Disentangling Recall and Reasoning in Transformer Models through Layer-wise Attention and Activation Analysis

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Fartale, Harshwardhan, Kattamuri, Ashish, Raja, Rahul, Vats, Arpita, Prasad, Ishita, Moharir, Akshata Kishore
Format:	Preprint
Published:	2025
Subjects:	Machine Learning Artificial Intelligence
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Internet

https://arxiv.org/abs/2510.03366