search

rapidsai / clx

A collection of RAPIDS examples for security analysts, data scientists, and engineers to quickly get started applying RAPIDS and GPU acceleration to real-world cybersecurity use cases.
Apache License 2.0
168 stars 68 forks source link

[QST] Easy way to run Cybert in non GPU environment #505

Open isaacmg opened 2 years ago

isaacmg commented 2 years ago

What is your question?

Hi, first of all thanks for the work on Cybert it looks like an awesome paper and model. However, I was wondering if it would be possible to run it in inference mode on a CPU. I tried to look at your example code https://github.com/rapidsai/clx/blob/branch-22.08/notebooks/cybert/cybert_log_parsing.ipynb and it seems heavily connected into CUDF and other Rapids tools. Particularly your SubWordTokenizer class seems highly coupled. Our anomaly detection models need to run in inference mode on a virtual range that doesn't have access to GPUs. Also I tried to install directly just from HuggingFace repo and ran into issues. Are there any solutions? Thanks.

github-actions[bot] commented 2 years ago

This issue has been labeled inactive-30d due to no recent activity in the past 30 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed. This issue will be labeled inactive-90d if there is no activity in the next 60 days.