SemaTune uses LLM guidance with semantic context to tune up to 41 Linux OS parameters, delivering 72.5% performance gains over defaults and 153.3% over non-LLM baselines on 13 workloads while avoiding degraded states.
{OPPerTune}:{Post-Deployment} configuration tuning of services made easy
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.OS 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
SemaTune: Semantic-Aware Online OS Tuning with Large Language Models
SemaTune uses LLM guidance with semantic context to tune up to 41 Linux OS parameters, delivering 72.5% performance gains over defaults and 153.3% over non-LLM baselines on 13 workloads while avoiding degraded states.