RELISH architecture offers efficient LLM text regression

Researchers have introduced RELISH, a novel and efficient architecture for performing text regression with large language models. Unlike existing methods that decode numeric targets as text or aggregate multiple outputs, RELISH directly predicts scalar values by iteratively refining a latent state. This approach, which uses minimal additional parameters, consistently outperforms previous baselines across various datasets and LLM backbones. AI

IMPACT This new architecture offers a more parameter-efficient approach to text regression with LLMs, potentially improving performance in applications requiring scalar value prediction.

RANK_REASON The cluster contains a research paper detailing a novel architecture for LLM text regression. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• arXiv
• CORE Recommender
• DagsHub
• Hugging Face
• RELISH
• Yiheng Su

• paper
• model release

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