Tag
This paper proposes recall-based prompting strategies (Self-Recall and Question-Recall) to improve LLM knowledge cutoff adherence, outperforming existing methods on counterfactual questions and introducing a Multi-cutoff Historical Event Benchmark (MHEB) for robustness evaluation.