#吐槽
【生日会 父女VR】我对你来说，重要吗？ UP主: YuriHimself - https://www.bilibili.com/video/BV1nYBWBaEih
via Nostr@cxplay
【生日会 父女VR】我对你来说，重要吗？ UP主: YuriHimself - https://www.bilibili.com/video/BV1nYBWBaEih
via Nostr@cxplay
静读天下 Pro 折扣 50%, 原价 US$10.99, 现价 US$5.49.
https://play.google.com/store/apps/details?id=com.flyersoft.moonreaderp
#折扣 #Android
via @[email protected]
https://play.google.com/store/apps/details?id=com.flyersoft.moonreaderp
#折扣 #Android
via @[email protected]
#吐槽
In reply to nevent1q…jemd
_________________________
有时候别人也会叫他们粉头/网红/KOL/水怪话题制造者/操盘手/黑红嘴/顶流/社交媒体影响者. OG 在英文语境里面主要是褒义居多, 比如这里的台湾歌手光良就是一个 OG, 也可能是 Nostr 中文社区里唯一的 OG, 和 OG 相反的是 "素人".
不过咱们这边对 OG 的贬低越来越多了, 最近的 "老资历" 就是用来嘲讽那些领域里的 OG 的, 真假 OG 并无所谓, 都会被冠上这个帽子.
via Nostr@cxplay
In reply to nevent1q…jemd
_________________________
有时候别人也会叫他们粉头/网红/KOL/水怪话题制造者/操盘手/黑红嘴/顶流/社交媒体影响者. OG 在英文语境里面主要是褒义居多, 比如这里的台湾歌手光良就是一个 OG, 也可能是 Nostr 中文社区里唯一的 OG, 和 OG 相反的是 "素人".
不过咱们这边对 OG 的贬低越来越多了, 最近的 "老资历" 就是用来嘲讽那些领域里的 OG 的, 真假 OG 并无所谓, 都会被冠上这个帽子.
via Nostr@cxplay
#吐槽
In reply to nevent1q…mywq
_________________________
我的经验里是普通人的内容输入和输出比是 10:1, 除非去刻意运营, 不然社区内容要么取决于人数要么取决于 OG 的活跃程度. 目前这里的中文社区既没有可数的 OG 也没有太多基数用户, 所以内容也就很少了.
况且大多数 OG 很在意自己的流量, 而 Nostr 的流量追踪和统计天然不可靠. :bili_doge:
via Nostr@cxplay
In reply to nevent1q…mywq
_________________________
我的经验里是普通人的内容输入和输出比是 10:1, 除非去刻意运营, 不然社区内容要么取决于人数要么取决于 OG 的活跃程度. 目前这里的中文社区既没有可数的 OG 也没有太多基数用户, 所以内容也就很少了.
况且大多数 OG 很在意自己的流量, 而 Nostr 的流量追踪和统计天然不可靠. :bili_doge:
via Nostr@cxplay
#article #read
In reply to nevent1q…kw8j
_________________________
[...] The Going Dark initiative, or ProtectEU as the Commission now calls it, wants to “enable law enforcement authorities to access encrypted data in a lawful manner”. This is a Chat Control 3.0 attempt.
The EU Commission and several member states are also looking for new rules on data retention. In a new ”Presidency outcome paper”, the member states discuss metadata retention: which websites you visit, and who is communicating with whom, when and how often. The ambition is “to have the broadest possible scope of application” and this time some member states also want the proposal to include VPN services. [...]
https://mastodon.online/@mullvadnet/115742530333573065
via Nostr@cxplay_clip
In reply to nevent1q…kw8j
_________________________
[...] The Going Dark initiative, or ProtectEU as the Commission now calls it, wants to “enable law enforcement authorities to access encrypted data in a lawful manner”. This is a Chat Control 3.0 attempt.
The EU Commission and several member states are also looking for new rules on data retention. In a new ”Presidency outcome paper”, the member states discuss metadata retention: which websites you visit, and who is communicating with whom, when and how often. The ambition is “to have the broadest possible scope of application” and this time some member states also want the proposal to include VPN services. [...]
https://mastodon.online/@mullvadnet/115742530333573065
via Nostr@cxplay_clip
#article #read
Problem when navigating routing - Support - Netlify Support Forums
https://answers.netlify.com/t/problem-when-navigating-routing/98037
via Nostr@cxplay_clip
Problem when navigating routing - Support - Netlify Support Forums
https://answers.netlify.com/t/problem-when-navigating-routing/98037
via Nostr@cxplay_clip
#吐槽
In reply to nevent1q…tpwg
_________________________
Make mute list subscribe-able, like adblock filter list.
via Nostr@cxplay
In reply to nevent1q…tpwg
_________________________
Make mute list subscribe-able, like adblock filter list.
via Nostr@cxplay
#吐槽
不止于过滤列表：基于大语言模型（LLM）的广告拦截新思路
https://adguard.com/zh_cn/blog/beyond-filter-lists-rethinking-ad-blocking-with-llms.html
TL;DR
通过扩展 AdGuard CSS 过滤规则中的的 :contains 伪类, 使其可以通过三种模式连接到 LLM 进行语义计算:
* :contains-meaning-embedding: 通过嵌入式模型检测目标文本与标准文本的向量相似度. (速度快/成本低/单一语言准确度高/难以处理多语言和复杂情况)
* :contains-meaning-prompt: 直接将选中的页面元素发给 LLM 询问是否符合过滤标准. (云托管模型准确度高速度也快/本地模型延迟高且准确性也低)
* :contains-meaning-vision: 将选中的页面元素截图发给 LLM 进行视觉判断. (云和本地模型准确都高但延迟也一样高)
原型插件: https://github.com/maximtop/ai-blocker-experiment
#AdGuard #LLM
via Nostr@cxplay
不止于过滤列表：基于大语言模型（LLM）的广告拦截新思路
https://adguard.com/zh_cn/blog/beyond-filter-lists-rethinking-ad-blocking-with-llms.html
TL;DR
通过扩展 AdGuard CSS 过滤规则中的的 :contains 伪类, 使其可以通过三种模式连接到 LLM 进行语义计算:
* :contains-meaning-embedding: 通过嵌入式模型检测目标文本与标准文本的向量相似度. (速度快/成本低/单一语言准确度高/难以处理多语言和复杂情况)
* :contains-meaning-prompt: 直接将选中的页面元素发给 LLM 询问是否符合过滤标准. (云托管模型准确度高速度也快/本地模型延迟高且准确性也低)
* :contains-meaning-vision: 将选中的页面元素截图发给 LLM 进行视觉判断. (云和本地模型准确都高但延迟也一样高)
原型插件: https://github.com/maximtop/ai-blocker-experiment
#AdGuard #LLM
via Nostr@cxplay
AdGuard
不止于过滤列表：基于大语言模型（LLM）的广告拦截新思路
在这项研究中，AdGuard 开发者 Maxim Topciu 深入探讨了多种运用大语言模型（LLM）实现自动广告拦截的方法，摆脱了对传统过滤列表的依赖，并展示了可供用户自行试用验证的功能原型。
#article #read
In reply to nevent1q…x0kd
_________________________
Range header is ignore on large files - General - Cloudflare Community
https://community.cloudflare.com/t/range-header-is-ignore-on-large-files/650787
via Nostr@cxplay_clip
In reply to nevent1q…x0kd
_________________________
Range header is ignore on large files - General - Cloudflare Community
https://community.cloudflare.com/t/range-header-is-ignore-on-large-files/650787
via Nostr@cxplay_clip
#article #read
cloudflare缓存与绕过缓存 | 雪饼的狗窝
https://zhouxuebin.club/blog/posts/cloudflare-bypass-cache/7fd52515/
via Nostr@cxplay_clip
cloudflare缓存与绕过缓存 | 雪饼的狗窝
https://zhouxuebin.club/blog/posts/cloudflare-bypass-cache/7fd52515/
via Nostr@cxplay_clip
#article #read
The Impossibility of Perfectly Caching HTTP Range Requests - Kevin Cox
https://kevincox.ca/2021/06/04/http-range-caching/
via Nostr@cxplay_clip
The Impossibility of Perfectly Caching HTTP Range Requests - Kevin Cox
https://kevincox.ca/2021/06/04/http-range-caching/
via Nostr@cxplay_clip
#吐槽
In reply to nevent1q…7j3y
_________________________
I don't think so, because the mainstream now is *aaS, although they might increase their prices as a result.
via Nostr@cxplay
In reply to nevent1q…7j3y
_________________________
I don't think so, because the mainstream now is *aaS, although they might increase their prices as a result.
via Nostr@cxplay
#吐槽
In reply to nevent1q…uv9h
_________________________
在 Hacker News 刷到了自托管数据库对比 AWS RDS 的, 里面的热评就很好地解释了这个责任的负担: 宜家和迪士尼也宕机了, 得等 AWS 它们自己恢复. 潜台词是: 与我无关.
Go ahead, self-host Postgres | Hacker News
https://news.ycombinator.com/item?id=46336947
类似的, 自己建边缘节点和直接用 Cloudflare 也是一样, 有能力自己建, 还得有能力自己承担责任.
以前对自托管的要求只有: 如果只需要对自己负责, 那就无所谓. 直到现在只是出于好玩和节省成本才去自托管了, 如果需要超长期使用和更高的可靠性(责任转移), 那我也会选代托管.
via Nostr@cxplay
In reply to nevent1q…uv9h
_________________________
在 Hacker News 刷到了自托管数据库对比 AWS RDS 的, 里面的热评就很好地解释了这个责任的负担: 宜家和迪士尼也宕机了, 得等 AWS 它们自己恢复. 潜台词是: 与我无关.
Go ahead, self-host Postgres | Hacker News
https://news.ycombinator.com/item?id=46336947
类似的, 自己建边缘节点和直接用 Cloudflare 也是一样, 有能力自己建, 还得有能力自己承担责任.
以前对自托管的要求只有: 如果只需要对自己负责, 那就无所谓. 直到现在只是出于好玩和节省成本才去自托管了, 如果需要超长期使用和更高的可靠性(责任转移), 那我也会选代托管.
via Nostr@cxplay
#吐槽
如何提高 CDN 缓存效率和命中率: 将 blob 细细切作臊子
Blossom 存分块, Nostr 中继存 manifest, 感觉会很好玩(很滥用), 让我想起了那些拿假切片薅图床的邪门东西.
hzrd149/cherry-tree: Chunked blobs on blossom
https://github.com/hzrd149/cherry-tree
via Nostr@cxplay
如何提高 CDN 缓存效率和命中率: 将 blob 细细切作臊子
Blossom 存分块, Nostr 中继存 manifest, 感觉会很好玩(很滥用), 让我想起了那些拿假切片薅图床的邪门东西.
hzrd149/cherry-tree: Chunked blobs on blossom
https://github.com/hzrd149/cherry-tree
via Nostr@cxplay
GitHub
GitHub - hzrd149/cherry-tree: Chunked blobs on blossom
Chunked blobs on blossom. Contribute to hzrd149/cherry-tree development by creating an account on GitHub.
#article #read
In reply to nevent1q…qqet
_________________________
TikTok Deal Done And It’s Somehow The Shittiest Possible Outcome, Making Everything Worse | Techdirt
https://www.techdirt.com/2025/12/19/tiktok-deal-done-and-its-somehow-the-shittiest-possible-outcome-making-everything-worse/
via Nostr@cxplay_clip
In reply to nevent1q…qqet
_________________________
TikTok Deal Done And It’s Somehow The Shittiest Possible Outcome, Making Everything Worse | Techdirt
https://www.techdirt.com/2025/12/19/tiktok-deal-done-and-its-somehow-the-shittiest-possible-outcome-making-everything-worse/
via Nostr@cxplay_clip
