TirZ

LogFormer_1 is a lightweight Transformer-based log anomaly detection system designed for industrial log analysis and AIOps. The model learns normal log event patterns from historical system logs and identifies unusual behavior using next-event prediction — a technique adapted from language modeling to the domain of system log monitoring.

TirZ M1 is a GPT-style decoder-only transformer language model comprising 417,887,232 trainable parameters, designed and trained from scratch by Tirth Patel (@Tirth9978) under the OpenTirZ organization. Built upon the foundational principles established in the landmark "Attention Is All You Need" paper .