SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2402.15052 · ACL · 2024

T MBENCH: Benchmarking Theory of Mind in Large Language Models

Minlie Huang, Bosi Wen, Jinfeng Zhou, Jincenzi Wu, Mengting Hu, Zhuang Chen, Gongyao Jiang, Guanqun Bi, Yaru Cao, Yunghwei Lai, Zexuan Xiong

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 2 functions out of this paper's own repositories and ran 2 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
zhchen18/tombench canonical 2 of 2
FunctionStatusWhere it lives
extract_answer Ran zhchen18/tombench/get_results.py
code served (permissive licence) · get_code("048418fb06a97048")
most_common_element Ran zhchen18/tombench/get_results.py
code served (permissive licence) · get_code("fb24308f1c903d0a")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Theory of Mind (ToM) is the cognitive capability to perceive and ascribe mental states to oneself and others. Recent research has sparked a debate over whether large language models (LLMs) exhibit a form of ToM. However, existing ToM evaluations are hindered by challenges such as constrained scope, subjective judgment, and unintended contamination, yielding inadequate assessments. To address this gap, we introduce T MBENCH with three key characteristics: a systematic evaluation framework encompassing 8 tasks and 31 abilities in social cognition, a multiple-choice question format to support automated and unbiased evaluation, and a build-from-scratch bilingual inventory to strictly avoid data leakage. Based on T MBENCH, we conduct extensive experiments to evaluate the ToM performance of 10 popular LLMs across tasks and abilities. We find that even the most advanced LLMs like GPT-4 lag behind human performance by over 10% points, indicating that LLMs have not achieved a human-level theory of mind yet. Our aim with T MBENCH is to enable an efficient and effective evaluation of LLMs' ToM capabilities, thereby facilitating the development of LLMs with inherent social intelligence.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2402.15052")
get_code_for_paper("2402.15052")
have("2402.15052")

Connect an agent — have() is free.