Accelerating LLM Decoding with Speculative Sampling
Accelerating LLM Decoding with Speculative Sampling
Why it matters
An independent formulation of the same technique, with a 2–2.5× speedup
Source — Primary sources first; secondary reports are labelled “Press”
Paperarxiv.orghttps://arxiv.org/abs/2302.01318Added in Week 35, 2026 · See that week