Entailment Accuracy is a key evaluation metric for fact consistency detection tasks in the fields of large language models and reasoning. It utilizes natural language inference models to determine whether generated content is entailed by the source text, thereby assessing fact consistency. It is typically reported as a percentage or a score between 0 and 1.

This metric is widely used in relevant benchmarks and papers, effectively reflecting model performance on this dimension. Higher values indicate better performance. It helps researchers and engineers quantify differences in actual model performance in fact consistency detection scenarios, facilitating horizontal comparisons across different methods or model versions.