一作ACL 2026Findings· Findings of ACL
Mask Tokens as Prophet: Fine-Grained Cache Eviction for Efficient d LLM Inference
Jianuo Huang · Yaojie Zhang · Yicun Yang · Benhao Huang · Linfeng Zhang
摘要
Diffusion large language models (dLLMs) present a promising alternative to dominant autoregressive models (ARMs) by the ability of parallel decoding at the expense of substantial computation and memory costs. Specifically, the cache mechanism for bidirectional attention in dLLMs demands large mem...