https://github.com/sileod/llm-theory-of-mind

Testing Theory of Mind (ToM) in language models with epistemic logic
https://github.com/sileod/llm-theory-of-mind

chatgpt dataset english epistemic-logic epistemic-reasoning gpt-4 language-model llm modal-logic muddy-children natural-language natural-language-inference natural-language-processing nli nlp pal public-announcement-logic social-reasoning theory-of-mind tom

Last synced: 3 days ago
JSON representation

Testing Theory of Mind (ToM) in language models with epistemic logic

• Host: GitHub
• URL: https://github.com/sileod/llm-theory-of-mind
• Owner: sileod
• License: apache-2.0
• Created: 2023-01-25T15:46:22.000Z (almost 2 years ago)
• Default Branch: master
• Last Pushed: 2023-12-13T10:22:18.000Z (about 1 year ago)
• Last Synced: 2023-12-13T11:34:20.813Z (about 1 year ago)
• Topics: chatgpt, dataset, english, epistemic-logic, epistemic-reasoning, gpt-4, language-model, llm, modal-logic, muddy-children, natural-language, natural-language-inference, natural-language-processing, nli, nlp, pal, public-announcement-logic, social-reasoning, theory-of-mind, tom
• Language: Python
• Homepage:
• Size: 610 KB
• Stars: 11
• Watchers: 4
• Forks: 2
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• awesome_ai_agents - Llm-Theory-Of-Mind - Testing Theory of Mind (ToM) in language models with epistemic logic (Building / Datasets)
• awesome_ai_agents - Llm-Theory-Of-Mind - Testing Theory of Mind (ToM) in language models with epistemic logic (Building / Datasets)

README

        
# MindGames

Code associated to the MindGames article

We generate epistemic reasoning problems (akin to muddy children or drinking logicians) using modal logic to target theory of mind (tom) in natural language processing models.

🤗 https://hf.co/datasets/sileod/mindgames

To appear at EMNLP 2023 Findings !

```

@article{sileo2023mindgames,

  title={MindGames: Targeting Theory of Mind in Large Language Models with Dynamic Epistemic Modal Logic},

  author={Sileo, Damien and Lernould, Antoine},

  journal={arXiv preprint arXiv:2305.03353},

  year={2023}

}

```

Contact (interested in logic+NLP colabs) [email protected]