INTELLIGENT LOG MANAGEMENT WITH LARGE LANGUAGE MODELS FOR ANOMALY DETECTION AND RESOLUTION
Prashant Lakhera · 2024
This paper presents a novel approach to intelligent log management, using large language models (LLMs) to detect and resolve anomalies in real-time. By extending the capabilities of a model like GPT for log analysis, we can enable the automatic detection of anomalies with actionable recommendations. The model was trained on a dataset of system logs with hyperparameters optimized for anomaly detection accuracy. The evaluation shows that the model successfully identified 85% of anomalies with a false positive rate of 5%, significantly outperforming traditional log analysis tools. Further, the model provides actionable recommendations in 90% of cases, which reduce the Mean Time to Recovery (MTTR) by 40%.