Table 8: 
The ratio of outputs with each error type for each compared system. Lower percentage is better.
ModelUnnatural Language %Missing Information %Node Mistranslation %Edge Mistranslation %
S2S 62.0 60.0 34.0 52.0 
GraphLSTM 54.0 54.0 28.0 46.0 
Graph Transformer 48.0 44.0 28.0 40.0 
ModelUnnatural Language %Missing Information %Node Mistranslation %Edge Mistranslation %
S2S 62.0 60.0 34.0 52.0 
GraphLSTM 54.0 54.0 28.0 46.0 
Graph Transformer 48.0 44.0 28.0 40.0 
Close Modal

or Create an Account

Close Modal
Close Modal