The option to run the ERG with ubertagging is ubertagging=0.001
, according to the docs. But I am assuming first I have to compile the grammar properly, giving it a path to the pretrained model? Since when I use the default supplied config.tdl
and then run ACE with the ubertagging=0.001
option (what does 0.001 stand for, btw?), then I get a
WARNING: cannot enable übertagging; no model present in this grammar image.
I see the models are stored in the ut
folder in the ERG releases, but what is the correct way of including them?