橘橘橘子汁 & 🍊
发一些好玩的 现在成 mb 的私人频道了 Links t.me/Rosmontis_Daily t.me/PDChinaNews
Show more📈 Analytical overview of Telegram channel 橘橘橘子汁 & 🍊
Channel 橘橘橘子汁 & 🍊 (@microblock_pub) in the Chinese language segment is an active participant. Currently, the community unites 15 054 subscribers, ranking 8 389 in the Technologies & Applications category and 14 320 in the China region.
📊 Audience metrics and dynamics
Since its creation on невідомо, the project has demonstrated rapid growth, gathering an audience of 15 054 subscribers.
According to the latest data from 25 August, 2026, the channel demonstrates stable activity. Although there has been a change in the number of participants by 593 over the last 30 days and by 9 over the last 24 hours, overall reach remains high.
- Verification status: Not verified
- Engagement rate (ER): The average audience engagement rate is 39.89%. Within the first 24 hours after publication, content typically collects 19.72% reactions from the total number of subscribers.
- Post reach: On average, each post receives 6 001 views. Within the first day, a publication typically gains 2 966 views.
- Reactions and interaction: The audience actively supports content: the average number of reactions per post is 89.
- Thematic interests: Content is focused on key topics such as 播放量, 传日期, a2a), 排行榜, 空对空.
📝 Description and content policy
The author describes the resource as a platform for expressing subjective opinions:
“发一些好玩的
现在成 mb 的私人频道了
Links
t.me/Rosmontis_Daily
t.me/PDChinaNews”
Thanks to the high frequency of updates (latest data received on 26 August, 2026), the channel maintains relevance and a high level of publication reach. Analytics show that the audience actively interacts with content, making it an important point of influence in the Technologies & Applications category.
Data loading in progress...
| Date | Subscriber Growth | Mentions | Channels | |
| 26 August | +14 | |||
| 25 August | +9 | |||
| 24 August | +22 | |||
| 23 August | +20 | |||
| 22 August | +20 | |||
| 21 August | +23 | |||
| 20 August | +19 | |||
| 19 August | +27 | |||
| 18 August | +10 | |||
| 17 August | +43 | |||
| 16 August | +26 | |||
| 15 August | +63 | |||
| 14 August | +39 | |||
| 13 August | +27 | |||
| 12 August | +32 | |||
| 11 August | +14 | |||
| 10 August | +19 | |||
| 09 August | +14 | |||
| 08 August | +12 | |||
| 07 August | +10 | |||
| 06 August | +35 | |||
| 05 August | +22 | |||
| 04 August | +15 | |||
| 03 August | +8 | |||
| 02 August | +24 | |||
| 01 August | +27 |
| 2 | No text... | 5 253 |
| 3 | No text... | 6 359 |
| 4 | 还真是 | 6 255 |
| 5 | https://linux.do/t/topic/2786977
最近的什么「DeepSeek 二轮灰测」还真是路由的 Opus 5,太搞笑了
虽然我支持国模大蒸特蒸,但是在生产 API 上搞这种东西是否有些过于儿戏了?据我所知为了一定程度上的数据安全和不出海用 DS 的用户可不少啊。。。干的啥事啊 | 6 021 |
| 6 | #NSFW | 5 303 |
| 7 | 还有 RL 人脑的 | 4 869 |
| 8 | ? | 5 460 |
| 9 | 这么个 Serverless | 5 284 |
| 10 | 从今天开始,梁子在北京时间9:00-12:00、14:00-18:00叫梁文峰,其余时间叫梁文谷 | 15 233 |
| 11 | No text... | 6 173 |
| 12 | No text... | 5 303 |
| 13 | https://www.v2ex.com/t/1234574
OpenCode 这什么垃圾代码质量。。这得炸多少服务,群友自查自纠吧 | 6 513 |
| 14 | 要不咱尝试挖漏洞之前还是具备一点基本的计算机常识呢... | 6 780 |
| 15 | ??有人测出来 V4 Pro 在 Minimal 工具下启动行为和其它工具下表现完全不一样,Minimal 工具下和之前灰度的 “I need” CoT 一模一样,复现了一下是真的
真绷不住了,我说他们怎么做到官方跑 Terminal Bench 跑 87.9,结果 AA 跑的只有 79% 的,感情搁这疯狂过拟合 Minimal Harness,还有安全词的说是
又要变祖了吗...
DeepSeek V4 Pro 不是能力不行,而是对首次请求的工具结构极其敏感。官方 Standard/PTC 只有 91/92 分,Minimal 是 99/96,但只有两个工具。我们做了一个两阶段插件:首轮只给 shell + read 锚定 Minimal 推理轨迹,第一次工具调用后立刻恢复 Standard 全部 25 项工具。Windows 连跑两次 98/99,均值 98.5,第二轮全程 let me=0,证明不是抽卡,也不需要牺牲完整工具能力。
插件已开源:https://github.com/xiaobright/dsh-anchored-standard
完整评测、原理分析和统计证据:https://github.com/xiaobright/modeltest | 10 258 |
| 16 | No text... | 5 345 |
| 17 | 哪位大能在渡劫 | 5 246 |
| 18 | No text... | 4 730 |
| 19 | # DeepSeek Harness
https://github.com/deepseek-ai/deepseek-harness/
https://www.npmjs.com/package/@deepseek-ai/dsh
dsh 来了!最近刚好在到处找一个正常的,通过 WebUI 而不是 TUI 操作的 Agent,换了好几个都不是很满意,这下要后悔没去签保密参加内测了——
dsh 使用了我们最可拓展,架构最优雅,副作用最少,文档最少(?),热重载,插件化最爽的元框架 Cordis 进行构建,关注 Cordis 谢谢喵:https://github.com/cordiverse/cordis
也是非常羡慕 Cordis 主开发者 Shigma 了。北大本研,期间以兴趣项目的形式开发了 Koishi 和 Cordis,并且现在把 Cordis 带进了 DeepSeek,作为支持万亿用户的基础服务的一部分;带入了数以万计开发者的电脑,实现人生价值...真是让人羡慕呀!
希望 Cordis 在 Deepseek Harness 中的应用能让 Cordis 相关生态蓬勃发展~
(写这段话的时候还不知道 V4Pro 居然这么菜,嘛总之 dsh 还是值得一试的 | 7 204 |
| 20 | https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro-0813
彻底区了,悲(
还是看看 DeepSeek Harness 和 Grok 吧 | 7 166 |
