Paper Image

Large language models struggle with long-range dependencies

Published on:

8 November 2023

Primary Category:

Computation and Language

Paper Authors:

Jiaqi Li,

Mengmeng Wang,

Zilong Zheng,

Muhan Zhang

Bullets

Key Details

New benchmark evaluates large language models on long documents

Models do well on short tasks like QA and cloze but struggle on long reasoning

Commercial models outperform open-source but still limited

Increasing context size helps but doesn't solve core issues

More work needed for true long document understanding

AI generated summary

Large language models struggle with long-range dependencies

This paper evaluates large language models on their ability to understand long documents. The models excel at short tasks like question answering but struggle with long-range reasoning. The benchmark reveals limitations in commercial and open-source models.

Answers from this paper

Comments

No comments yet, be the first to start the conversation...

Sign up to comment on this paper

Sign Up