Keep on to blur preview images; turn off to show them clearly
大家感恩节快乐，MkDollar网站今天内测上线啦！ 网址：https://t.co/QzhUwYuD1k MkDollar的含义就是挣美刀，目前已完成的功能有： 功能1：用户可以在后台提交自己的产品，查看产品的DR值、流量趋势和分布，将来会支持监控DR变更发送通知等功能。 功能免费，提交只要输入域名，信息抓取纯自动化。
🔥 The best AI SaaS boilerplate - https://t.co/VyNtTs0jSX 🚀 The best directory boilerplate with AI - https://t.co/wEvJ1Dd8aR 🎉 https://t.co/bh1RxeERuY & https://t.co/zubXJCoY92 & https://t.co/tfQf8T7gGF
离谱，晚饭点的小满手工粉的外卖，商家怕汤凉，里面有个自热包，跟自嗨锅自热米饭那种一样。倒点水开始加热，还没热完，厨房突然响起尖锐的报警声，然后米家推送天然气泄露的消息。顿时吓尿，立刻以最快的速度开门开窗，然后检查煤气是否关紧。排查一圈没看到什么问题，也没闻到什么异味。通风一会后报警声也没了。我就觉得是不是小米的 bug 。然后就关门关窗，准备吃粉，还没吃到嘴，又开始报警。这下是真的慌了。又把门窗打开，再三检查还是没问题。就在我想要不要找物业的时候，突然觉得该不会跟这自热包加热有关吧？因为其他也没什么特殊的。上小红书一搜，还真是，很多人遇到同样的问题，吃自嗨锅引起天然气报警。震惊！然后把天然气监控的电源拔了，接着吃粉。
脚踏实地做梦/独立开发/降临派/新手奶爸👨🍼
2/n: 'I am so dumb' moment (comparable to the 'the aha moment' of DeepSeek R1) The most important technological breakthrough in DeepSeekMath-V2 paper is not the IMO Gold level performance!!!! Then what is is? It the ability to grant model capabilities to reliably verify its own sampled generation. This has been very hard for LLMs (even the reasoning ones) Quote: "When a proof generator fails to produce a completely correct proof in one shot – common for challenging problems from competitions like IMO and CMO – iterative verification and refinement can improve results (to an extent). This involves analyzing the proof with an external verifier and prompting the generator to address identified issues. However, we observed a critical limitation: when prompted to both generate and analyze its own proof in one shot, the generator tends to claim correctness even when the external verifier easily identify flaws. In other words, while the generator can refine proofs based on external feedback, it fails to evaluate its own work with the same rigor as the dedicated verifier. This observation motivated us to endow the proof generator with genuine verification capabilities." @gm8xx8 @teortaxesTex @rohanpaul_ai @ai_for_success
3/n: DeepSeekMath-V2 model was literally threatened not to cheat. You can read it in the prompt template. Liang Wengfeng is a strict parent!!!
RT @wei_xiao949736: @dyng692058 @yihui_indie 现身说法！熠辉老师的课真的太赞了，我一个零基础的小白，通过两个月的时间已经做了几个小网站了。这是我目前见过编程课中，最友好，最有性价比的课程了。 而且熠辉老师还一对一的给指导和答疑，这个价…
我的AI编程课(https://t.co/HVZn3ItASW) |B站up主 | 分享创造 + 无限迭代ing
My current, not particularly novel stance on AI coding: AI is amazing technology. I pair with AIs pretty much every time I touch code now, and I can’t imagine the idea of working without it. Having a researcher, thought partner, and code generator that knows syntax and patterns of every programming language is incredible. Responsible coding with AI makes me a much better, more productive developer. But to those who want to vibe code an entire complex SaaS application, I would simply say—try it. Go build your app with AI. See how the codebase evolves over time and witness things get messier, harder, and more error prone day after day after day. Six months in, you are going to seriously regret not paying more attention to the code.
Went from CTO to solopreneur, now building an empire one business at a time. Growing https://t.co/o8SD3o1h6p, https://t.co/NVUx34raAP, and https://t.co/2uIpQjuNAC
1/n: What most are missing is that the DeepSeek team has cracked the much feared and sought self improvement loop for the most complex domain for LLMs - proof generation for difficult problems. Will be interesting to see where the loop breaks. Some body wake up Eliezer Yudkowsky! @teortaxesTex
2/n: 'I am so dumb' moment (comparable to the 'the aha moment' of DeepSeek R1) The most important technological breakthrough in DeepSeekMath-V2 paper is not the IMO Gold level performance!!!! Then what is is? It the ability to grant model capabilities to reliably verify its own sampled generation. This has been very hard for LLMs (even the reasoning ones) Quote: "When a proof generator fails to produce a completely correct proof in one shot – common for challenging problems from competitions like IMO and CMO – iterative verification and refinement can improve results (to an extent). This involves analyzing the proof with an external verifier and prompting the generator to address identified issues. However, we observed a critical limitation: when prompted to both generate and analyze its own proof in one shot, the generator tends to claim correctness even when the external verifier easily identify flaws. In other words, while the generator can refine proofs based on external feedback, it fails to evaluate its own work with the same rigor as the dedicated verifier. This observation motivated us to endow the proof generator with genuine verification capabilities." @gm8xx8 @teortaxesTex @rohanpaul_ai @ai_for_success