New SCRIBE framework uses LLMs to interpret brain-grounded semantics

Researchers have developed SCRIBE, a new framework designed to create interpretable, word-level semantic representations of text grounded in a neuroscience-informed semantic space. This framework utilizes an ensemble of open-source large language models to capture how context influences meaning. A study involving 921 participants indicated that SCRIBE accurately reflects human judgments of semantic features and predicts neural responses in language-selective brain regions, outperforming non-contextual methods. AI

IMPACT This framework could advance the understanding of how language is processed in the brain by providing more contextually aware semantic representations.

RANK_REASON The cluster contains a research paper detailing a new framework for semantic representation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• alphaXiv
• arXiv
• CatalyzeX
• DagsHub
• Gotit.pub
• Hugging Face
• ScienceCast
• SCRIBE

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →