AutoToM: Scaling Model-based Mental Inference via Automated Agent Modeling

Paper | Project Page | Tweet

AutoToM is an automated agent modeling method for scalable, robust, and interpretable mental inference. It achieves SOTA on five benchmarks, produces human-like confidence estimates, and supports embodied decision-making.

Example Usage

To run AutoToM on MMToM-QA, with the default settings of reduced hypotheses and backwards inference:

python ProbSolver.py --automated --dataset_name "MMToM-QA"

To run AutoToM on ToMi-1st with a specified model input:

python ProbSolver.py --dataset_name "ToMi-1st" --assigned_model "['State', 'Observation', 'Belief']"

Requirements

• Install relevant packages:

  • run pip install -r requirements.txt

• Set your OPENAI_API_KEY:

  • On macOS and Linux: export OPENAI_API_KEY='your-api-key'

  • On Windows: set OPENAI_API_KEY='your-api-key'

Testing AutoToM with customized questions

Please check out playground.ipynb. Simply replace the story and choices with your customized input to see how AutoToM discover Bayesian models and conduct inverse planning!

Citation

Please cite the paper and star this repo if you find it useful, thanks!

@article{zhang2025autotom,
  title={AutoToM: Automated Bayesian Inverse Planning and Model Discovery for Open-ended Theory of Mind},
  author={Zhang, Zhining and Jin, Chuanyang and Jia, Mung Yao and Shu, Tianmin},
  journal={arXiv preprint arXiv:2502.15676},
  year={2025}
}