logbert

logbert copied to clipboard

The experimental data of the paper cannot be reproduced

9

hi, guo I have tried many times. The following results are always the same, which is far from the results in the paper. Is there any difference between the results...

chinahappyking

scikt-learn==1.1.0 should be scik**i**t-learn==1.1.0

ghost

Is there any way to index (mark) anomaly log sequence back into parsed file usually named `log_structured` from drain? I tried to use following workflow for `LineId` from `structured` file...

shoaib-intro

thatmee

Update typo in requirements.txt

1

just-in-chang

E.g., perhaps MIT like your dependency logalizer?

Antymon

When I run **python deeplog.py train**, I got an error of **PermissionError: [Errno 1] Operation not permitted**. I don't have superuser privilege for my laptop. Is there an alternative command...

htcml

在BGL数据集上训练的时候出现以下错误： ValueError: test_size=68 should be either positive and smaller than the number of samples 34 or a float in the (0, 1) range 请问该如何解决？

lianzhaoy

关于log key中的<*>如何参与Bert的MLM训练

2

您好，logbert这篇论文让我获益匪浅。与LAnoBERT进行了一些对比，关于log key中的参与预训练任务我有以下疑惑： 1）在数据预处理阶段，train数据只包含日志Sequence中的log key ID（对应代码中的EventID），请问bert预训练过程中如何从vocab中关联token呢？ 2）您论文中的input应该是log key，请问log key中的是否参与训练呢？若参与训练，请问随机mask（如果mask到）之后，如何做loss呢？若不参与训练，请问在mask阶段应该如何处理呢？ 期待您的答疑解惑，如若上述理解有偏差，请您见谅。

zora-triumph

cannot reproduce results

3

Hi, I tried to run your latest version code and here is the result I got on HDFS dataset: best threshold: 0, best threshold ratio: 0.2 TP: 2999, TN: 546622,...

mmichaelzhang