Imagen de Portada

Dissecting the Ullman Variations with a SCALPEL: Why do LLMs fail at Trivial Alterations to the False Belief Task?

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Pi, Zhiqiang, Vadaparty, Annapurna, Bergen, Benjamin K., Jones, Cameron R.
Formato:	Preprint
Publicado:	2024
Materias:	Computation and Language
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Large Language Models Pass the Turing Test
por: Jones, Cameron R., et al.
Publicado: (2025)

Does GPT-4 pass the Turing test?
por: Jones, Cameron R., et al.
Publicado: (2023)

Why do language models perform worse for morphologically complex languages?
por: Arnett, Catherine, et al.
Publicado: (2024)

Lies, Damned Lies, and Distributional Language Statistics: Persuasion and Deception with Large Language Models
por: Jones, Cameron R., et al.
Publicado: (2024)

LLMs and people both learn to form conventions -- just not with each other
por: Jones, Cameron R., et al.
Publicado: (2026)

GPT-4 is judged more human than humans in displaced and inverted Turing tests
por: Rathi, Ishika, et al.
Publicado: (2024)

Language Statistics and False Belief Reasoning: Evidence from 41 Open-Weight LMs
por: Trott, Sean, et al.
Publicado: (2026)

People cannot distinguish GPT-4 from a human in a Turing test
por: Jones, Cameron R., et al.
Publicado: (2024)

Bigram Subnetworks: Mapping to Next Tokens in Transformer Language Models
por: Chang, Tyler A., et al.
Publicado: (2025)

Do Large Language Models Exhibit Spontaneous Rational Deception?
por: Taylor, Samuel M., et al.
Publicado: (2025)

Emergent inabilities? Inverse scaling over the course of pretraining
por: Michaelov, James A., et al.
Publicado: (2023)

EVOKE: Emotion Vocabulary Of Korean and English
por: Jung, Yoonwon, et al.
Publicado: (2026)

Triviality Corrected Endogenous Reward
por: Wang, Xinda, et al.
Publicado: (2026)

The Dynamics of Delusion: Modeling Bidirectional False Belief Amplification in Human-Chatbot Dialogue
por: Mehta, Ashish, et al.
Publicado: (2026)

Dissecting Multiplication in Transformers: Insights into LLMs
por: Qiu, Luyu, et al.
Publicado: (2024)

Accommodation and Epistemic Vigilance: A Pragmatic Account of Why LLMs Fail to Challenge Harmful Beliefs
por: Cheng, Myra, et al.
Publicado: (2026)

Revenge of the Fallen? Recurrent Models Match Transformers at Predicting Human Language Comprehension Metrics
por: Michaelov, James A., et al.
Publicado: (2024)

A Bit of a Problem: Measurement Disparities in Dataset Sizes Across Languages
por: Arnett, Catherine, et al.
Publicado: (2024)

Characterizing Learning Curves During Language Model Pre-Training: Learning, Forgetting, and Stability
por: Chang, Tyler A., et al.
Publicado: (2023)

SCALPEL: Selective Capability Ablation via Low-rank Parameter Editing for Large Language Model Interpretability Analysis
por: Fu, Zihao, et al.
Publicado: (2026)

On the Limitations of Large Language Models (LLMs): False Attribution
por: Adewumi, Tosin, et al.
Publicado: (2024)

Language Model Behavioral Phases are Consistent Across Architecture, Training Data, and Scale
por: Michaelov, James A., et al.
Publicado: (2025)

Why Do LLMs Struggle in Strategic Play? Broken Links Between Observations, Beliefs, and Actions
por: Sobotka, Jan, et al.
Publicado: (2026)

Why do LLMs attend to the first token?
por: Barbero, Federico, et al.
Publicado: (2025)

Not quite Sherlock Holmes: Language model predictions do not reliably differentiate impossible from improbable events
por: Michaelov, James A., et al.
Publicado: (2025)

Quiet Feature Learning in Algorithmic Tasks
por: Naidu, Prudhviraj, et al.
Publicado: (2025)

False Sense of Security: Why Probing-based Malicious Input Detection Fails to Generalize
por: Wang, Cheng, et al.
Publicado: (2025)

Assessing Problem Decomposition in CS1 for the GenAI Era
por: Srinath, Samvrit, et al.
Publicado: (2025)

Goldfish: Monolingual Language Models for 350 Languages
por: Chang, Tyler A., et al.
Publicado: (2024)

Explaining and Mitigating Crosslingual Tokenizer Inequities
por: Arnett, Catherine, et al.
Publicado: (2025)

Pearmut: Human Evaluation of Translation Made Trivial
por: Zouhar, Vilém, et al.
Publicado: (2026)

CS1-LLM: Integrating LLMs into CS1 Instruction
por: Vadaparty, Annapurna, et al.
Publicado: (2024)

LLMs Struggle to Reject False Presuppositions when Misinformation Stakes are High
por: Sieker, Judith, et al.
Publicado: (2025)

TagRouter: Learning Route to LLMs through Tags for Open-Domain Text Generation Tasks
por: Chen, Zhou, et al.
Publicado: (2025)

Vulnerability of LLMs' Stated Beliefs? LLMs Belief Resistance Check Through Strategic Persuasive Conversation Interventions
por: Huang, Fan, et al.
Publicado: (2026)

Zero-Shot Belief: A Hard Problem for LLMs
por: Murzaku, John, et al.
Publicado: (2025)

On the Acquisition of Shared Grammatical Representations in Bilingual Language Models
por: Arnett, Catherine, et al.
Publicado: (2025)

Drawing Your Programs: Exploring the Applications of Visual-Prompting with GenAI for Teaching and Assessment
por: Smith IV, David H., et al.
Publicado: (2026)

Dissecting Bias in LLMs: A Mechanistic Interpretability Perspective
por: Chandna, Bhavik, et al.
Publicado: (2025)

Rectifying Belief Space via Unlearning to Harness LLMs' Reasoning
por: Niwa, Ayana, et al.
Publicado: (2025)