<div align="center">
<img width="300px" height="auto" src="./docs/figures/TiZero.png">
</div>
<div align="center">
<img weight="300px" height="auto" src="./docs/figures/screen_800.png">
</div>
[](https://opensource.org/licenses/Apache-2.0)
[](https://pypi.org/project/tizero/)
[](https://tizero.readthedocs.io/en/latest/?badge=latest)
### Introduction
Reinforcement learning agent for Google Research Football.
Code accompanying the paper
"TiZero: Mastering Multi-Agent Football with Curriculum Learning and Self-Play" (AAMAS 2023). [[paper](https://arxiv.org/abs/2302.07515)] [[videos](https://www.youtube.com/watch?v=U9REh0otmVU)].
<div align="center">
<img height="300px" height="auto" src="./docs/figures/football_trueskill.png">
</div>
### Installation
- Follow the instructions in [gfootball](https://github.com/google-research/football#on-your-computer) to set up the environment.
- `pip install gfootball openrl "openrl[selfplay]"`
- `pip install tizero` (or clone this repo and `pip install -e .`).
- test the installation by `python3 -m gfootball.play_game --action_set=full`.
### Evaluate JiDi submissions locally
You can evaluate your agent locally using tizero:
```bash
tizero eval --left_agent submission_dir1 --right_agent submission_dir2 --total_game 10
```
For example, you can evaluate tizero with random agent as below:
```bash
tizero eval --left_agent submission/tizero --right_agent submission/random_agent --total_game 10
```
For evaluations for JiDi submissions on other games, please refer to the [Arena](https://openrl-docs.readthedocs.io/en/latest/arena/index.html) of OpenRL
and this [example](https://github.com/OpenRL-Lab/openrl/tree/main/examples/snake) for the snake game.
### Show a saved dump file
- show detailed infomation of a match via: `tizero show dump_file`
- show keypoints of a mactch via: `tizero keypoint dump_file`
You can download an example dump file from [here](http://jidiai.cn/daily_6484285/daily_6484285.dump).
Then execute: `tizero show daily_6484285.dump` or `tizero keypoint daily_6484285.dump`. Then you will see a GUI as below:
<div align="center">
<img weight="300px" height="auto" src="./docs/_static/images/show_dump.png">
</div>
### Convert dump file to video
After the installation, you can use tizero to convert a dump file to a video file.
The usage is `tizero dump2video <dump_file> <output_dir> --episode_length <the length> --render_type <2d/3d>`.
You can download an example dump file from [here](http://jidiai.cn/daily_6484285/daily_6484285.dump).
And then execute `tizero dump2video daily_6484285.dump ./` in your terminal. By default, the episode length is 3000 and the render type is 2d.
Wait a minute, you will get a video file named `daily_6484285.avi` in your current directory.
### Submit TiZero to JIDI(及第评测平台)
<div align="center">
<img width="400px" height="auto" src="./docs/figures/jidi.png">
</div>
JIDI is a public evaluation platform for RL agents. You can submit your agent of GRF at: [http://www.jidiai.cn/env_detail?envid=34](http://www.jidiai.cn/env_detail?envid=34).
We provide several agents under `./submission/` directory, which can be submitted to JIDI directly:
- `./submission/tizero`: the final model of TiZero for JIDI submission, which ranked 1st on October 28th, 2022.
- `./submission/random_agent`: the random agent for JIDI submission.
### Cite
Please cite our paper if you use our codes or our weights in your own work:
```
@article{lin2023tizero,
title={TiZero: Mastering Multi-Agent Football with Curriculum Learning and Self-Play},
author={Lin, Fanqi and Huang, Shiyu and Pearce, Tim and Chen, Wenze and Tu, Wei-Wei},
journal={arXiv preprint arXiv:2302.07515},
year={2023}
}
```
Raw data
{
"_id": null,
"home_page": "https://github.com/OpenRL-Lab/TiZero",
"name": "tizero",
"maintainer": "",
"docs_url": null,
"requires_python": ">=3.8",
"maintainer_email": "",
"keywords": "reinforcement-learning multi-agent google research football",
"author": "openrl contributors",
"author_email": "huangsy1314@163.com",
"download_url": "https://files.pythonhosted.org/packages/04/e2/6d713d360d76e9504571d815125a033492ec0b9c5e4be89ecd5c799fd040/tizero-0.0.3.tar.gz",
"platform": null,
"description": "<div align=\"center\">\n<img width=\"300px\" height=\"auto\" src=\"./docs/figures/TiZero.png\">\n</div>\n\n<div align=\"center\">\n<img weight=\"300px\" height=\"auto\" src=\"./docs/figures/screen_800.png\">\n</div>\n\n[](https://opensource.org/licenses/Apache-2.0)\n[](https://pypi.org/project/tizero/)\n\n[](https://tizero.readthedocs.io/en/latest/?badge=latest)\n\n\n### Introduction\n\nReinforcement learning agent for Google Research Football.\n\nCode accompanying the paper \n\"TiZero: Mastering Multi-Agent Football with Curriculum Learning and Self-Play\" (AAMAS 2023). [[paper](https://arxiv.org/abs/2302.07515)] [[videos](https://www.youtube.com/watch?v=U9REh0otmVU)]. \n\n<div align=\"center\">\n<img height=\"300px\" height=\"auto\" src=\"./docs/figures/football_trueskill.png\">\n</div>\n\n### Installation\n\n- Follow the instructions in [gfootball](https://github.com/google-research/football#on-your-computer) to set up the environment.\n- `pip install gfootball openrl \"openrl[selfplay]\"`\n- `pip install tizero` (or clone this repo and `pip install -e .`).\n- test the installation by `python3 -m gfootball.play_game --action_set=full`.\n\n### Evaluate JiDi submissions locally\n\nYou can evaluate your agent locally using tizero:\n\n```bash\ntizero eval --left_agent submission_dir1 --right_agent submission_dir2 --total_game 10\n```\n\nFor example, you can evaluate tizero with random agent as below:\n\n```bash\ntizero eval --left_agent submission/tizero --right_agent submission/random_agent --total_game 10\n```\n\nFor evaluations for JiDi submissions on other games, please refer to the [Arena](https://openrl-docs.readthedocs.io/en/latest/arena/index.html) of OpenRL \nand this [example](https://github.com/OpenRL-Lab/openrl/tree/main/examples/snake) for the snake game.\n\n### Show a saved dump file\n\n- show detailed infomation of a match via: `tizero show dump_file`\n- show keypoints of a mactch via: `tizero keypoint dump_file`\n\nYou can download an example dump file from [here](http://jidiai.cn/daily_6484285/daily_6484285.dump). \n\nThen execute: `tizero show daily_6484285.dump` or `tizero keypoint daily_6484285.dump`. Then you will see a GUI as below:\n\n<div align=\"center\">\n<img weight=\"300px\" height=\"auto\" src=\"./docs/_static/images/show_dump.png\">\n</div>\n\n### Convert dump file to video\n\nAfter the installation, you can use tizero to convert a dump file to a video file.\nThe usage is `tizero dump2video <dump_file> <output_dir> --episode_length <the length> --render_type <2d/3d>`.\n\nYou can download an example dump file from [here](http://jidiai.cn/daily_6484285/daily_6484285.dump). \nAnd then execute `tizero dump2video daily_6484285.dump ./` in your terminal. By default, the episode length is 3000 and the render type is 2d.\nWait a minute, you will get a video file named `daily_6484285.avi` in your current directory.\n\n### Submit TiZero to JIDI(\u53ca\u7b2c\u8bc4\u6d4b\u5e73\u53f0)\n\n<div align=\"center\">\n<img width=\"400px\" height=\"auto\" src=\"./docs/figures/jidi.png\">\n</div>\n\n\nJIDI is a public evaluation platform for RL agents. You can submit your agent of GRF at: [http://www.jidiai.cn/env_detail?envid=34](http://www.jidiai.cn/env_detail?envid=34).\n\nWe provide several agents under `./submission/` directory, which can be submitted to JIDI directly:\n\n- `./submission/tizero`: the final model of TiZero for JIDI submission, which ranked 1st on October 28th, 2022.\n- `./submission/random_agent`: the random agent for JIDI submission.\n\n\n### Cite\n\nPlease cite our paper if you use our codes or our weights in your own work:\n\n```\n@article{lin2023tizero,\n title={TiZero: Mastering Multi-Agent Football with Curriculum Learning and Self-Play},\n author={Lin, Fanqi and Huang, Shiyu and Pearce, Tim and Chen, Wenze and Tu, Wei-Wei},\n journal={arXiv preprint arXiv:2302.07515},\n year={2023}\n}\n```\n",
"bugtrack_url": null,
"license": "",
"summary": "Toolkit and agents for Google Research Football",
"version": "0.0.3",
"project_urls": {
"Code": "https://github.com/OpenRL-Lab/TiZero",
"Documentation": "https://tizero.readthedocs.io/",
"Homepage": "https://github.com/OpenRL-Lab/TiZero"
},
"split_keywords": [
"reinforcement-learning",
"multi-agent",
"google",
"research",
"football"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "0ca524e6ec60de76b99b1a23ee59d6a76cc3d7021d2a631ccd1d74dcb6e73329",
"md5": "3d006cc86537ffd790a2552764f54ac7",
"sha256": "37e9b57ed26e60a18c5352eaa5ffb2901f96388952ee436630000e69157d7566"
},
"downloads": -1,
"filename": "tizero-0.0.3-py3-none-any.whl",
"has_sig": false,
"md5_digest": "3d006cc86537ffd790a2552764f54ac7",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.8",
"size": 89355,
"upload_time": "2023-09-01T09:59:41",
"upload_time_iso_8601": "2023-09-01T09:59:41.511860Z",
"url": "https://files.pythonhosted.org/packages/0c/a5/24e6ec60de76b99b1a23ee59d6a76cc3d7021d2a631ccd1d74dcb6e73329/tizero-0.0.3-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "04e26d713d360d76e9504571d815125a033492ec0b9c5e4be89ecd5c799fd040",
"md5": "0d8b9a1e850f144c458fb48799860212",
"sha256": "123f2d7ff2fbe143e7551637b73f62dd6b598134613e982aa1ffff032b0fa9e3"
},
"downloads": -1,
"filename": "tizero-0.0.3.tar.gz",
"has_sig": false,
"md5_digest": "0d8b9a1e850f144c458fb48799860212",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.8",
"size": 50100,
"upload_time": "2023-09-01T09:59:43",
"upload_time_iso_8601": "2023-09-01T09:59:43.082967Z",
"url": "https://files.pythonhosted.org/packages/04/e2/6d713d360d76e9504571d815125a033492ec0b9c5e4be89ecd5c799fd040/tizero-0.0.3.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2023-09-01 09:59:43",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "OpenRL-Lab",
"github_project": "TiZero",
"travis_ci": false,
"coveralls": false,
"github_actions": false,
"lcname": "tizero"
}
JSON
