imagining syntax experiment results

← depth_long

n-layer_3

done · peak unseen_mismatch 93.1

Figures

accuracy_at_iter_01200.png
accuracy_at_iter_01200.png
accuracy_at_iter_01500.png
accuracy_at_iter_01500.png
accuracy_at_iter_02400.png
accuracy_at_iter_02400.png
accuracy_at_iter_03600.png
accuracy_at_iter_03600.png
accuracy_at_iter_04800.png
accuracy_at_iter_04800.png
waypoint_evolution.png
waypoint_evolution.png

Results

paramseen_matchseen_mismatchunseen_matchunseen_mismatch
0.9 100.0 ± 0.151.1 ± 23.496.6 ± 2.854.7 ± 18.9
1.0 100.0 ± 0.163.2 ± 16.695.4 ± 3.157.2 ± 16.3
1.1 99.9 ± 0.179.4 ± 11.095.8 ± 2.464.5 ± 11.4
1.2 99.9 ± 0.189.3 ± 4.591.8 ± 3.471.4 ± 4.0
1.3 99.9 ± 0.193.0 ± 3.691.5 ± 3.871.0 ± 9.6
1.4 99.8 ± 0.194.9 ± 3.489.0 ± 3.769.9 ± 11.5
1.5 99.8 ± 0.197.2 ± 2.389.2 ± 3.278.1 ± 7.9
1.6 99.6 ± 0.396.7 ± 2.389.1 ± 3.071.6 ± 11.9
1.7 99.6 ± 0.296.8 ± 2.687.9 ± 4.067.0 ± 11.8
1.8 99.8 ± 0.297.2 ± 1.686.5 ± 3.769.4 ± 8.8
1.9 99.8 ± 0.297.8 ± 1.085.4 ± 1.764.6 ± 9.3
2.0 99.8 ± 0.197.5 ± 2.786.3 ± 2.667.8 ± 10.3
2.1 99.7 ± 0.297.6 ± 1.985.7 ± 2.865.4 ± 10.1
2.5 99.8 ± 0.199.1 ± 0.483.4 ± 2.168.3 ± 8.0
3.0 99.8 ± 0.199.3 ± 0.679.7 ± 3.568.1 ± 7.7

Downloads