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
All requirements can be found in requirements.txt. You can install all required packages in a new environment with pip install -r requirements.txt.
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 |
The usage will be updated soon...