6