This is the repository of our ACL 2024 main paper "ESCoT: Towards Interpretable Emotional Support Dialogue Systems".
Our ESD-CoT dataset is organized under the data folder and is split into three JSON files: train, val, and test. Each file contains samples structured as follows:
{
"id": ,
"original_data": {
"dialog": [
{
"speaker": "seeker",
"content": "Hi, I'm having a really hard time managing my schoolwork and extracurricular activities. I feel like there's just not enough hours in the day."
},
...
{
"speaker": "seeker",
"content": "Yeah, I can try that."
}
],
"strategy": "Providing Suggestions",
"response": "Great, and let's touch base next week to see if the list has been helpful. In the meantime, have you considered talking to your teacher or a guidance counselor about feeling overwhelmed?"
},
"cot_data": {
"emotion": "The seeker feels overwhelmed and stretched thin.",
"emotion_stimuli": "The seeker is struggling to manage schoolwork...",
"individual_appraisal": "The seeker thinks they are not able to do anything well...",
"recognized_strategy": "Providing Suggestions",
"strategy_reason": "To address the seeker's feeling of being overwhelmed and..."
}
}Additionally, we provide instructional format training data in the data/ablation_data folder.
Download the LLAMA2-7B-CHAT model.
The training of LLAMA2-CHAT model is based on the SFT trainer of Transformer Reinforcement Learning.
Run bash scripts/supervised_finetune_llama2_cot.sh to train your model.
Run bash scripts/supervised_finetune_llama2_cot_ablation.sh for Ablation Study model training.
Run bash scripts/test_llama2_chat_sft_cot.sh or scripts/test_llama2_inference_cot.sh.
If you use our codes or your research is related to our work, please kindly cite our paper:
@inproceedings{zhang-etal-2024-escot,
title = "{ESC}o{T}: Towards Interpretable Emotional Support Dialogue Systems",
author = "Zhang, Tenggan and Zhang, Xinjie and Zhao, Jinming and Zhou, Li and Jin, Qin",
booktitle = "Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
year = "2024"
}Please contact zhangxinjie827@ruc.edu.cn for any problems.