deadbits / vigil-llm

⚡ Vigil ⚡ Detect prompt injections, jailbreaks, and other potentially risky Large Language Model (LLM) inputs
https://vigil.deadbits.ai/
Apache License 2.0
270 stars 32 forks source link

Add Gandalf Lakera datasets #40

Closed deadbits closed 8 months ago

deadbits commented 9 months ago

Add Gandalf datasets

deadbits commented 9 months ago

Created new HF dataset repo for the Lakera dataset ada-002 embeddings https://huggingface.co/datasets/deadbits/vigil-gandalf-instruction-bypass-ada-002

All prompts were converted to lowercase before embedding. I'll create the sbert embeddings later today.

deadbits commented 8 months ago

I might get around to this later but OpenAI embeddings are good for now. I'll re-open this later if needed