Negligible in Size, Significant in Effect: On Scale Vectors in Large Language Models 文章

ArXiv CS.AI2026-05-27NEWSen作者: Mingze Wang, Shuchen Zhu, Yuxin Fang, Binghui Li, Kai Shen, Shu Zhong
大语言模型
查看原文 →
关系图谱

Negligible in Size, Significant in Effect: On Scale Vectors in Large Language Models · 相关技术

相关技术

ORMODELLMlanguage modelweight decayUCTStraight-Through EstimatorStanReferring expression comprehension (REC)Pre-trainingOWSNarrative Abstraction BenchmarkMaliLMMlarge language modelsHISForEffort Metric AttentionDiTARGANN