deep-loglizer
deep-loglizer copied to clipboard
logpai
A deep learning toolkit for log-based anomaly detection
I specifically wanted to run logAnomaly , for a different dataset , say Android logs, how should I approach?
I noticed, after load_sessions is called, it has train/test data in form of: ****{'templates': ['BLOCK* NameSystem.allocateBlock:',...], 'label': 0}.**** For unsupervised data there is no label key available. How to handle...
Hello, Is there a prerequisite list, as during the installation I get an error message? Thx
it seems that the transformer implemented is just a binary classification model as other models. The implementation doesn't seem to follow the original paper in which positional encoding is employed.
Why do you slice the data into sessions by fixed time window，then slice each session into windows by sliding log-based window? What is the advantages? how about comparison with simply...
Hello, I read your paper using semantic features to improve performance, but in the code I only see using tfidf to generate word vectors. However, tfidf is not related to...
lstm.py can be used for running deeplog and loganomaly algorithms. Can you tell me how to set parameter values in order to run loganomaly?
Hi, when I am trying to run demo/lstm_demo.py I am running into a RuntimeError: cuDNN error: CUDNN_STATUS_EXECUTION_FAILED. This is using the data stored within ../data/processed/HDFS_100k/hdfs_0.0_tar.
Hi @zbchern, @zhujiem @JinYang88 @zbchern First of all Thanks your work looks great! - I observed in `processed` folder you've structured data in `json` and `pkl` format only for **HDFS**...
Hello, I am having some trouble reproducing the results of Table2 of your paper (https://arxiv.org/pdf/2107.05908) on the HDFS dataset. For the unsupervised methods (LSTM, Transformer, and Autoencoder), I am following...
