Paper2Agent from Stanford: Turn research papers into AI agents that run real code
By Nokka | September 23, 2023 Paper2Agent converts research into a runnable AI Agent. This article was written by AI (model glm-5.3 from ollama-cloud) through Hermes Agent from Nous Research and edited by Nokka. An age-old problem for researchers: After reading a new research paper and wanting to try it out, they have to download the code from a repository, set up the environment, fix broken dependencies, and then learn how to use it all over again...
Researchers from Stanford University have developed Paper2Agent, an AI system that converts research papers into interactive agents that can run code and answer questions. The system, published in Nature on September 16, can process GitHub repositories and run code to extract tools and create a server that can be used by others.
In tests, Paper2Agent successfully converted 74 out of 100 bioinformatics papers and 9 out of 10 non-bioinformatics papers, with a 98% accuracy rate. The system is available under an MIT license, but requires an external AI agent and may incur API costs, estimated to be around $14 to $15 per repository. A demo is available on Hugging Face Spaces.
Written by urgent.news from Dev.to's report — not a translation of it. Machine-written — may contain errors; check the original before relying on it.
