Published
empathy-conversations
Multi-turn dialogue centred on emotional support — ICU waiting rooms, caregiver burnout, end-of-life conversations, loneliness inside relationships, and the ordinary exchanges where tone matters more than information.
Released as AntEngage's first open artefact. The intent is narrow and deliberate: most conversational corpora optimise for task completion, so models trained on them learn to resolve rather than to sit with a problem. This set is the opposite case.
| Rows | 4,008 conversations |
|---|---|
| Turns | 79,616 total · 19.9 average · 20 maximum |
| Language | English |
| Format | JSONL · CSV |
| Tasks | Dialogue modelling · generation |
| Provenance | Fully AI-synthesised by AntEngage's own generation pipeline, then quality-verified by an automated cross-checker. No real user data. |
| License | CC BY-SA 4.0 |
Limitations
- English only. It does not yet cover the Indic and code-switch settings AntEngage's production systems run in.
- Emotional-support scenarios are a narrow slice of conversation; do not treat coverage here as coverage of dialogue generally.
- Not a clinical resource, and not validated for any diagnostic or therapeutic use.
- Fully synthetic, and verified by an automated checker rather than by clinicians or people with lived experience. It reflects the distributional habits of the models that generated it as much as how people in distress actually speak, so treat it as material for training tone rather than as evidence about real conversations.
Cite
@dataset{antengage_empathy_2026,
title = {AntEngage Empathy Conversation Dataset},
author = {Pradhan, Dibyaprakash},
year = {2026},
publisher = {AntEngage Technology Private Limited},
doi = {10.7910/DVN/LWUFLG},
license = {CC BY-SA 4.0}
}
Mirrors · 6
Canonical listing, changelog, and future releases: antengage.com/datasets.