Your curated collection of saved posts and media
π₯ Two days. Hundreds of #AI practitioners. One community. #PyTorchCon North America is coming to San Jose, California, October 20-21. Register by July 31 to save $400 & join the conversations shaping #OpenSource AI. ποΈ https://t.co/AVHdaIFT20 https://t.co/34wK6B9hyD
The Keras team is gonna be at the AI Systems DevLabs in Sunnvale tomorrow. Come say hi if you're around! https://t.co/tHtGXd53fC
For our free newsletter this week, we write about the humanoid robot safety race. β¨@IrenaCronin and I write this newsletter every week. Β Humanoid robots are moving closer to real-world use, but safety will matter more than impressive demos. To be trusted in factories, warehouses, hospitals, and public spaces, robots need strong testing, safety software, clear standards, and predictable behavior around people. The companies that prove their robots can operate safely and reliably will have the real advantage. Read and subscribe: https://t.co/HHwYy7NoAl
1-bit GLM-5.2 GGUF vs. Claude 4.8 Opus vs. GPT-5.5 We gave 3 models the same prompt and compared one-shot outputs. The 1-bit GLM-5.2 GGUF ran locally on a Mac Studio M3 Ultra with 256GB RAM at ~21.6 tok/s. Which output do you like best? GGUF: https://t.co/BMkxswdj5N https://t.co/UoXsCSh4Gn
GLM-5.2 can now be run locally!π₯ The 2-bit model retains ~82% accuracy after we shrunk it from 1.51TB to 238GB (-84% size). Run on a 256GB Mac or RAM/VRAM setups. GLM-5.2 is the strongest open model to date. Guide: https://t.co/bI7FeeKHDd GGUF: https://t.co/BMkxswdj5N https:/
Live Stream: Welcome to open source AI Lots of new folk are starting out on their journey with open models. Come join our livestream with all your questions about local models, open coding agents, and owning your AI. Thursday 8am PST/ 5pm CEST HF X and YT https://t.co/UlLUTXrsZb
Todayβs mission includes a demo of a new vehicle that will enable affordable, routine access to the microgravity environment for scientific research and in-space manufacturing. After demonstrating controlled flight, the spacecraft will splash down in the Pacific Ocean https://t.co/NLwhigtSWC
Microsoft Teams just crossed a line I did not expect this year. It now runs an AI employee that does the work, not one that answers questions. Here is what is new: https://t.co/XyNjkfSuly
Here is the part most people have not seen. Once it is in your Teams or Slack workspace, it starts coming to you. It remembers what our team has already done, understands what we are working toward, and moves before every instruction is spelled out. Screenshot below. https://t.co/xZSsq8Hg8Z
Try free at https://t.co/u5rYmO58TP. $100 in credits, no card. Tracked link: https://t.co/M7jYo8EBdu

Thanks @lablabai #Ai #Agents #Aiagents https://t.co/yls08NERwe
History suggests that competition alone rarely creates stability. During the Cold War, trust came from verification, not goodwill. The challenge with AI is that we're still debating what exactly should be verified. That's a policy problem as much as a technology problem. https://t.co/RGtRWGpbmB

https://t.co/r5YBxKIzbf
AWS loves startups! At NY Summit, established founders and aspiring ones convened at the Startup Zone. Building is easier (and more fun) when we do it together. We have the tools, the programs, and the people to take your startup to the next level. https://t.co/o3ZRXUrpe5
@ReutersJapan https://t.co/ENIfMqsITW
On my way to 9M impressions on X! Thanks everyone for keeping it real against AI hype! Last 3 months have been a blast. Answering rumours: β’ No, Iβm not an AI. β’ Realist: neither pro nor anti. Become a subscriber: exclusive access and perks! Eg: send over your AI research
Most AI code review tools look at one repo at a time. But the bug usually isn't in the code that changed. It's in what that change quietly breaks three repos away. @QodoAI just shipped Cross Repo Review to solve this. I tested it on my own repos. Here's what it caught. https://t.co/p73GYRQ3Hh
Modern teams work across dozens, sometimes hundreds, of repos. Change a shared utility, an API contract, or a data schema in one repo, and you can silently break consumers in another. Single-repo review misses this entirely. So the breakage shows up post-merge, in production. https://t.co/kCz5nWVi0J
Cross Repo Review maps how your repos depend on each other. Then on every PR, it reads the related repos and surfaces downstream consumers, breaking changes, and blast radius before you merge. It tracks four kinds of dependencies: Code, Service, Data, and Pipeline. https://t.co/sDCAwuFmfV
I tried this on my own setup. I run two repos: a Claude orchestrator backend and a desktop Electron app built on top of it. I renamed one field in the backend. A clean, consistent rename. Single-repo review saw nothing wrong. I ran that PR through Cross Repo Review. Here's the finding that surfaced:
Cross Repo Review flagged it: that rename silently breaks the desktop app in the other repo, which still expects the old field name. Every command between them would hang and time out. I never touched the app. The diff looked safe in isolation. The blast radius told a different story.
Start a free trial: https://t.co/nCJQvaqay1 Thanks, @QodoAI, for the partnership on this post.
On my way to 9M impressions on X! Thanks everyone for keeping it real against AI hype! Last 3 months have been a blast. Answering rumours: β’ No, Iβm not an AI. β’ Realist: neither pro nor anti. Become a subscriber: exclusive access and perks! Eg: send over your AI research paper for independent analysis. https://t.co/IBAYnVNves Trailblazing the manifold.
Goldman reckons that AI will create about $10trn of discounted value for the world, or up to $22trn. But markets have priced in $27trn of additional value creation. So the size of the "bubble" has now reached $5trn or up to $17trn... https://t.co/ggCK6f75Xz
Nice to see the Drivetrain Approach getting some love again! :D https://t.co/XcOjrJ0As3
Did a close read of @jeremyphoward et al's Drivetrain Approach Define goal β Consider levers β Consider data β Implement assembly line Most important is having actions that fulfill the objective Predictions are more data, not actions Key takeaways π§΅
Published by @OReillyMedia here: https://t.co/sNMLLCSHAa
Maybe the most revealing thing about AI companions isn't the AI. It's what they tell us about the human need to be heard, understood and connected. That's a very old story. https://t.co/YNf692QvkT
https://t.co/ha4FkMtp2q
@Ferbin08 Why I put more lights on the walls. AI can see the edges well. https://t.co/85mtG6lw6J
i dont think people realize how big this is Seedance 2.5 now can generate 30s 4K videos from one prompt.. with up to 50 ref.. one click filmmaking is here https://t.co/ltsoz3zgSD
The second official Engineer Worldβs Fair Hackathon with @aiDotEngineer Join us for 48 hours of building at the frontier of recursive self-improvement (RSI): systems that learn from their own outputs and become more capable through continuous iteration. Excited to be partnering with @GoogleDeepMind, @MongoDB, @MiniMax_AI, @LiveKit, and @digitalocean. π $10,000+ in prizes and partner credits June 27β28 hosted at @SHACK15sf Apply belowπ
how it feels to code[x] https://t.co/8Uyhk4IEhn
Post your best codex billboard
how it feels to code[x] https://t.co/8Uyhk4IEhn
No, Elon Musk and DOGE did not cause "millions of deaths of children." Anthony Fauci, however, very much did. https://t.co/UAiYJ9U9hc
Today, on my final day as Director of National Intelligence, Iβm releasing never-before-seen communications and documents exposing howΒ Dr.Β FauciΒ provided millions in US taxpayer dollars to fund dangerous gain-of-function research at the Wuhan lab, worked withΒ politicizedΒ elements