ConvLab-3 is a flexible dialog system platform based on a unified data format for task-oriented dialog (TOD) datasets. The unified format serves as the adapter between TOD datasets and models: datasets are first transformed to the unified format and then loaded by models. In this way, the cost of adapting
- 2023.8.6: Add LLM-based models.
- 2023.2.26: Update ConvLab on PyPI to 3.0.1 to reflect bug fixes.
- 2022.11.30: ConvLab-3 release.
You can install ConvLab-3 in one of the following ways according to your need. We use torch>=1.10.1,<=1.13 and transformers>=4.17.0,<=4.24.0. Higher versions of torch and transformers may also work.
For the latest and most configurable version, we recommend installing ConvLab-3 in development mode.
Clone the newest repository:
git clone --depth 1 https://github.com/ConvLab/ConvLab-3.gitInstall ConvLab-3 via pip:
cd ConvLab-3
pip install -e .To use ConvLab-3 as an off-the-shelf tool, you can install via:
pip install convlabNote that the data directory will not be included due to the package size limitation.
We also provide Dockerfile for building docker. Basically it uses the requirement.txt and then installs ConvLab-3 in development mode.
# create image
docker build -t convlab .
# run container
docker run -dit convlab
# open bash in container
docker exec -it CONTAINER_ID bash- Getting Started (Have a try on Colab!)
- Introduction to Unified Data Format
- Utility functions for unified datasets
- RL Toolkit
- Interactive Tool [demo video]
Current datasets in unified data format: (DA-U/DA-S stands for user/system dialog acts)
| Dataset | Dialogs | Goal | DA-U | DA-S | State | API result | DataBase |
|---|---|---|---|---|---|---|---|
| Camrest | 676 | ✅ | ✅ | ✅ | ✅ | ✅ | |
| WOZ 2.0 | 1200 | ✅ | ✅ | ||||
| KVRET | 3030 | ✅ | ✅ | ✅ | |||
| DailyDialog | 13118 | ✅ | |||||
| Taskmaster-1 | 13175 | ✅ | ✅ | ✅ | |||
| Taskmaster-2 | 17303 | ✅ | ✅ | ✅ | |||
| MultiWOZ 2.1 | 10438 | ✅ | ✅ | ✅ | ✅ | ✅ | |
| Schema-Guided | 22825 | ✅ | ✅ | ✅ | ✅ | ||
| MetaLWOZ | 40203 | ✅ | |||||
| CrossWOZ (zh) | 6012 | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
| Taskmaster-3 | 23757 | ✅ | ✅ | ✅ | ✅ |
Unified datasets are available under data/unified_datasets directory as well as Hugging Face Hub. We will continue adding more datasets listed in this issue. If you want to add a listed/custom dataset to ConvLab-3, you can create an issue for discussion and then create pull-request. We will list you as the contributors and highly appreciate your contributions!
We list newly integrated models in ConvLab-3 that support unified data format and obtain strong performance. You can follow the link for more details about these models. Other models can be used in the same way as in ConvLab-2.
| Task | Models | Input | Output |
|---|---|---|---|
| Response Generation | T5RG, LLMs | Context | Response |
| Goal-to-Dialogue | T5Goal2Dialogue | Goal | Dialog |
| Natural Language Understanding | T5NLU, BERTNLU, MILU, LLMs | Context | DA-U |
| Dialog State Tracking | T5DST, SUMBT, SetSUMBT, TripPy, LLMs | Context | State |
| RL Policy | DDPT, PPO, PG | State, DA-U, DB | DA-S |
| Word-Policy | LAVA | Context, State, DB | Response |
| Natural Language Generation | T5NLG, SC-GPT, LLMs | DA-S | Response |
| End-to-End | SOLOIST | Context, DB | State, Response |
| User simulator | TUS, GenTUS, LLMs | Goal, DA-S | DA-U, (Response) |
Trained models are available on Hugging Face Hub.
We welcome contributions from community. Please see issues to find what we need.
- If you want to add a new dataset, model, or other feature, please describe the dataset/model/feature in an issue with corresponding issue template before creating pull-request.
- Small change like fixing a bug can be directly made by a pull-request.
.
├── convlab # Source code, installed in pypi package
│ ├── dialog_agent # Interface for dialog agent and session
│ ├── base_models
│ │ ├── llm # LLM-based models
│ │ │ ├── user_simulator # LLM-based user simulator and RG
│ │ │ ├── dst # LLM-based DST
│ │ │ ├── nlu # LLM-based NLU
│ │ │ └── nlg # LLM-based NLG
│ │ │
│ │ └── t5 # T5 models with a unified training script
│ │ ├── goal2dialogue # T5Goal2Dialogue
│ │ ├── dst # T5DST
│ │ ├── nlu # T5NLU
│ │ ├── nlg # T5NLG
│ │ └── rg # T5RG
│ │
│ ├── nlu # NLU models, interface, and evaluation script
│ │ ├── jointBERT # BERTNLU
│ │ ├── milu # MILU
│ │ └── svm # SVMNLU*
│ │
│ ├── laug # Language understanding AUGmentation (LAUG) toolkit
│ │
│ ├── dst # DST models, interface, and evaluation script
│ │ ├── rule # RuleDST
│ │ ├── setsumbt # SetSUMBT, has uncertainty estimates
│ │ ├── sumbt # SUMBT
│ │ ├── trippy # TripPy
│ │ ├── trade # TRADE*
│ │ ├── comer # COMER*
│ │ ├── mdbt # MDBT*
│ │ └── dstc9 # scripts for DSTC9 cross-lingual DST evaluation
│ │
│ ├── policy # Policy models, interface, and RL toolkit
│ │ ├── vector # vectorizer class
│ │ ├── plot_results # RL plotting tool
│ │ ├── mle # MLE (imitation learning) policy
│ │ ├── pg # Policy Gradient
│ │ ├── ppo # Proximal Policy Optimization
│ │ ├── vtrace_DPT # DDPT
│ │ ├── lava # LAVA
│ │ ├── rule # Rule policies and rule-based user simulators
│ │ ├── tus # TUS
│ │ ├── genTUS # GenTUS
│ │ ├── dqn # DQN*
│ │ ├── gdpl # GDPL*
│ │ ├── vhus # VHUS*
│ │ ├── hdsa # HDSA*
│ │ ├── larl # LARL*
│ │ └── mdrg # MDRG*
│ │
│ ├── nlg # NLG models, interface, and evaluation script
│ │ ├── scgpt # SC-GPT
│ │ ├── sclstm # SC-LSTM
│ │ └── template # TemplateNLG*
│ │
│ ├── e2e # End2End models
│ │ ├── soloist # SOLOIST
│ │ ├── damd # DAMD*
│ │ └── sequicity # Sequicity*
│ │
│ ├── evaluator # Evaluator for interactive evaluation
│ ├── human_eval # Human evaluation with AMT
│ ├── task # Goal generators for MultiWOZ, CrossWOZ, and Camrest
│ ├── util
│ │ └── unified_datasets_util.py # Utility function for unified data format
│ └── deploy # Deploy system for human conversion
│
├── data # Data dir, not included in pypi package
│ ├── ... # ConvLab-2 data, not available for pypi installation
│ └── unified_datasets # Unified datasets, available for pypi installation
├── examples
│ └── agent_examples # Examples of building user and system agents
└── tutorials # Tutorials*: models do not support unified datasets, only support MultiWOZ.
ConvLab-3 is maintained and developed by Tsinghua University Conversational AI group (THU-COAI), the Dialogue Systems and Machine Learning Group at Heinrich Heine University, Düsseldorf, Germany and Microsoft Research (MSR).
We would like to thank all contributors of ConvLab:
Yan Fang, Zhuoer Feng, Jianfeng Gao, Qihan Guo, Kaili Huang, Minlie Huang, Sungjin Lee, Bing Li, Jinchao Li, Xiang Li, Xiujun Li, Jiexi Liu, Lingxiao Luo, Wenchang Ma, Mehrad Moradshahi, Baolin Peng, Runze Liang, Ryuichi Takanobu, Dazhen Wan, Hongru Wang, Jiaxin Wen, Yaoqin Zhang, Zheng Zhang, Qi Zhu, Xiaoyan Zhu, Carel van Niekerk, Christian Geishauser, Hsien-chin Lin, Nurul Lubis, Xiaochen Zhu, Michael Heck, Shutong Feng, Milica Gašić.
If you use ConvLab-3 in your research, please cite:
@article{zhu2022convlab3,
title={ConvLab-3: A Flexible Dialogue System Toolkit Based on a Unified Data Format},
author={Qi Zhu and Christian Geishauser and Hsien-chin Lin and Carel van Niekerk and Baolin Peng and Zheng Zhang and Michael Heck and Nurul Lubis and Dazhen Wan and Xiaochen Zhu and Jianfeng Gao and Milica Gašić and Minlie Huang},
journal={arXiv preprint arXiv:2211.17148},
year={2022},
url={http://arxiv.org/abs/2211.17148}
}
Apache License 2.0