Pinned
Proud to announce:
π« garak - an LLM vulnerability scannerπ«
π Check if a model is susceptible to common attacks
π¦ Supports HuggingFace, OpenAI, ggml, Cohere, ...
π§ >70 probes: prompt injection, false claims, toxicity, encoding evasion, ..
