Closed MGlauer closed 3 months ago
The semantic loss currently assumes that the chebi version from the chebi100 config is used. Instead, it should derive the required information from the data configs passed to the trainer.
data
The semantic loss currently assumes that the chebi version from the chebi100 config is used. Instead, it should derive the required information from the
dataconfigs passed to the trainer.