This repository provides the official PyTorch implementation of the research paper 'A Probabilistic Framework for Discovering New Intents'
We use anaconda to create python environment:
conda create --name python=3.9
Install all required libraries:
pip install -r requirements.txt
Get the pre-trained BERT model and convert it into Pytorch.
Set the path of the uncased-bert model (parameter "bert_model" in init_parameter.py).
Run the experiments by:
sh scripts/run.sh
To compare with MTP, please add the parameter --load_mtp and move the checkpoint to the path.
We use the MTP pretrain checkpoint https://github.com/fanolabs/NID_ACLARR2022. It gives pretrain ckpt in step1 , we further pretrain the checkpoint using its step2 method.
The model architecture of Our method:
Our code implementation is on the basis of https://github.com/thuiar/DeepAligned-Clustering