Enregistré dans:
Détails bibliographiques
Auteurs principaux: Akhtar, Ryyan, Pahwa, Payal, Arora, Monika
Format: Preprint
Publié: 2026
Sujets:
Accès en ligne:https://arxiv.org/abs/2604.07098
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866911671464755200
author Akhtar, Ryyan
Pahwa, Payal
Arora, Monika
author_facet Akhtar, Ryyan
Pahwa, Payal
Arora, Monika
contents Large language models often fail on tasks they seem to already understand. In our experiments, this appears to be less about missing knowledge and more about certain internal circuits not being strongly activated during inference. We explore Selective Neuron Amplification, which increases the influence of task relevant neurons without changing the model's parameters. The method works at inference time and does not permanently alter the model. SNA helps mainly when the model is uncertain, while having low effect when the model is already confident. This suggests that some model failures are due to weak activation rather than lack of capability.
format Preprint
id arxiv_https___arxiv_org_abs_2604_07098
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Selective Neuron Amplification in Transformer Language Models
Akhtar, Ryyan
Pahwa, Payal
Arora, Monika
Machine Learning
Computation and Language
Large language models often fail on tasks they seem to already understand. In our experiments, this appears to be less about missing knowledge and more about certain internal circuits not being strongly activated during inference. We explore Selective Neuron Amplification, which increases the influence of task relevant neurons without changing the model's parameters. The method works at inference time and does not permanently alter the model. SNA helps mainly when the model is uncertain, while having low effect when the model is already confident. This suggests that some model failures are due to weak activation rather than lack of capability.
title Selective Neuron Amplification in Transformer Language Models
topic Machine Learning
Computation and Language
url https://arxiv.org/abs/2604.07098