Task Conditioned BERT for Joint Intent Detection and Slot-filling

Predict intent and slot at the same time from one BERT model ; total_loss = intent_loss + coef * slot_loss (Change coef with --slot_loss_coef

BERT - Sv Slot Value Matching Slot Gate Refer Prediction Candidate value Vp Slot Name query BERT - Sv Slot - Level Schema Graph hotel

bert stevens poker A Bert-based Joint Model of Intent Recognition and Slot Filling Cross-correlation Abstract: Intent recognition and slot filling are two key steps in natural

bert for joint intent classification and slot filling In this work, propose a joint intent classification and slot filling model based on BERT