Dense2MoE: Pushing the Pareto Frontier of On-Device LLMs via Unified Pruning and Upcycling 文章

ArXiv CS.AI2026-05-27NEWSen作者: Fengfa Li, Hongjin Ji, Yifeng Ding, Lei Ren, Chen Wei
查看原文 →
关系图谱

Dense2MoE: Pushing the Pareto Frontier of On-Device LLMs via Unified Pruning and Upcycling · 相关技术

相关技术

Multi-layer perceptronsTTEODELLM远程代码执行（RCE）研究法辛烷值attentionWALUnifiedUCTTokenStraight-Through EstimatorStanPromiseParts-of-Speech (POS) tagsNATMoEMLPMITLMMICTHISForFFIEffort Metric AttentionDenseCLIANNICE