AdaSD: Adaptive Speculative Decoding for Efficient Language Model Inference 文章

ArXiv CS.CL2026-05-27NEWSen作者: Kuan-Wei Lu, Ding-Yong Hong, Pangfeng Liu, Jan-Jan Wu
大语言模型
查看原文 →
关系图谱

AdaSD: Adaptive Speculative Decoding for Efficient Language Model Inference · 相关技术

相关技术

ginORMODEPLALLMlanguage modeltarget modeldraft modelTokenStanSpeculative decodingRPAParts-of-Speech (POS) tagsPPRMITLMMlarge language modelsICTHISForFFIEffort Metric AttentionDiTDAPTCodingCamouflaged object detectionARGANNAGIADAS