turkish-over-refusal-set fevziegeyurtsevenler · COMPLETED
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.
github.com/fevziegeyurtsevenler/turkish-over-refusal-set · ★ 0 · Forks 0 · Size 105 KB
SUMMARY
Technologies 4
Scored 4
Observed 4
Practices 6
Evidence 4
Skips 0
COVERAGE
Analyzed 14 files · 4 commits · 0 API calls
TECHNOLOGIES & DEPTH
JSON LANGUAGE Depth 70
1 files · PRODUCTION
Markdown LANGUAGE Depth 70
3 files · PRODUCTION
YAML LANGUAGE Depth 70
1 files · PRODUCTION
Python LANGUAGE Depth 70
4 files · PRODUCTION
PRACTICES
documentation · observedautomated_tests · observedcontinuous_integration · observedcontainerization · absentlinting · absentformatting · absent
ACTIVITY & OWNERSHIP
First commit 2026-07-26
Last commit 2026-08-08
Active months 2
Commits 4