paper-with-me

홈 › Papers

Layers at Similar Depths Generate Similar Activations Across LLM Architectures

2025-04-03 · Christopher Wolfram, Aaron Schein

How do the latent spaces used by independently-trained LLMs relate to one another? We study the nearest neighbor relationships induced by activations at different layers of 24 open-weight LLMs, and find that they 1) tend to vary from layer to layer within a model, and 2) are approximately shared between corresponding layers of different models. Claim 2 shows that these nearest neighbor relationships are not arbitrary, as they are shared across models, but Claim 1 shows that they are not "obvious" either, as there is no single set of nearest neighbor relationships that is universally shared. Together, these suggest that LLMs generate a progression of activation geometries from layer to layer, but that this entire progression is largely shared between models, stretched and squeezed to fit into different architectures.

📄 PDF Abstract BibTeX arXiv:2504.08775

Code (0)

등록된 구현이 없습니다.

Methods 이 논문이 사용한 방법론

SET Dynamic Sparse Training method where weight mask is updated randomly periodically

Similar Papers 제목 키워드 기반

A Case Study of Deep-Learned Activations via Hand-Crafted Audio Features

2019-07-03 · Olga Slizovskaia, Emilia Gómez, Gloria Haro

The explainability of Convolutional Neural Networks (CNNs) is a particularly challenging task in all areas of application, and it is notably under-researched in music and audio domain. In this paper, we approach explaina…

Sparse Autoencoders Bridge The Deep Learning Model and The Brain

2025-06-10 · Ziming Mao, Jia Xu, Zeqi Zheng, Haofang Zheng 외

We present SAE-BrainMap, a novel framework that directly aligns deep learning visual model representations with voxel-level fMRI responses using sparse autoencoders (SAEs). First, we train layer-wise SAEs on model activa…

Deep Learning

A Library of Mirrors: Deep Neural Nets in Low Dimensions are Convex Lasso Models with Reflection Features

2024-03-02 · Emi Zeger, Yifei Wang, Aaron Mishkin, Tolga Ergen 외

We prove that training neural networks on 1-D data is equivalent to solving convex Lasso problems with discrete, explicitly defined dictionary matrices. We consider neural networks with piecewise linear activations and d…

Time Series

Causality $\neq$ Decodability, and Vice Versa: Lessons from Interpreting Counting ViTs

2025-10-10 · Lianghuan Huang, Yingshan Chang arxiv

Mechanistic interpretability seeks to uncover how internal components of neural networks give rise to predictions. A persistent challenge, however, is disentangling two often conflated notions: decodability--the recovera…

Object Counting

Fruit classification using deep feature maps in the presence of deceptive similar classes

2020-07-12 · Mohit Dandekar, Narinder Singh Punn, Sanjay Kumar Sonbhadra, Sonali Agarwal

Autonomous detection and classification of objects are admired area of research in many industrial applications. Though, humans can distinguish objects with high multi-granular similarities very easily; but for the machi…

ClassificationGeneral Classification