English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Retrieval-augmented language models (RALMs) hold promise to produce languageunderstanding systems that are are factual, efficient, and up-to-date. Animportant desideratum of RALMs, is that retrieved information helps modelperformance when it is relevant, and does not harm performance when it is not.This is particularly important in multi-hop reasoning scenarios, where misuseof irrelevant evidence can lead to cascading errors. However, recent work hasshown that retrieval augmentation can sometimes have a negative effect onperformance. In this work, we present a thorough analysis on five open-domainquestion answering benchmarks, characterizing cases when retrieval reducesaccuracy. We then propose two methods to mitigate this issue. First, a simplebaseline that filters out retrieved passages that do not entail question-answerpairs according to a natural language inference (NLI) model. This is effectivein preventing performance reduction, but at a cost of also discarding relevantpassages. Thus, we propose a method for automatically generating data tofine-tune the language model to properly leverage retrieved passages, using amix of relevant and irrelevant contexts at training time. We empirically showthat even 1,000 examples suffice to train the model to be robust to irrelevantcontexts while maintaining high performance on examples with relevant ones.

Making Retrieval-Augmented Language Models Robust to Irrelevant Context