Official repository of our NAACL 2025 main conference paper:
DEMO.ipynb shows an example of using EmoDynamiX. You could integrate EmoDynamiX with any LLM you like to make your own strategy-controlled ESC agent!
Model | URL |
---|---|
EmoDynamiX on ESConv | link |
EmoDynamiX on AnnoMI | link |
Pretrained Submodules | link |
Unzip to the project root directory.
Test on ESConv:
./test_roberta_hg_esconv.sh
Test on AnnoMI:
./test_roberta_hg_annomi.sh
Train on ESConv:
./train_roberta_hg_esconv.sh
Train on AnnoMI:
./train_roberta_hg_annomi.sh
If you find our work useful, please cite our paper:
@misc{wan2024emodynamixemotionalsupportdialogue,
title={EmoDynamiX: Emotional Support Dialogue Strategy Prediction by Modelling MiXed Emotions and Discourse Dynamics},
author={Chenwei Wan and Matthieu Labeau and Chloé Clavel},
year={2024},
eprint={2408.08782},
archivePrefix={arXiv},
primaryClass={cs.CL},
url={https://arxiv.org/abs/2408.08782},
}