Personalized Knowledge-grounded Dialogue

This repository contains the code for our paper There Are a Thousand Hamlets in a Thousand People's Eyes: \ Enhancing Knowledge-grounded Dialogue with Personal Memory

Set up

All requirements can be found in requirements.txt. You can install all required packages in a new environment with pip install -r requirements.txt.

Prepare for the dataset

Due to some issues, please directly contact the author if you want a copy of our dataset.

Some statistics for the dataset:

	Train	Valid	Test
# Dialogue	217,095	11,186	6,236
# Utterances	1,442,975	74,480	41,519
# Knowledges	5,459,744	290,349	148,057
# User	48,858	5,603	3,281
# Memory facets	490,460	70,494	38,354
Utterance length	34.15	33.95	33.60
Knowledge length	54.54	52.39	53.17
Memory facet length	42.10	40.21	40.60

Usage

The usage will be updated soon...