Copertina

Not quite Sherlock Holmes: Language model predictions do not reliably differentiate impossible from improbable events

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Michaelov, James A., Estacio, Reeka, Zhang, Zhien, Bergen, Benjamin K.
Natura:	Preprint
Pubblicazione:	2025
Soggetti:	Computation and Language Artificial Intelligence
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

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

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

How Open Must Language Models be to Enable Reliable Scientific Inference?
di: Michaelov, James A., et al.
Pubblicazione: (2026)

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

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

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

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

Kallini et al. (2024) do not compare impossible languages with constituency-based ones
di: Hunter, Tim
Pubblicazione: (2024)

A novel language model for predicting serious adverse event results in clinical trials from their prospective registrations
di: Hu, Qixuan, et al.
Pubblicazione: (2025)

Design of reliable technology valuation model with calibrated machine learning of patent indicators
di: Lee, Seunghyun, et al.
Pubblicazione: (2024)

Supervising the search process produces reliable and generalizable information-seeking agents
di: Xiong, Guangzhi, et al.
Pubblicazione: (2025)

Improving Large Language Model (LLM) fidelity through context-aware grounding: A systematic approach to reliability and veracity
di: Talukdar, Wrick, et al.
Pubblicazione: (2024)

N-gram-like Language Models Predict Reading Time Best
di: Michaelov, James A., et al.
Pubblicazione: (2026)

Language models are better than humans at next-token prediction
di: Shlegeris, Buck, et al.
Pubblicazione: (2022)

Training language models to be warm and empathetic makes them less reliable and more sycophantic
di: Ibrahim, Lujain, et al.
Pubblicazione: (2025)

LeoTask: a fast, flexible and reliable framework for computational research
di: Zhang, Changwang, et al.
Pubblicazione: (2015)

Dissociation of Faithful and Unfaithful Reasoning in LLMs
di: Yee, Evelyn, et al.
Pubblicazione: (2024)

Is GPT-4 a reliable rater? Evaluating Consistency in GPT-4 Text Ratings
di: Hackl, Veronika, et al.
Pubblicazione: (2023)

Disaggregation Reveals Hidden Training Dynamics: The Case of Agreement Attraction
di: Michaelov, James A., et al.
Pubblicazione: (2025)

It's All in The [MASK]: Simple Instruction-Tuning Enables BERT-like Masked Language Models As Generative Classifiers
di: Clavié, Benjamin, et al.
Pubblicazione: (2025)

Ploutos: Towards interpretable stock movement prediction with financial large language model
di: Tong, Hanshuang, et al.
Pubblicazione: (2024)

When is the consistent prediction likely to be a correct prediction?
di: Nguyen, Alex, et al.
Pubblicazione: (2024)

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

Surrogate modeling for interpreting black-box LLMs in medical predictions
di: Han, Changho, et al.
Pubblicazione: (2026)

Sustaining AI safety: Control-theoretic external impossibility, intrinsic necessity, and structural requirements
di: Mazzu, James M.
Pubblicazione: (2026)

On the Representations of Entities in Auto-regressive Large Language Models
di: Morand, Victor, et al.
Pubblicazione: (2025)

KICGPT: Large Language Model with Knowledge in Context for Knowledge Graph Completion
di: Wei, Yanbin, et al.
Pubblicazione: (2024)

Language Model Memory and Memory Models for Language
di: Badger, Benjamin L.
Pubblicazione: (2026)

CT Open: An Open-Access, Uncontaminated, Live Platform for the Open Challenge of Clinical Trial Outcome Prediction
di: Wang, Jianyou, et al.
Pubblicazione: (2026)

How do Large Language Models Handle Multilingualism?
di: Zhao, Yiran, et al.
Pubblicazione: (2024)

Revealing emergent human-like conceptual representations from language prediction
di: Xu, Ningyu, et al.
Pubblicazione: (2025)

What is the best model? Application-driven Evaluation for Large Language Models
di: Lian, Shiguo, et al.
Pubblicazione: (2024)

Modeling Language as a Sequence of Thoughts
di: Borazjanizadeh, Nasim, et al.
Pubblicazione: (2025)

Direct Alignment of Language Models via Quality-Aware Self-Refinement
di: Yu, Runsheng, et al.
Pubblicazione: (2024)

Token-Level Uncertainty-Aware Objective for Language Model Post-Training
di: Liu, Tingkai, et al.
Pubblicazione: (2025)

MAD-Sherlock: Multi-Agent Debate for Visual Misinformation Detection
di: Lakara, Kumud, et al.
Pubblicazione: (2024)

Emotions Where Art Thou: Understanding and Characterizing the Emotional Latent Space of Large Language Models
di: Reichman, Benjamin, et al.
Pubblicazione: (2025)

Language models show human-like content effects on reasoning tasks
di: Dasgupta, Ishita, et al.
Pubblicazione: (2022)

On the Effectiveness of Incremental Training of Large Language Models
di: Li, Miles Q., et al.
Pubblicazione: (2024)

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