llama.cpp b11368：为 draft 与 MTP 推测解码加入概率采样

AI 摘要

llama.cpp 发布 b11368 版本，为简单 draft 与 MTP 推测解码引入概率采样，使 drafter 采用概率采样、target 通过拒绝采样进行验证。该改动在语法约束请求下回退到 argmax 采样，并新增默认关闭（greedy）的开关以启用概率 draft 采样，同时修复了 draft sampler 与 target 共享 rng 流等问题。发布同时提供 macOS、Linux、Windows、Android 等多平台预编译包。

正文节选

<details open> spec : add probabilistic sampling for simple draft and MTP (#27694) * Make the drafter probabilistic and the target verify by rejection sampling * Drop stale spec_draft_q before drafting * Fallback to argmax sampling for grammar-constrained requests and adding flag for enabling probabilistic draft sampling. Default flag value is greedy. * Support grammar-constrained requests in rejection sampling * Fix - renormalize distribution after masking * copy rng on sampler copy and

发布时间：2026-10-03 10:35

抓取时间：2026-10-03 11:16

来源机构：ggml-org