[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-74129":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":10,"languages":9,"totalLinesOfCode":9,"stars":11,"forks":12,"watchers":13,"openIssues":14,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":14,"stars7d":16,"stars30d":17,"stars90d":15,"forks30d":15,"starsTrendScore":18,"compositeScore":19,"rankGlobal":9,"rankLanguage":9,"license":20,"archived":21,"fork":21,"defaultBranch":22,"hasWiki":21,"hasPages":21,"topics":23,"createdAt":9,"pushedAt":9,"updatedAt":24,"readmeContent":25,"aiSummary":26,"trendingCount":15,"starSnapshotCount":15,"syncStatus":27,"lastSyncTime":28,"discoverSource":29},74129,"HyperAgents","facebookresearch\u002FHyperAgents","facebookresearch","Self-referential self-improving agents that can optimize for any computable task",null,"Python",2569,332,29,11,0,23,104,33,29.57,"Other",false,"main",[],"2026-06-12 02:03:22","\u003Cdiv align=\"center\">\n\n\u003C!-- Logo\u002FBanner placeholder - uncomment and add your image -->\n\u003C!-- \u003Cimg src=\"assets\u002Fbanner.png\" alt=\"HyperAgents Banner\" width=\"800\"> -->\n\n\u003Ch1>HyperAgents\u003C\u002Fh1>\n\n\u003Cp>Self-referential self-improving agents that can optimize for any computable task\u003C\u002Fp>\n\n\u003Cp>\n\u003Ca href=\"LICENSE.md\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-CC%20BY--NC--SA%204.0-lightgrey.svg?style=for-the-badge\" alt=\"License: CC BY-NC-SA 4.0\">\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.19461\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FarXiv-2603.19461-b31b1b.svg?style=for-the-badge&logo=arxiv\" alt=\"arXiv\">\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fai.meta.com\u002Fresearch\u002Fpublications\u002Fhyperagents\u002F\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F-Blog-%238D6748?style=for-the-badge&logo=Website&logoColor=white\">\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fx.com\u002Fjennyzhangzt\u002Fstatus\u002F2036099935083618487\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Ftwitter-%230077B5.svg?&style=for-the-badge&logo=twitter&logoColor=white&color=00acee\">\u003C\u002Fa>\n\u003C\u002Fp>\n\n---\n\n\u003C\u002Fdiv>\n\n## Setup\n```bash\n# API keys, put these into .env file\nOPENAI_API_KEY=...\nANTHROPIC_API_KEY=...\nGEMINI_API_KEY=...\n```\n\n```bash\n# Install things\nsudo dnf install -y python3.12-devel\nsudo dnf install -y graphviz graphviz-devel cmake ninja-build bzip2-devel zlib-devel ncurses-devel libffi-devel\n```\n\n```bash\n# Create virtual environment\npython3.12 -m venv venv_nat\nsource venv_nat\u002Fbin\u002Factivate\npip install -r requirements.txt\npip install -r requirements_dev.txt\n# To build the docker container\ndocker build --network=host -t hyperagents .\n```\n\n```bash\n# Setup initial agents\nbash .\u002Fsetup_initial.sh\n```\n\n## Running HyperAgents\n\n```bash\n# See the script for args, and baseline selections\npython generate_loop.py --domains \u003Cdomain>\n```\n\nBy default, outputs will be saved in `outputs\u002F` directory.\n\n## File Structure\n- `agent\u002F` code for using foundation models\n- `analysis\u002F` scripts used for plotting and analysis\n- `domains\u002F` code for each domain\n- `utils\u002F` common code used in the repo\n- `run_meta_agent.py` script to help run the meta agent and get the diffs\n- `meta_agent.py` main implementation of the meta agent\n- `task_agent.py` main implementation of the task agent\n- `generate_loop.py` entry point for running the algorithm\n\n## Logs from Experiments\n\nThe experiment logs can be downloaded here: https:\u002F\u002Fdrive.google.com\u002Fdrive\u002Ffolders\u002F164fKQWgLM18foOzSnpv0F_I3TNpX8u8-?usp=sharing\n\n## Safety Consideration\n> [!WARNING]  \n> This repository involves executing untrusted, model-generated code. We strongly advise users to be aware of the associated safety risks. While it is highly unlikely that such code will perform overtly malicious actions under our current settings and with the models we use, it may still behave destructively due to limitations in model capability or alignment. By using this repository, you acknowledge and accept these risks.\n\n## Citing\nIf you find this project useful, please consider citing:\n```bibtex\n@misc{zhang2026hyperagents,\n      title={Hyperagents}, \n      author={Jenny Zhang and Bingchen Zhao and Wannan Yang and Jakob Foerster and Jeff Clune and Minqi Jiang and Sam Devlin and Tatiana Shavrina},\n      year={2026},\n      eprint={2603.19461},\n      archivePrefix={arXiv},\n      primaryClass={cs.AI},\n      url={https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.19461}, \n}\n```\n\n","HyperAgents 是一个能够自我改进并优化任何可计算任务的自引用代理系统。项目采用 Python 语言开发，通过利用先进的基础模型实现智能体之间的交互与迭代学习，以达到持续提升性能的目的。其核心功能包括支持多种领域的任务优化、基于元智能体和任务智能体的双层架构设计以及灵活的任务领域扩展能力。适用于需要高度定制化解决方案的研究场景或复杂问题求解环境，如自动编程、游戏策略生成等。此外，项目提供了详尽的实验记录及分析工具，便于研究者深入探究智能体的行为模式与进化路径。",2,"2026-06-11 03:48:55","high_star"]