One of the grand challenges of artificial intelligence is developing agents capable of conducting scientific research and discovering new knowledge. While frontier models have already been used to aid human scientists, e.g. for brainstorming ideas or writing code, they still require extensive manual supervision or are heavily constrained to a specific task.
We're excited to introduce The AI Scientist, the first comprehensive system for fully automatic scientific discovery, enabling Foundation Models such as Large Language Models (LLMs) to perform research independently.
We further provide all runs and data from our paper here, where we run each base model on each template for ~50 ideas. We highly recommend reading through some of the Claude papers, (especially the diffusion ones), to get a sense of its strengths and weaknesses. Here are some example papers generated by The AI Scientist 📝:
Note: Caution! This codebase will execute LLM-written code. There are various risks and challenges associated with this autonomy. This includes e.g. the use of potentially dangerous packages, web access, and potential spawning of processes. Use at your own discretion. Please make sure to containerize and restrict web access appropriately.
This code was designed for NVIDIA GPUs with CUDA using PyTorch. Support for other GPU architectures may be possible by following PyTorch guidelines. Current templates would likely take an infeasible amount of time on CPU-only machines. All code is designed to be run on Linux, other operating systems will likely require major adjustments.
When installing texlive-full, you may need to hold Enter.
Supported Models and API Keys
We support a wide variety of models including open-weight and API-only models. In general, we recommend only using frontier models above the capability of the original GPT-4.
OpenAI API (GPT-4)
By default, this uses the OPENAI_API_KEY environment variable.
Anthropic API (Claude Sonnet 3.5)
By default, this uses the ANTHROPIC_API_KEY environment variable.
Claude models via Bedrock
For Claude models provided by Amazon Bedrock, please install these additional packages:
Next, set up a valid authentication for a Google Cloud project, for example by providing region and project ID like so:
bash
1exportCLOUD_ML_REGION="REGION"# for Model Garden call2exportANTHROPIC_VERTEX_PROJECT_ID="PROJECT_ID"# for Model Garden call3exportVERTEXAI_LOCATION="REGION"# for Aider/LiteLLM call, as per https://docs.litellm.ai/docs/providers/vertex#set-vertex-project--vertex-location4exportVERTEXAI_PROJECT="PROJECT_ID"# for Aider/LiteLLM call as per https://docs.litellm.ai/docs/providers/vertex#set-vertex-project--vertex-location
DeepSeek API (DeepSeek-Coder-V2)
By default, this uses the DEEPSEEK_API_KEY environment variable.
OpenRouter API (Llama3.1)
By default, this uses the OPENROUTER_API_KEY environment variable.
Semantic Scholar API (Literature Search)
Our code can also optionally use a Semantic Scholar API Key (S2_API_KEY) for higher throughput if you have one, though in principle it should work without it.
Be sure to provide the key for the model used for your runs, e.g.
Here, and below, we give instructions for setting up the data and baseline evaluations for each template. You can only run setup steps for templates you are interested in. This is necessary to run on your machine as training times may vary depending on your hardware.
1# Set up NanoGPT baseline run2# NOTE: YOU MUST FIRST RUN THE PREPARE SCRIPTS ABOVE!3cd templates/nanoGPT && python experiment.py --out_dir run_0 && python plot.py
Create NanoGPT_lite baseline run. We use this for sanity-checking
bash
1# NOTE: YOU MUST FIRST RUN THE PREPARE SCRIPTS ABOVE!2cd templates/nanoGPT_lite && python experiment.py --out_dir run_0 && python plot.py
Setup 2D Diffusion
bash
1# Set up 2D Diffusion2git clone https://github.com/gregversteeg/NPEET.git
3cd NPEET
4pip install.5pip install scikit-learn
67# Set up 2D Diffusion baseline run8cd templates/2d_diffusion && python experiment.py --out_dir run_0 && python plot.py
Setup Grokking
bash
1# Set up Grokking2pip install einops
34# Set up Grokking baseline run5cd templates/grokking && python experiment.py --out_dir run_0 && python plot.py
Run AI Scientist Paper Generation Experiments
Note: please ensure the setup steps above are completed.
bash
1conda activate ai_scientist
2# Run the paper generation.3python launch_scientist.py --model "gpt-4o-2024-05-13" --experiment nanoGPT_lite --num-ideas 24python launch_scientist.py --model "claude-3-5-sonnet-20240620" --experiment nanoGPT_lite --num-ideas 25python launch_scientist.py --model "ollama/mistral-nemo" --experiment nanoGPT_lite --num-ideas 2
If you have more than 1 GPU, use the parallel option to parallelize ideas across multiple GPUs.
Getting an LLM Generated Paper Review
python
1import openai
2from ai_scientist.perform_review import load_paper, perform_review
34client = openai.OpenAI()5model ="gpt-4o-2024-05-13"67# Load paper from pdf file (raw text)8paper_txt = load_paper("report.pdf")9# Get the review dict of the review10review = perform_review(11 paper_txt,12 model,13 client,14 num_reflections=5,15 num_fs_examples=1,16 num_reviews_ensemble=5,17 temperature=0.1,18)1920# Inspect review results21review["Overall"]# overall score 1-1022review["Decision"]# ['Accept', 'Reject']23review["Weaknesses"]# List of weaknesses (str)
If there is an area of study you would like The AI Scientist to explore, it should be very easy to create your own templates. In general, follow the structure of the existing templates, which consists of:
experiment.py -- This is a single file where the 'meat' of the content is. It takes in an argument for out_dir, which is where it should create the folder and save the relevant information from the run.
plot.py -- This should take in the information from the run folders and create plots. The code should be clear and easy to edit.
prompt.json -- Put information about your template here.
seed_ideas.json -- Put example ideas here. You can also try to generate ideas without any examples, and then pick the best one or two to put here.
latex/template.tex -- We recommend using our latex folder, but be sure to replace the pre-loaded citations with ones that you would expect to be more relevant.
Template Resources
We provide 3 templates, which heavily use code from other repositories, which we credit below. (Normally, we would do this in the files themselves, but it's unclear how this would affect The AI Scientist since it would be visible).
The NanoGPT template used code from NanoGPT and this PR.
We would like to thank the developers of the open-source models and packages for their contributions and for making their work available.
Citing The AI Scientist
If you use The AI Scientist in your research, please cite it as follows:
@article{lu2024aiscientist,
title={The {AI} {S}cientist: Towards Fully Automated Open-Ended Scientific Discovery},
author={Lu, Chris and Lu, Cong and Lange, Robert Tjarko and Foerster, Jakob and Clune, Jeff and Ha, David},
journal={arXiv preprint arXiv:2408.06292},
year={2024}
}
FAQ
We recommend reading our paper in the first instance for any questions you have on The AI Scientist.
Why am I missing files when running The AI Scientist?
Make sure you have completed all the setup and preparation steps before the main experiment script.
Why has a PDF or a review not been generated?
The AI Scientist finishes an idea with a success rate that depends on both the template, the base foundation model, and the complexity of the idea. We advise referring to our main paper. The highest success rates are observed with Claude Sonnet 3.5.
Reviews are best done with GPT-4o, all other models have issues with positivity bias or failure to conform to required outputs.
What is the cost of each idea generated?
Typically less than $15 per paper with Claude Sonnet 3.5. We recommend DeepSeek Coder V2 for a much more cost-effective approach. A good place to look for new models is the Aider leaderboard.
How do I change the base conference format associated with the write-ups?
Change the base template.tex files contained within each template.
How do I run The AI Scientist for different subject fields?
Please refer to the instructions for different templates. In this current iteration, this is restricted to ideas that can be expressed in code. However, lifting this restriction would represent exciting future work! :)
How do I add support for a new foundation model?
Please see this PR for an example of how to add a new model, e.g. this time for Claude via Bedrock.
We do not advise any model that is significantly weaker than GPT-4 level for The AI Scientist.
Why do I need to run the baseline runs myself?
These appear as run_0 and should be run per machine you execute The AI Scientist on for accurate run-time comparisons due to hardware differences.
Containerization
We include a community-contributed Docker image that may assist with your containerization efforts in experimental/Dockerfile.
