Compare with HuggingFace's HMTL’s results on the same query. Or compare with CoreNLP’s results on the same query. Or compare with spaCy’s displaCy results on entity recognition.

This is a demo of CoReNer, our new multi-task model for document understanding. The model combines Named Entity Recognition, Entity Mention Detection, Relation Extraction and Coreference Resolution. These 4 tasks are all fundamentals for NLP and NLU applications such as conversational agents. Moreover, these 4 tasks can benefit each other when trained together in a multi-task learning framework.

This demo is based on the HMTL demo by HuggingFace 🤗. The model used for this demo is a RoBERTa base based CoReNer model, trained on the OntoNotes (NER, EMD, CR) and CoNLL04 datasets (NER, RE). The model checkpoint is available at HuggingFace's model hub: aiola/roberta-base-corener. Check out the model card for more details.