paper-with-me

Papers

Number Representations in LLMs: A Computational Parallel to Human Perception

2025-02-22 · H. V. AlquBoj, Hilal AlQuabeh, Velibor Bojkovic, Tatsuya Hiraoka, Ahmed Oumar El-Shangiti, Munachiso Nwadike, Kentaro Inui

Humans are believed to perceive numbers on a logarithmic mental number line, where smaller values are represented with greater resolution than larger ones. This cognitive bias, supported by neuroscience and behavioral studies, suggests that numerical magnitudes are processed in a sublinear fashion rather than on a uniform linear scale. Inspired by this hypothesis, we investigate whether large language models (LLMs) exhibit a similar logarithmic-like structure in their internal numerical representations. By analyzing how numerical values are encoded across different layers of LLMs, we apply dimensionality reduction techniques such as PCA and PLS followed by geometric regression to uncover latent structures in the learned embeddings. Our findings reveal that the model's numerical representations exhibit sublinear spacing, with distances between values aligning with a logarithmic scale. This suggests that LLMs, much like humans, may encode numbers in a compressed, non-uniform manner.

📄 PDF Abstract BibTeX arXiv:2502.16147

Code (1)

halquabeh/llm_natural_log 공식 구현 pytorch

Tasks

Dimensionality Reduction

Methods 이 논문이 사용한 방법론

PCA Principle Components Analysis (PCA) is an unsupervised method primary used for dimensionality reduction within machine learning. PCA is calculated via a singular value…

Similar Papers 제목 키워드 기반

Human Behavioral Benchmarking: Numeric Magnitude Comparison Effects in Large Language Models

2023-05-18 · Raj Sanjay Shah, Vijay Marupudi, Reba Koenen, Khushi Bhardwaj 외

Large Language Models (LLMs) do not differentially represent numbers, which are pervasive in text. In contrast, neuroscience research has identified distinct neural representations for numbers and words. In this work, we…

Benchmarking

Language models struggle with compartmentalization

2026-05-19 · Thomas Vincent Howe, David Wingate arxiv

In the training data used by large language models (LLMs), the same latent concept is often presented in multiple distinct ways: the same facts appear in English and Swahili; many functions can be expressed in both Pytho…

ParallelPARC: A Scalable Pipeline for Generating Natural-Language Analogies

2024-03-02 · Oren Sultan, Yonatan Bitton, Ron Yosef, Dafna Shahaf

Analogy-making is central to human cognition, allowing us to adapt to novel situations -- an ability that current AI systems still lack. Most analogy datasets today focus on simple analogies (e.g., word analogies); datas…

Multiple-choice

Harmonizing Program Induction with Rate-Distortion Theory

2024-05-08 · Hanqi Zhou, David G. Nagy, Charley M. Wu

Many aspects of human learning have been proposed as a process of constructing mental programs: from acquiring symbolic number representations to intuitive theories about the world. In parallel, there is a long-tradition…

Program induction

Adaptive Token Boundaries: Integrating Human Chunking Mechanisms into Multimodal LLMs

2025-05-03 · Dongxing Yu

Recent advancements in multimodal large language models (MLLMs) have demonstrated remarkable capabilities in processing diverse data types, yet significant disparities persist between human cognitive processes and comput…

ChunkingQuestion AnsweringVisual Question Answering