Changling Li leads EvalAwareBench to measure when LLMs know they are being tested

In a new paper and open releases, Li and collaborators decompose evaluation awareness, test nine models across four benchmarks, and publish a factor-controlled dataset and code.

By · Published

Illustration of a small stylized AI figure in a display being examined, representing Changling Li's EvalAwareBench testing LLMs.

In a new paper and open releases, Li and collaborators decompose evaluation awareness, test nine models across four benchmarks, and publish a factor-controlled dataset and code.