Conceptevaluation format gap
Evaluation Format Gap
A systematic performance discrepancy observed in LLMs between multiple-choice tasks (recognition) and open-ended generation tasks (recall). This gap reflects how different evaluation formats influence the measurement of model capabilities.
Timeline (last 90 days)
Sep 24
The researchers interpreted the gap between selection and generation as an 'evaluation format gap,' rather than direct evidence that the model's lexical knowledge is intact.
Unverified50%