距离Anthropic发布Claude的“模型上下文协议”（MCP）才一天，就像是AI的万能接口。 现在，AI连接工具和数据变得异常简单。 人们已经开始疯狂地用它来完成各种工作。 以下是10个令人惊叹的例子：

这两天付鹏的文章刷屏了，里面提到了分配是个大问题。在AI时代这个问题似乎更无解了。 AI对于工作效率的提升巨大，意味着每个AI进入的行业，企业都可以用更少的岗位换取更多更好的产出。 目前受这类威胁的还只是中产白领，10年内TSLA的Optimus量产后，估计不少蓝领岗位也会被替换掉。

2023-2024对我来说应该连在一块看，可能是生命中最兼具“广度”和“厚度”的两年：横祸、斗争、失意、恐惧、探索、受助、觉醒，最终归于平静和快乐。有懂占星的朋友说男人三十土星回归，懂八门遁甲的算出青龙折足招灾失财。我只知道一定要感激这个赛季，感谢一出又一出drama提升了我生命的认知。2025，阿门

🚀 Introducing DeepSeek-V3! Biggest leap forward yet: ⚡ 60 tokens/second (3x faster than V2!) 💪 Enhanced capabilities 🛠 API compatibility intact 🌍 Fully open-source models & papers 🐋 1/n

How to train a 670B parameter model. Let's talk about the DeepSeek v3 report + some comparisons with what Meta did with Llama 405B

使用 reasoning model 的新范式：要思想对话（dialogue），不要无脑闲聊（chat） 昨天晚上和o1 的一段对话，4 个来回；虽然是在手机上敲的，但也得分行分段有标点。 o1 说完后，我会仔细阅读。然后把新想法用文字组织起来。往往是 一到几分钟后。

DeepSeek是一个战术上比较成功，但战略上很蠢的例子。很取巧地汇报最后一次训练成功的单次炼丹成本，战术上赢得了掌声、关注、国内资本青睐。战略上使得美国对东方大国的警惕大增。以后连H20都别想买了，而国内显卡没法用。这可能导致东方大国在未来很长一段时间军备竞赛落后，烧了国内其他AI公司的路

🚀 Meet Kimi-VL and Kimi-VL-Thinking! 🌟 Our latest open source lightweight yet powerful Vision-Language Model with reasoning capability. ✨ Key Highlights: 💡 An MoE VLM and an MoE Reasoning VLM with only ~3B activated parameters 🧠 Strong multimodal reasoning (36.8% on

Why does RL work for enhancing agentic reasoning? This paper studies what actually works when using RL to improve tool-using LLM agents, across three axes: data, algorithm, and reasoning mode. Instead of chasing bigger models or fancy algorithms, the authors find that real,

This Aviral Kumar lecture is AMAZING.

🚀 No More Train–Inference Mismatch! We demonstrate bitwise consistent on-policy RL with TorchTitan (training) + vLLM (inference) — the first open-source run where training and inference numerics match exactly. It only takes 3 steps: 1️⃣ Make vLLM batch-invariant (same seq →

Glad to introduce our research on understanding the "mathematical principles" behind reinforcement learning (RL) with LLMs, and how stabilization techniques work 🧠 📄 huggingface.co/papers/2512.01… 👇 Thread below

Hot topics in RL On-policy RL Everyone faces training rollout mismatch - Truncated importance sampling: fengyao.notion.site/off-policy-rl#… - IcePop: doubled-ended importance ratio clipping - Rollout Routing Replay: arxiv.org/abs/2510.11370 Efficient rollout systems design PipelineRL:

Single rollout RL for multimodal RL. It is similar to the previous approach of single rollout RL (arxiv.org/abs/2509.13232) but they were able to stabilize this only after applying advantage shaping with an entropy bonus.

I'm Boris and I created Claude Code. Lots of people have asked how I use Claude Code, so I wanted to show off my setup a bit. My setup might be surprisingly vanilla! Claude Code works great out of the box, so I personally don't customize it much. There is no one correct way to

A few random notes from claude coding quite a bit last few weeks. Coding workflow. Given the latest lift in LLM coding capability, like many others I rapidly went from about 80% manual+autocomplete coding and 20% agents in November to 80% agent coding and 20% edits+touchups in

x.com/i/article/2025…

x.com/i/article/2025…