DogeDesigner@cb_doge
54AI 编辑部评分,满分 100

X 开源 For You 时间线推荐算法

2026-08-14 01:41· 17分钟前
AI 导读

𝕏 宣布开源 For You 时间线推荐算法,公开帖子排名与过滤机制,涵盖核心排序系统、模型配置、时间线过滤器及影响可见性的内容标签。同时推出“Under the Hood”页面,符合条件的用户可查看账号或帖子是否被施加可见性限制标签并下载数据。试点面向随机选取的、注册满一年且上月发帖超 10 次的账号,并邀请独立推荐系统专家审查。

BREAKING: X is open-sourcing the For You timeline and introducing a new level of transparency.

The released code will explain how posts are ranked and filtered, including:

• Core ranking systems • Model configurations • Timeline filters • Content labels that can affect a post's visibility

X is also launching an "Under the Hood" page that will allow eligible users to see whether visibility-limiting labels were applied to their accounts or posts, and download that data.

The pilot will initially include a randomized group of accounts that are at least one year old and posted 10+ times during the previous month. X also invited independent recommender-system experts to review and pressure-test the release.

This is a major step toward helping users understand how their reach is determined, whether the system is fair, and why certain content appears in their feeds.

来源:DogeDesigner · x.com

X 开源 For You 时间线推荐算法

DogeDesigner · @cb_doge · X·2026-08-14 01:41·17分钟前
AI 导读

𝕏 宣布开源 For You 时间线推荐算法,公开帖子排名与过滤机制,涵盖核心排序系统、模型配置、时间线过滤器及影响可见性的内容标签。同时推出“Under the Hood”页面,符合条件的用户可查看账号或帖子是否被施加可见性限制标签并下载数据。试点面向随机选取的、注册满一年且上月发帖超 10 次的账号,并邀请独立推荐系统专家审查。

BREAKING: X is open-sourcing the For You timeline and introducing a new level of transparency.

The released code will explain how posts are ranked and filtered, including:

• Core ranking systems • Model configurations • Timeline filters • Content labels that can affect a post's visibility

X is also launching an "Under the Hood" page that will allow eligible users to see whether visibility-limiting labels were applied to their accounts or posts, and download that data.

The pilot will initially include a randomized group of accounts that are at least one year old and posted 10+ times during the previous month. X also invited independent recommender-system experts to review and pressure-test the release.

This is a major step toward helping users understand how their reach is determined, whether the system is fair, and why certain content appears in their feeds.

来源:DogeDesigner· x.com