imagining syntax

← ascent_v1

s4_v07

done · peak unseen_mismatch 99.6

Referenced by

α schedule

α 4.0 ×19 α 4.0 ×19 α 4.0 ×19 α 4.0 ×19 α 4.0 ×19 α 4.0 ×19 α 4.0 ×19 α 4.0 ×19 α 0.7 ×19 α 0.7 ×19 α 0.7 ×19 α 0.7 ×19 α 0.7 ×19 α 0.7 ×19 α 0.7 ×19 α 0.7 ×19 α 2.6429 ×19 α 2.6429 ×19 α 2.6429 ×19 α 2.6429 ×19 α 2.6429 ×19 α 2.6429 ×19 α 2.6429 ×19 α 2.6429 ×19 α 0.7 ×19 α 0.7 ×19 α 0.7 ×19 α 0.7 ×19 α 0.7 ×19 α 0.7 ×19 α 0.7 ×19 α 0.7 ×19 α 1.2857 ×19 α 1.2857 ×19 α 1.2857 ×19 α 1.2857 ×19 α 1.2857 ×18 α 1.2857 ×18 α 1.2857 ×18 α 1.2857 ×18 α 0.7 ×18 α 0.7 ×18 α 0.7 ×18 α 0.7 ×18 α 0.7 ×18 α 0.7 ×18 α 0.7 ×18 α 0.7 ×18

direction
mixed
α range
0.7–4.0
stages
48

Figures

schedule.png
schedule.png
trajectory.png
trajectory.png

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

iterationseen_matchseen_mismatchunseen_matchunseen_mismatch
25 51.5 ± 1.649.7 ± 1.649.9 ± 2.049.7 ± 2.1
50 51.1 ± 1.749.9 ± 1.350.1 ± 1.550.4 ± 1.5
75 52.9 ± 2.150.9 ± 1.950.6 ± 1.750.0 ± 2.1
100 58.7 ± 5.555.5 ± 4.252.5 ± 4.051.1 ± 2.2
125 72.0 ± 10.167.7 ± 9.261.9 ± 9.357.9 ± 7.3
150 80.1 ± 7.877.2 ± 8.168.6 ± 9.464.3 ± 7.9
175 94.2 ± 6.788.5 ± 10.290.4 ± 9.183.7 ± 11.2
200 99.5 ± 1.991.7 ± 13.299.4 ± 1.990.2 ± 14.0
225 99.9 ± 0.488.5 ± 21.799.8 ± 0.887.9 ± 21.7
250 100.0 ± 0.089.1 ± 24.8100.0 ± 0.088.7 ± 24.8
275 100.0 ± 0.091.4 ± 23.6100.0 ± 0.090.7 ± 23.5
300 100.0 ± 0.091.8 ± 23.1100.0 ± 0.091.1 ± 22.9
325 98.9 ± 1.292.9 ± 16.996.0 ± 3.986.8 ± 16.2
350 97.7 ± 1.593.3 ± 8.493.6 ± 4.686.1 ± 9.4
375 96.8 ± 1.694.5 ± 2.891.7 ± 4.286.0 ± 7.0
400 96.4 ± 1.693.7 ± 2.389.4 ± 4.583.6 ± 5.9
450 95.1 ± 1.693.3 ± 1.986.1 ± 4.380.4 ± 5.7
500 100.0 ± 0.099.9 ± 0.2100.0 ± 0.099.6 ± 0.6
550 100.0 ± 0.099.9 ± 0.299.9 ± 0.199.2 ± 1.1
600 100.0 ± 0.099.8 ± 0.3100.0 ± 0.298.1 ± 2.6
650 99.8 ± 0.298.7 ± 1.197.8 ± 1.591.8 ± 6.2
700 99.8 ± 0.398.4 ± 1.497.2 ± 2.289.4 ± 7.4
750 99.8 ± 0.298.4 ± 1.397.5 ± 1.588.5 ± 7.5
800 100.0 ± 0.098.6 ± 2.799.9 ± 0.293.7 ± 6.8
850 100.0 ± 0.098.6 ± 2.199.9 ± 0.392.8 ± 8.2
900 100.0 ± 0.096.3 ± 6.199.6 ± 0.587.5 ± 10.0

Downloads