OpenAGI: When LLM Meets Domain Experts
About OpenAGI
OpenAGI is a research platform from Rutgers University, in the paper OpenAGI: When LLM Meets Domain Experts, which appeared in the datasets and benchmarks track of NeurIPS 2023. Its authors are Yingqiang Ge, Wenyue Hua, Kai Mei, Jianchao Ji, Juntao Tan, Shuyuan Xu, Zelong Li and Yongfeng Zhang.
The idea is that a language model takes a task described in natural language, chooses suitable expert models and runs them. The authors propose Reinforcement Learning from Task Feedback, which uses task results to improve the model's planning in a self-improving loop. The platform combines standard benchmark tasks with open-ended tasks that can be extended with new models, tools, plugins or APIs.
The code, dataset, benchmarks, evaluation methods and a demo interface are released at github.com/agiresearch/OpenAGI.
Key features
- Language model selects and runs expert models
- Reinforcement Learning from Task Feedback
- Benchmark and open-ended tasks
- Extensible with new models, tools and APIs
- NeurIPS 2023 paper
- Open-source code and demo
Links
More from AGI Research at Rutgers University
AIOS: AI Agent Operating System
All content and agents from AGI Research at Rutgers University
Frequently asked questions about OpenAGI
- What is OpenAGI?
- OpenAGI: When LLM Meets Domain Experts. It is listed on botshub under autonomous agent projects.
- Who makes OpenAGI?
- OpenAGI is made by AGI Research at Rutgers University.
- Is OpenAGI free or open source?
- OpenAGI is tagged as open source on botshub. Check the official site for current terms.
- What can OpenAGI do?
- Key features include language model selects and runs expert models, reinforcement Learning from Task Feedback, benchmark and open-ended tasks, extensible with new models, tools and APIs and neurIPS 2023 paper.
- What are alternatives to OpenAGI?
- Other autonomous agent projects on botshub include AutoGPT, Mini AGI, Multi GPT and OpenAgents. See all alternatives to OpenAGI.
