/LogFormer

[AAAI 2024] LogFormer: A Pre-train and Tuning Pipeline for Log Anomaly Detection

Primary LanguagePython

LogFormer

[AAAI 2024] LogFormer: A Pre-train and Tuning Pipeline for Log Anomaly Detection

Data

Training data can be download from LogHub

Updates

01/23. We release the base code version for LogFormer, which is a strong baseline for log anomaly detection.

Data processing

  1. Downloading data into log_data/
  2. parse_log.py
  3. preprocess_xxx.py

Run

  1. First run train_transformer.py
  2. Then run tune_transformer.py

Citation

If you feel helpful, please cite our paper.

@inproceedings{guo2024logformer,
  title={Logformer: A pre-train and tuning pipeline for log anomaly detection},
  author={Guo, Hongcheng and Yang, Jian and Liu, Jiaheng and Bai, Jiaqi and Wang, Boyang and Li, Zhoujun and Zheng, Tieqiao and Zhang, Bo and Peng, Junran and Tian, Qi},
  booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
  volume={38},
  number={1},
  pages={135--143},
  year={2024}
}