Skip to content

Latest commit

 

History

History
18 lines (17 loc) · 867 Bytes

README.md

File metadata and controls

18 lines (17 loc) · 867 Bytes

TraceVAE

This is the source code for "Unsupervised Anomaly Detection on Microservice Traces through Graph VAE".

Usage

  1. pip3 install -r requirements.txt.
  2. Convert the dataset with python3 -m tracegnn.cli.data_process preprocess -i [input_path] -o [dataset_path]. The sample dataset is under sample_dataset. (Note: This sample dataset only shows data format and usage, and cannot be used to evaluate model performance. Please replace it with your dataset.) sample:
python3 -m tracegnn.cli.data_process preprocess -i sample_dataset -o sample_dataset
  1. Train the model with bash train.sh [dataset_path]:
bash train.sh sample_dataset
  1. Evaluate the model with bash teset.sh [model_path] [dataset_path]. The default model path is under results/train/models/final.pt:
bash test.sh results/train/models/final.pt sample_dataset