The 'Lost-in-the-Middle' phenomenon, as described by Liu et al. (2023a), is a manifestation of position bias where a model's performance significantly degrades when crucial information is placed in the middle of a long input context, even if the context window is large enough to contain it.

Authors

• Person: Not available Organization: arXiv
  A Survey on the Theory and Mechanism of Large Language Models

Sources

• A Survey on the Theory and Mechanism of Large Language Models arxiv.org arXiv via serper

Referenced by nodes (1)

• Large Language Models concept