Tag: lost-in-the-middle
All the articles with the tag "lost-in-the-middle".
Long Context, and Where It Breaks: Lost in the Middle
Updated:A model that accepts a long prompt doesn't use all of it equally. The lost-in-the-middle result reproduced on a 1.5B model: a key-value lookup that's perfect at 2,000 tokens and loses things in the middle by 8,000. Why the needle-in-a-haystack test misses it, what RULER and Context Rot measure instead, and the causes: softmax dilution worked out, attention sinks measured, RoPE's reach, and training on short documents. Then what to do about it: put things where they're read, retrieve less, compact, train longer.