quester-one / smartplay

SmartPlay is a benchmark for Large Language Models (LLMs). Uses a variety of games to test various important LLM capabilities as agents. SmartPlay is designed to be easy to use, and to support future development of LLMs.

SmartPlay

The SmartPlay repository is a benchmark and methodology for evaluating the abilities of large language models (LLMs) as agents. It consists of six different games, including Rock-Paper-Scissors, Tower of Hanoi, and Minecraft, each featuring a unique setting that provides up to 20 evaluation settings and infinite environment variations. The games in SmartPlay challenge a subset of nine important capabilities of an intelligent LLM agent, including reasoning with object dependencies, planning ahead, spatial reasoning, learning from history, and understanding randomness. The distinction between the set of capabilities each game tests allows for the analysis of each capability separately. SmartPlay serves as a rigorous testing ground for evaluating the overall performance of LLM agents and as a roadmap for identifying gaps in current methodologies.

Currently included games are:

• Rock-Paper-Scissors
• 2-armed bandit
• Tower of Hanoi
• Messenger-EMMA
• Crafter
• MineDojo Creative Tasks

For more information, please refer to the paper.

Table of Contents

• Introduction
• Games in SmartPlay
• Getting Started
• Using SmartPlay
• Citing SmartPlay
• Contributing
• Legal Notices

Games in SmartPlay

Getting Started

First consider setting up a conda environment by running

conda env create --name SmartPlay --file environment.yml

SmartPlay requires MineDojo, please follow the official documentation to install MineDojo first before proceeding.

Then run

pip install -e .

For completeness, we also provide conda environment scripts and requirements.txt in the root directory.

Using SmartPlay

Guidelines to use the benchmark are provided in:

examples/experiments.py

To see all games available in the SmartPlay benchmark, run the following code:

import smartplay
print(smartplay.env_list)

See MineDojo Documentation for a description of the MineDojo Creative tasks.

Citing SmartPlay

@misc{wu2023smartplay,
      title={SmartPlay : A Benchmark for LLMs as Intelligent Agents}, 
      author={Yue Wu and Xuan Tang and Tom M. Mitchell and Yuanzhi Li},
      year={2023},
      eprint={2310.01557},
      archivePrefix={arXiv},
      primaryClass={cs.LG}
}

Contributing

This project welcomes contributions and suggestions. Most contributions require you to agree to a Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us the rights to use your contribution. For details, visit https://cla.opensource.microsoft.com.

When you submit a pull request, a CLA bot will automatically determine whether you need to provide a CLA and decorate the PR appropriately (e.g., status check, comment). Simply follow the instructions provided by the bot. You will only need to do this once across all repos using our CLA.

This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact [email protected] with any additional questions or comments.

Legal Notices

Microsoft and any contributors grant you a license to the Microsoft documentation and other content in this repository under the Creative Commons Attribution 4.0 International Public License, see the LICENSE file, and grant you a license to any code in the repository under the MIT License, see the LICENSE-CODE file.

Microsoft, Windows, Microsoft Azure and/or other Microsoft products and services referenced in the documentation may be either trademarks or registered trademarks of Microsoft in the United States and/or other countries. The licenses for this project do not grant you rights to use any Microsoft names, logos, or trademarks. Microsoft's general trademark guidelines can be found at http://go.microsoft.com/fwlink/?LinkID=254653.

Privacy information can be found at https://privacy.microsoft.com/en-us/

Microsoft and any contributors reserve all other rights, whether under their respective copyrights, patents, or trademarks, whether by implication, estoppel or otherwise.