# Lost in the middle

> Lost in the middle: The finding (Liu et al., 2023) that language models often use information at the start or end of a long context better than…

Canonical: https://llmdojo.dev/glossary/lost-in-the-middle

The finding (Liu et al., 2023) that language models often use information at the start or end of a long context better than information in the middle, giving a U-shaped accuracy curve.

Learn it in the lesson: [Position matters: lost in the middle](https://llmdojo.dev/learn/prompt-context-engineering/context-selection/lost-in-the-middle) (Prompt and Context Engineering · Choosing Context: Selection, Ordering and Compression).

## Related
- [Needle in a haystack](https://llmdojo.dev/glossary/needle-in-a-haystack)
- [Reciprocal rank fusion](https://llmdojo.dev/glossary/reciprocal-rank-fusion)
- [BM25](https://llmdojo.dev/glossary/bm25)
- [Glossary of every term](https://llmdojo.dev/glossary)