This is a
RoBERTa-large model trained on the
Tweet-FID dataset (
"TWEET-FID: An Annotated Dataset for Multiple Foodborne Illness Detection Tasks", Ruofan Hu et al, 2022 ) which is a collection of Twitter to detect incidents of foodborne illnesses.
The model is enriched with a multi class classification head to perform the custom task called Entity Mention Detection (EMD).
The objective is to determine predefined entities (food, location, symptom, other) in a given text related to a food risk.
