Skip to content

Latest commit

 

History

History
15 lines (11 loc) · 412 Bytes

README.md

File metadata and controls

15 lines (11 loc) · 412 Bytes

Fusion-BERT

Using BERT and Siamese structure to do the text matching task.

Model

This model uses BERT for sentence pair task to get [CLS] vector $\vec{a}$, and uses BERT for single text classification task to get mean token vector $\vec{b}$ and $\vec{c}$, at last concat three similar information to predict. model

How to use

sh run.sh

Result

result