Phased DMD: Few-step Distribution Matching Distillation via Score Matching within SubintervalsJan 1, 2026·Xiangyu Fan,Zesong Qiu,Zhuguanyu Wu,Fanzhou Wang,Zhiqian Lin,Tianxiang Ren,Dahua LinRuihao Gong,Lei Yang· 0 min read Cite URLTypeConference paperPublicationInternational Conference on Learning Representations (ICLR)Last updated on Jan 1, 2026 AuthorsRuihao Gong ← OmniFit: Bridging Modalities via Layer-Adaptive Token Compression for Omnimodal Large Language Models Jan 1, 2026PiLLM: Resource-efficient LLM Inference Using Workload Prediction Jan 1, 2026 →
,·
0 min read