AIOps: Log Anomaly Detection with LLMs

A laptop-CPU system that reads 16 million real system log lines, finds the abnormal ones and explains them. Drain turns raw lines into templates. DeepLog, classic detectors and an LLM-embedding detector find anomalous sessions and windows, and a local LLM writes root-cause summaries that are checked automatically before an engineer reads them. Everything is measured on Loghub HDFS and BGL in time order.

Category: AIOps / Machine Learning for Systems / Large Language Models

CSE Project 22 from ProTechSC Academy, with graphical abstract, video demonstration and project guidance.