Loading...
Loading...
Available on 1 platform
Sign in to view source links and access this dataset
MetaTruth is a benchmark dataset for evaluating metacognitive failures in large language models. It contains 105 tasks designed to test five specific mechanisms of failure across 15 different models. The dataset was sourced from Kaggle, but the original author, organization, and last update date are unknown.
License is unknown; users must verify licensing terms before use.