x.com/i/article/2021…

Introducing MolmoSpaces, a large-scale, fully open platform + benchmark for embodied AI research. 🤖 230k+ indoor scenes, 130k+ object models, & 42M annotated robotic grasps—all in one ecosystem.

New art project. Train and inference GPT in 243 lines of pure, dependency-free Python. This is the *full* algorithmic content of what is needed. Everything else is just for efficiency. I cannot simplify this any further. gist.github.com/karpathy/8627f…

x.com/i/article/2016…

residual encoding of high dimensional continuous data with temporal structure mentioned!!!

We are about to enter the age of simulation.

Predicting latents is both simple AND scalable. Reconstructing pixel seems unnecessary and against human intuition. The same idea motivated us to look at JEPA-like representation learning from behavior data in RLDP (openreview.net/forum?id=jdL6W…) ICLR 2026 #ICLR2026

AlphaFold won a Nobel Prize and is being used to design new drugs. How do we get more AI models like it? My new piece introduces "Natural Law Models" (NLMs): AI models trained on scientific data to uncover the rules governing nature. I also lay out what policymakers and funders

Congrats on the launch Simile ! (and I am excited to be involved as a small angel.) Simile is working on a really interesting, imo under-explored dimension of LLMs. Usually, the LLMs you talk to have a single, specific, crafted personality. But in principle, the native,

这个项目好像有意思？ 它把大模型预训练的知识都写成skills了， 这意味着初中辍学生，都可以做自己的预训练。 搞大模型根本不需要博士学位.... github.com/Orchestra-Rese…

牛逼，斯坦福小镇论文的作者 Joon 创业了，公司叫 Simile，融了 1 亿u，做斯坦福小镇的延伸，利用 agent 模拟人类行为。 我很看好 Joon，他的论文我基本都读了。最近很火的 AI 论坛 Moltbook，其实 Joon 在22年的论文就提出过原型 “SimReddit”，agent 论坛。 期待 Simile 的产品！

the wealth of knowledge in this post is crazy. what a valuable resource

Given that this topic 👇 of secularism in China, unexpectedly, seems to stir passions (even though I've unsurprisingly seen many pretty ignorant takes), let me add a small something. The man who probably best summed up this topic - ever - was Pascal, the legendary 17th Century

我参与了中文版翻译工作。希望把关于 AI 时代经济与治理的讨论带给更多中文读者，欢迎大家指出任何翻译/术语建议。虽然AI已经能做大部分翻译任务，但翻译过程中还是有很大量的人类对齐工作，尤其一些概念中/英差距很大，又要兼顾原作者表达的语气和方式，整个工作体验还是很有意思的。

Excellent article - having advised the UK Gov on AI for Science I cannot rec more V good push back against the "LLM solves everything" school-of-thought Succinctly explains steps for more AlphaFolds (new datasets, training scientists and more compute) Also calls out OpenBInd!

x.com/i/article/2022…

Introducing Nested Learning: A new ML paradigm for continual learning buff.ly/iBqCgnR #AI #MachineLearning #DeepLearning #LLMs #DataScience

来自另一条时间线的记忆？🧠 与Rizwan Virk的深度对话，探讨“曼德拉效应”，以及为什么如此多人记得在这个现实中从未发生的事件。 在《The Simulated Multiverse》中，Rizwan 探索了一个问题：这些是否是不同模拟运行之间“泄漏”的记忆？ 🎥 点击观看完整对话

The golden age of discovery is here garryslist.org/posts/demis-ha…

Latent space discrete diffusion or EBM sampling is the future. Default mode thinking is not sequential.