This is the origin Pytorch implementation of CAT in the following paper: [CAT: Beyond Efficient Transformer for Content-Aware AnomalyDetection in Event Sequences].
Figure 1. The architecture of CAT.
Dependencies can be installed using the following command:
pip install -r requirements.txt
The log datasets used in the paper can be found in the repo loghub. In this repository, an small sample of the HDFS dataset is proposed for a quick hands-up.
For generating the Log template files, please refer to the official implementation repo of logparser.
The simplest way of running CAT is to run
python main_cat.py --data HDFS.
If you have any questions, feel free to contact Shengming Zhang through Email ([email protected]) or Github issues.