What a benchmark scores with nothing in it: content-free members against each task's own majority baseline, plus a judged arm and a judge-disagreement control. Preregistered. WIP.
-
Updated
Sep 17, 2026 - Python
#
evaluation-integrity
Here are 2 public repositories matching this topic...
Reproducible code and metrics for measuring exploitability and contamination in LLM evaluation harnesses.
benchmarking reproducibility ai-safety large-language-models prompt-injection llm-evaluation llm-as-judge reward-hacking benchmark-contamination evaluation-integrity
-
Updated
Aug 6, 2026 - Python
Add this topic to your repo
To associate your repository with the evaluation-integrity topic, visit your repo's landing page and select "manage topics."