Last updated
Last updated
garak is designed to help discover situations where a language model generates outputs that one might not want it to. If you know nmap or metasplot for traditional netsec/infosec analysis, then garak aims to operate in a similar space for language models.
garak has probes that try to look for different "vulnerabilities". Each probs sends specific prompts to models, and gets multiple generations for each prompt. LLM output is often stochastic, so a single test isn't very informative. These generations are then processed by "detectors", which will look for "hits". If a detector registers a hit, that attempt is registered as failing. Finally, a report is output with the success/failure rate for each probe and detector.
No. The scores from any probe don't operate on any kind of normalised scale. Higher passing percentage is better, but that's it. No meaningful comparison can be made between scores from different probes.
Each detector is different. Most use a classifier (either locally or via API) to judge the response, or look for keywords that are (or are not) present in the language model output, but there are many different failure modes. Your best bet might be to look in the technical or the .
Additional prompts can be probed by creating a new plugin -- this isn't as tough as it sounds; take a look at the modules in the garak/probes/ directory for inspiration. The reference documentation includes a section on .
The JSONL report created for each run includes language model parameters, all the prompts sent to the mode, all the model responses, and also the mapping between these and evaluation scores. There's a JSONL report analysis script in analyze/analyze_log.py.
Not immediately, but if you have the Gradio skills, get in touch!
Perhaps - please , including a description of the vulnerability, example prompts, and tag it "new plugin" and "probes".
Would love to! Please , tagging it "new plugin" and "generators". You can even check out our guide to .
We keep a live official FAQ in the - but also just reach out and !