newsHacker NewsTrust 72 · CommunityPublished 6d agoLive · 6d ago
DSpark: Speculative decoding accelerates LLM inference [pdf]
717points294comments
Covers
Covers (incoming)
Related across the graph
paperWhen are likely answers right? On Sequence Probability and Correctness in LLMspaperBlockPilot: Instance-Adaptive Policy Learning for Diffusion-based Speculative Decodingrepoalibaba/rtp-llmreposgl-project/SpecForgepaperDepth Exploration for LLM Decodingrepoguoqingbao/xinferrepodphnAI/aphrodite-enginerepoKaden-Schutt/hipfirepaperSpeculative decoding with draft models