hard20 pts · first answer countsragcontext-management

Stuffing 100 chunks into the prompt

To "improve recall", a team puts 100 retrieved chunks into a long-context prompt. Answer quality drops compared to using the top 8. What is the most likely reason?

A.Relevant facts get buried among many distractors; models often use information in the middle of long contexts less reliablyB.Long context models cannot read more than 8 chunksC.The vector database returns random results after 8D.Temperature automatically increases with prompt length

← Why long contexts eat GPU memory What LoRA trains →