跳到正文
ollama release(github.com)· github-actions[bot]·· 6 天前品質評分44

Ollama v0.40.17-rc1 加入批次式推測驗證

v0.40.17

重點導讀

多個請求同時產生內容時,逐一檢查草稿可能拖慢速度。

  • Ollama v0.40.17-rc1 的 mlxrunner 改成批次檢查,讓不同請求共用目標模型的驗證工作,同時保留各自的抽樣方式和快取回復。
  • 它也會分別學習同時處理多個請求時該寫多長的草稿,並處理取消請求、不同生成方式混用等情況,讓並行生成時的草稿和檢查更能配合。

AI 伴隨問答

自備 Key 深度分析

針對這篇報導有疑問?可直接點選快捷問題或向 AI 追問背景脈絡與技術細節:

正文

Release v0.40.17-rc1: mlxrunner: batch speculative verification (#19663) · ollama/ollama · GitHub

Skip to content

Navigation Menu

[](https://github.com/)

Sign in

  • Products

All Features

  • Developers
  • Businesses
  • Resources

Search/

Sign in

Sign up

Appearance settings

You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert

{{ message }}

1. 1 more breadcrumb items
2. ollama
3. ollama

Repository navigation

More items

ReleasesTags

v0.40.17-rc1

This commit was created on GitHub.com and signed with GitHub’s verified signature. GPG key ID: B5690EEEBB952194 Verified Learn about vigilant mode
* Compare# Choose a tag to compare ## Sorry, something went wrong. Filter Loading ## Sorry, something went wrong. ### Uh oh!

There was an error while loading. Please reload this page. ## No results found View all tags

v0.40.17-rc1: mlxrunner: batch speculative verification (#19663)

  • v0.40.17-rc1
  • db5266a
  • Compare# Choose a tag to compare ## Sorry, something went wrong. Filter Loading ## Sorry, something went wrong. ### Uh oh!

There was an error while loading. Please reload this page. ## No results found View all tags
* Verified# Verified

This commit was created on GitHub.com and signed with GitHub’s verified signature. GPG key ID: B5690EEEBB952194 Verified Learn about vigilant mode

Image 9: @dhiltgendhiltgen tagged this 16 Nov 16:43

  • mlxrunner: batch speculative verification

Keep drafting active for concurrent generation by sharing target verification while preserving per-request sampling and cache rollback. Learn draft depth for each concurrent workload separately from isolated requests, and preserve probes that short requests never execute.

Materialize owned target features before a reader can pause, and cover acceptance, cancellation, handoffs, mixed decoding and retained-memory boundaries.

  • address comments

Assets 2

Footer

[](https://github.com/) © 2026 GitHub,Inc.

Footer navigation

You can’t perform that action at this time.

來源: ollama release (github.com)
開啟原文

讀者討論

理性交流 · 即時討論