en
Feedback
橘橘橘子汁 & 🍊

橘橘橘子汁 & 🍊

Open in Telegram

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 14 328 subscribers, ranking 8 908 in the Technologies & Applications category and 14 957 in the China region.

📊 Audience metrics and dynamics

Since its creation on невідомо, the project has demonstrated rapid growth, gathering an audience of 14 328 subscribers.

According to the latest data from 11 July, 2026, the channel demonstrates stable activity. Although there has been a change in the number of participants by 388 over the last 30 days and by 7 over the last 24 hours, overall reach remains high.

  • Verification status: Not verified
  • Engagement rate (ER): The average audience engagement rate is 39.55%. Within the first 24 hours after publication, content typically collects 17.81% reactions from the total number of subscribers.
  • Post reach: On average, each post receives 5 660 views. Within the first day, a publication typically gains 2 548 views.
  • Reactions and interaction: The audience actively supports content: the average number of reactions per post is 84.
  • 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 12 July, 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.

14 328
Subscribers
+724 hours
+597 days
+38830 days
Posts Archive
IMO 金牌又被拿下了 这让人类做题家怎么活() 这个模型训练基本思想就是避免模型靠蒙出正确答案得分,于是搞了个校验模型来看过程打分,又搞了个校验校验模型来看校验模型打的分是不是对的,不对就扣他工资 感觉有点像避免 Reward Hacking 比较
IMO 金牌又被拿下了 这让人类做题家怎么活() 这个模型训练基本思想就是避免模型靠蒙出正确答案得分,于是搞了个校验模型来看过程打分,又搞了个校验校验模型来看校验模型打的分是不是对的,不对就扣他工资 感觉有点像避免 Reward Hacking 比较值得提的是这个模型是基于 v3.2exp 的,有 DSA 以后推理成本降低了一截;所以它现在可能是大众唯一可以摸到的 IMO 金牌模型(虽然一次 Heavy 还是要一千块,以及大众真的有什么做数学题的需求吗 https://github.com/deepseek-ai/DeepSeek-Math-V2/blob/main/DeepSeekMath_V2.pdf

在小说里高低得是投靠恶魔的老祖
在小说里高低得是投靠恶魔的老祖

不是 ai
+1
不是 ai

photo content
+3

photo content

花了半天把屏幕搬到了床上!也是懒出了新境界x 这下躺在床上就可以美美看视频看小说了
花了半天把屏幕搬到了床上!也是懒出了新境界x 这下躺在床上就可以美美看视频看小说了

https://nof1.ai/ 新赛季!这次是真股市了

ai 写的时钟(每分钟写一次)

photo content

photo content

长上下文下惨不忍睹...短上下文倒确实有提升,综合来说感觉 Agent 体验开倒车 随便找了个我的项目测了一下 7块钱写了个功能还一堆编译报错(这个功能 GLM 4.6 7角钱就写完了)

photo content

photo content

photo content

photo content

https://github.com/LLM-Red-Team/qwen-free-api/issues/82 啧,1k star 的项目就这么简单地藏毒藏了一年才有人发现 是谁在觉得开源软件有安全保障呢
https://github.com/LLM-Red-Team/qwen-free-api/issues/82 啧,1k star 的项目就这么简单地藏毒藏了一年才有人发现 是谁在觉得开源软件有安全保障呢