Temperature in SLMs: Impact on Incident Categorization in On-Premises Environments

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Pohlmann, Marcio, Severo, Alex, Almeida, Gefté, Kreutz, Diego, Heinrich, Tiago, Pereira, Lourenço
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912727361912832
author Pohlmann, Marcio
Severo, Alex
Almeida, Gefté
Kreutz, Diego
Heinrich, Tiago
Pereira, Lourenço
author_facet Pohlmann, Marcio
Severo, Alex
Almeida, Gefté
Kreutz, Diego
Heinrich, Tiago
Pereira, Lourenço
contents SOCs and CSIRTs face increasing pressure to automate incident categorization, yet the use of cloud-based LLMs introduces costs, latency, and confidentiality risks. We investigate whether locally executed SLMs can meet this challenge. We evaluated 21 models ranging from 1B to 20B parameters, varying the temperature hyperparameter and measuring execution time and precision across two distinct architectures. The results indicate that temperature has little influence on performance, whereas the number of parameters and GPU capacity are decisive factors.
format Preprint
id arxiv_https___arxiv_org_abs_2511_19464
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Temperature in SLMs: Impact on Incident Categorization in On-Premises Environments
Pohlmann, Marcio
Severo, Alex
Almeida, Gefté
Kreutz, Diego
Heinrich, Tiago
Pereira, Lourenço
Distributed, Parallel, and Cluster Computing
Artificial Intelligence
Cryptography and Security
Machine Learning
Performance
68T01
I.2
SOCs and CSIRTs face increasing pressure to automate incident categorization, yet the use of cloud-based LLMs introduces costs, latency, and confidentiality risks. We investigate whether locally executed SLMs can meet this challenge. We evaluated 21 models ranging from 1B to 20B parameters, varying the temperature hyperparameter and measuring execution time and precision across two distinct architectures. The results indicate that temperature has little influence on performance, whereas the number of parameters and GPU capacity are decisive factors.
title Temperature in SLMs: Impact on Incident Categorization in On-Premises Environments
topic Distributed, Parallel, and Cluster Computing
Artificial Intelligence
Cryptography and Security
Machine Learning
Performance
68T01
I.2
url https://arxiv.org/abs/2511.19464