XSTest-style over-refusal eval for Turkish LLM guards: 120 benign-but-scary pairs. Finding: ProtectAI over-refuses 59% of benign Turkish prompts vs 0.8% English (71x). Dataset + refusal scorer.
-
Updated
Jul 26, 2026 - Python
#
xstest
Here is 1 public repository matching this topic...
Improve this page
Add a description, image, and links to the xstest topic page so that developers can more easily learn about it.
Add this topic to your repo
To associate your repository with the xstest topic, visit your repo's landing page and select "manage topics."