Code implementation for EMNLP 24 findings: Pseudo Label Enhanced Prototypical Contrastive Learning Framework towards Uniformed Intent Discovery. Prepare the environment:

python==3.6.13
pip install -r requirements.txt

If you use the NIVIDIA A100 80GB PCIe GPU with PyTorch, the torch version should be updated

pip install torch==1.7.1+cu110 torchvision==0.8.2+cu110 torchaudio==0.7.2 -f https://download.pytorch.org/whl/torch_stable.html

Run the experiments by:

run.sh

You can change the parameters here.

Some code references the following repositories:

• DKT

• DeepAligned