Key points are not available for this paper at this time.
Abstract While recent language models have the ability to take long contexts as input, relatively little is known about how well they use longer context. We analyze the performance of language models on two tasks that require identifying relevant information in their input contexts: multi-document question answering and key-value retrieval. We find that performance can degrade significantly when changing the position of relevant information, indicating that current language models do not robustly make use of information in long input contexts. In particular, we observe that performance is often highest when relevant information occurs at the beginning or end of the input context, and significantly degrades when models must access relevant information in the middle of long contexts, even for explicitly long-context models. Our analysis provides a better understanding of how language models use their input context and provides new evaluation protocols for future long-context language models.
Building similarity graph...
Analyzing shared references across papers
Loading...
Nelson F. Liu
Kevin Lin
John Hewitt
Transactions of the Association for Computational Linguistics
SHILAP Revista de lepidopterología
Stanford University
University of California, Berkeley
Building similarity graph...
Analyzing shared references across papers
Loading...
Liu et al. (Mon,) studied this question.
www.synapsesocial.com/papers/69d7cc4605ee2ba81dbee0e0 — DOI: https://doi.org/10.1162/tacl_a_00638
Synapse has enriched 4 closely related papers on similar clinical questions. Consider them for comparative context: