AdaPLD: Adaptive Retrieval and Reuse for Efficient Model-Free Speculative Decoding 文章

ArXiv CS.CL2026-06-05NEWSen作者: Runheng Liu, Jincheng Xie, Wen Hu, Xingchen Xiao, Heyan Huang

AdaPLD: Adaptive Retrieval and Reuse for Efficient Model-Free Speculative Decoding · 相关技术