EMMA is a new conversation model designed for mixed-session conversations, incorporating Egocentric Memory trained on the MiSC dataset. This model focuses on handling dynamic interactions across sessions, where a main speaker engages with different partners.

🚨 This repository is for the adapter of EMMA's dialogue module, which is based on FLAN-T5-Large.

Model Description

Paper: Mixed-Session Conversation with Egocentric Memory
Dataset: MiSC
Code: GitHub

Citation Information

If you use EMMA in your research, please cite the following paper:

bibtex

@inproceedings{jang-etal-2024-mixed,
    title = "Mixed-Session Conversation with Egocentric Memory",
    author = "Jang, Jihyoung  and
      Kim, Taeyoung  and
      Kim, Hyounghun",
    editor = "Al-Onaizan, Yaser  and
      Bansal, Mohit  and
      Chen, Yun-Nung",
    booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2024",
    month = nov,
    year = "2024",
    address = "Miami, Florida, USA",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2024.findings-emnlp.689/",
    doi = "10.18653/v1/2024.findings-emnlp.689",
    pages = "11786--11815"
}

jihyoung/EMMA

Egocentric Memory Enhanced Mixed-Session Conversation Agent (EMMA)

Introduction

Model Description

Citation Information

jihyoung/
EMMA