Skip to Main Content
Table A.2

Accuracy of GRU as a function of the language and input data complexity. Training data in each run includes examples of complexity up to and including n; testing data (disjoint from training) contained examples of complexity exactly n. Random baseline is 0.25.

Test ex. complexity:34567
Right branching 0.17 0.42 0.21 0.37 0.25 
Left branching 0.98 0.95 1 0.92 0.94 
Left, no curriculum 0.45 0.3 0.34 0.34 0.31 
Left, steep curriculum 1 0.99 0.81 0.43 0.43 
Left, slow curriculum 1 1 0.997 1 0.997 
Test ex. complexity:34567
Right branching 0.17 0.42 0.21 0.37 0.25 
Left branching 0.98 0.95 1 0.92 0.94 
Left, no curriculum 0.45 0.3 0.34 0.34 0.31 
Left, steep curriculum 1 0.99 0.81 0.43 0.43 
Left, slow curriculum 1 1 0.997 1 0.997 
Close Modal

or Create an Account

Close Modal
Close Modal