imagining syntax experiment results

← depth_long

n-layer_4

done · peak unseen_mismatch 88.4

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 ± 0.053.9 ± 17.599.3 ± 1.052.6 ± 18.6
1.0 100.0 ± 0.074.4 ± 21.499.0 ± 1.067.6 ± 21.8
1.1 100.0 ± 0.182.1 ± 20.496.3 ± 3.470.6 ± 20.9
1.2 100.0 ± 0.189.4 ± 7.497.2 ± 2.276.8 ± 11.4
1.3 99.9 ± 0.192.6 ± 5.593.6 ± 3.273.3 ± 14.5
1.4 99.8 ± 0.294.4 ± 4.892.5 ± 2.974.6 ± 12.8
1.5 99.8 ± 0.293.2 ± 7.391.9 ± 2.965.7 ± 13.3
1.6 99.7 ± 0.291.8 ± 6.392.2 ± 1.962.1 ± 14.3
1.7 99.6 ± 0.395.6 ± 4.388.3 ± 2.868.9 ± 13.4
1.8 99.7 ± 0.395.1 ± 5.085.8 ± 5.065.4 ± 14.4
1.9 99.7 ± 0.297.1 ± 2.487.5 ± 3.267.6 ± 10.6
2.0 99.7 ± 0.297.4 ± 1.787.1 ± 3.664.1 ± 10.9
2.1 99.7 ± 0.195.2 ± 4.185.6 ± 1.561.7 ± 14.3
2.5 99.8 ± 0.198.6 ± 1.081.8 ± 2.662.3 ± 8.5
3.0 99.9 ± 0.198.3 ± 1.578.5 ± 2.155.4 ± 11.4

Downloads