The ESConv dataset. GitHub repo. Original paper. @inproceedings{liu-etal-2021-towards, title={Towards Emotional Support Dialog Systems}, author={Liu, Siyang and Zheng, Chujie and Demasi, Orianna and Sabour, Sahand and Li, Yu and Yu, Zhou and Jiang, Yong and Huang, Minlie}, booktitle={ACL}, year={2021} }

Overview

The approach for creating this dataset is outlined in our paper Steering Conversational Large Language Models for Long Emotional Support Conversations. This dataset specifically is generated using attention guided prompting on llama3-70b-instruct. We start from the original ESConv dataset and generate strategy conditioned continuations at random points of the conversation. For example, we cut a conversation at 10th turn when it is the emotional supporters turn to answer and… See the full description on the dataset page: https://huggingface.co/datasets/navidmadani/extended_esc.