← 返回卡片

原书出处

Build a Reasoning Model (From Scratch)

MEAP Version 5(早期访问版) · 核验版本的摘录快照 · 导入于 2026-11-14T08:12:58.960Z

is removed. The remaining k tokens are then renormalized and sampled from. In short, top-k keeps a fixed number of the most likely tokens, whereas top-p keeps a variable number of tokens depending on their cumulative mass. Top-k is simpler to implement, and I covered it in my Build a Large Language Model (From Scratch) book. However, top-p sampling has become more popular recently. Exercise

此预览只包含定位所需的摘录和上下文。

原书定位详情

来源:a19c7933-4eda-4176-aebd-30fe7e1ad292

正文位置:extracted/Build a Reasoning Model (From Scratch)/_epub-src/OEBPS/Text/chapter-4.html

EPUB SHA-256:52d4d07986020be70f5aac21762dfcf60fcba08db0ba29d4f63957caf12aa377