Datasetpublic

anezatra/empathetic-dialogues-sharegpt

EmpatheticDialogues - ShareGPT Processed Dataset Summary EmpatheticDialogues is a large-scale, open-domain dialogue dataset designed to help AI systems recognize, understand, and respond to human emotions more naturally. While humans can easily identify and acknowledge others’ feelings during conversation, this remains a major challenge for artificial dialogue agents due to the lack of high-quality empathetic datasets. This dataset introduces a new benchmark for… See the full description on the dataset page: https://huggingface.co/datasets/anezatra/empathetic-dialogues-sharegpt.

sourceHugging Faceapache-2.0updated 10mo agoView on Hugging Face

0likes52downloads

Dataset Card

EmpatheticDialogues - ShareGPT Processed

Dataset Summary

EmpatheticDialogues is a large-scale, open-domain dialogue dataset designed to help AI systems recognize, understand, and respond to human emotions more naturally. While humans can easily identify and acknowledge others’ feelings during conversation, this remains a major challenge for artificial dialogue agents due to the lack of high-quality empathetic datasets.

This dataset introduces a new benchmark for empathetic dialogue generation, containing over 25,000 conversations grounded in emotionally rich situations. Each conversation captures genuine emotional experiences and reactions, enabling AI models to develop a more human-like ability to perceive and express empathy.

It has been converted into the ShareGPT format and further developed for Eliza, an intelligent, human-like conversational character designed to produce natural, emotionally aware, and empathetic dialogue.

Data Structure

Fields

Field	Description
conv_id	Unique identifier for each conversation
situation	Description of the emotional situation or context
emotion	Emotion label corresponding to the conversation
conversations	List of dialogue messages exchanged between roles
├── from	The speaker role: `human` or `gpt`
└── value	The textual content of the message

Splits

Split	Examples	Size (bytes)	Description
Train	19531	13013173	Used for model training
Validation	2769	1942576	Used for tuning and validation
Test	2547	1783153	Used for final model evaluation

Total dataset size: 16738902 bytes Total number of dialogues: 25531

Usage Example

python

from datasets import load_dataset

ds = load_dataset("anezatra/empatheticdialogues-sharegpt", split="train")
print(ds[0])

References

Rashkin, H., Smith, E. M., Li, M., & Boureau, Y. L. (2018). Towards empathetic open-domain conversation models: A new benchmark and dataset. arXiv preprint arXiv:1811.00207. (https://arxiv.org/abs/1811.00207)