Sublime
An inspiration engine for ideas
Flowith 2.0 知识库 + Deepseek R1 ,输出效果真不错。(注:视频有加速)。 @flowith_ai @deepseek_ai https://t.co/HFm66NpQ3w
Wang Shuyix.com
和菜头转述西乔的这段提示词,太「邪恶」了。如果你觉得自己某篇文章写得很好,飘了,可以直接把文稿放在提示词后面丢给反思式大模型(例如 deepseek r1, o1 pro 等),然后你的「症状」大概率「一剂而愈」。 https://t.co/Wu83XwF5CV
各位大佬,有没有好看的静态博客主题推荐,2025 年想把博客搞好看一点。
A junior software engineer came up with this idea that we could offer people kind of an all-you-can-eat buffet of fast, free shipping.
David M. Rubenstein • How to Lead: Wisdom from the World's Greatest CEOs, Founders, and Game Changers




What does DeepSeek R1 & v3 mean for LLM data?
Contrary to some lazy takes I’ve seen, DeepSeek R1 was trained on a shit ton of human-generated data—in fact, the DeepSeek models are setting records for the disclosed amount of post-training data for open-source models:
- 600,000 reasoning data [1]
- 200,000 non-reasoning SFT data [2]
- human prefere... See more