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

Pre-trained language models (PLMs) contain vast amounts of factual knowledge,but how the knowledge is stored in the parameters remains unclear. This paperdelves into the complex task of understanding how factual knowledge is storedin multilingual PLMs, and introduces the Architecture-adapted MultilingualIntegrated Gradients method, which successfully localizes knowledge neuronsmore precisely compared to current methods, and is more universal acrossvarious architectures and languages. Moreover, we conduct an in-depthexploration of knowledge neurons, leading to the following two importantdiscoveries: (1) The discovery of Language-Independent Knowledge Neurons, whichstore factual knowledge in a form that transcends language. We designcross-lingual knowledge editing experiments, demonstrating that the PLMs canaccomplish this task based on language-independent neurons; (2) The discoveryof Degenerate Knowledge Neurons, a novel type of neuron showing that differentknowledge neurons can store the same fact. Its property of functional overlapendows the PLMs with a robust mastery of factual knowledge. We designfact-checking experiments, proving that the degenerate knowledge neurons canhelp the PLMs to detect wrong facts. Experiments corroborate these findings,shedding light on the mechanisms of factual knowledge storage in multilingualPLMs, and contribute valuable insights to the field. The code is available athttps://github.com/heng840/AMIG.

Journey to the Center of the Knowledge Neurons: Discoveries of Language-Independent Knowledge Neurons and Degenerate Knowledge Neurons