Your curated collection of saved posts and media
The stage is set. The tech is ready. Are you? 🚀 Join us tomorrow for #GoogleIO as we unveil the breakthroughs, tools, and innovations shaping the future of AI. Tune in live right here on @X from 10am PT: https://t.co/u4s3nkfrlT https://t.co/8DNJCycnf0
https://t.co/yGiqw0xbji
“favorable conditions never come” c.s. lewis https://t.co/ozyvXjhEWL
We're excited to announce we're starting a Journal Club. And our first meeting is scheduled for tomorrow! @__init_self will present her work, CorText: Brain-Language Fusion Enables Interactive Neural Readout and In-Silico Experimentation Tomorrow at 10:15am ET, join Discord! https://t.co/sObrmq0Vmw
We are starting a journal club in our open medical AI research community. The first talk is tomorrow! Join our Discord server, https://t.co/KIY8DYBCWE to participate!
We're excited to announce we're starting a Journal Club. And our first meeting is scheduled for tomorrow! @__init_self will present her work, CorText: Brain-Language Fusion Enables Interactive Neural Readout and In-Silico Experimentation Tomorrow at 10:15am ET, join Discord!
We're excited to announce we're starting a Journal Club. And our first meeting is scheduled for tomorrow! @__init_self will present her work, CorText: Brain-Language Fusion Enables Interactive Neural Readout and In-Silico Experimentation Tomorrow at 10:15am ET, join Discord! https://t.co/Gnl5IBTFQR
Our open medical AI research community is starting a journal club, our first talk is tomorrow. Join our Discord server, https://t.co/KIY8DYBCWE to participate!
We're excited to announce we're starting a Journal Club. And our first meeting is scheduled for tomorrow! @__init_self will present her work, CorText: Brain-Language Fusion Enables Interactive Neural Readout and In-Silico Experimentation Tomorrow at 10:15am ET, join Discord!
NVIDIA just released Nemotron CLIMB Proxy Models on Hugging Face Small decoder-only models (62M & 350M params) trained on 10T tokens for scaling law research, enabling prediction of larger model behavior without full-scale compute. https://t.co/Xtpa9UEiPz
for the facts, courtesy @RachelBitecofer: https://t.co/uQLjKKOyD7
4 levels of Hermes Agent setup: LEVEL 1: main agent You → Hermes Agent this is your main agent and your prototype area, where you test new workflows and refine them. it doubles as your orchestrator until you have something worth breaking out ---- LEVEL 2: specialized agents You → SEO Agent You → CMO Agent You → Ops Agent once a workflow is solid, break it out into its own agent with its own credentials, memory and scope. --- LEVEL 3: orchestrated team You → Orchestrator ↓ Specialist Agents bring the orchestrator back in. it now steers the company of agents you have built. ---- LEVEL 4: automated team Cron / Events → Orchestrator ↓ Agent Team add task lists so the team works async. cron and events fire jobs, the orchestrator routes them through the task bus, the team handles the work without you ---- take small steps, you DO NOT want to automate slop. if your output at level 1 is mediocre, you are about to scale mediocrity. 20 agents shipping low quality work at speed is worse than 3 shipping great work slowly. I would rather run fewer agents with better output than MAXXING the agent count and spitting out more of the same.
https://t.co/lrJAYG5iEE
Hermes Agent “Foundation Release” is a MASSIVE update. In just this release, @NousResearch added: - /handoff live session transfers - Native Windows support beta - OpenAI-compatible local proxy - DeepSeek V4 Flash FREE in Nous Portal - Grok + SuperGrok integrations - Native AI video generation - Faster browser automation - Background Computer Use - Cross-session Claude prompt caching - x_search for real-time X/Twitter search - Better multi-agent orchestration - Huge cold-start performance improvements Hermes is rapidly evolving from an AI agent into a true open-source Agentic AI Operating System. Full breakdown/demo: https://t.co/iljJztkeWs
Good thing we discovered this class M planet in the Nous system before it was too late https://t.co/b6YNteqXW1
OK, one more scene and crew confession. https://t.co/ZMOju5OMij
My colleagues wrote up a great post on using Goals in Codex. They go through when to use them, what changes when a Goal is active, and how to write Goals that give Codex a clear outcome, constraints and verification criteria. Also how we designed Goals at the architecture level if you’re curious. https://t.co/QQfjW2EbPO
Grok Build has Command Palette with Ctrl+P https://t.co/OuikQpUmMZ
Elon Musk’s lawyer, Marc Toberoff, on today’s verdict in the OpenAI case: “I have a one-word reaction: Appeal. This war is not over. We firmly believe what happened with OpenAI was wrong on a very basic level that you can't raise millions of dollars in a publicly subsidized charity, and when it suits you, just turn into a for-profit operation where the officers and directors of the charity enrich themselves to the tune of billions, and that's what actually happened here, and that's just wrong."
The New Third Rail of American Politics: Americans don't agree on much anymore, but they agree that they hate AI data centers Millions of Americans increasingly believe they are being asked to sacrifice affordability, stability, and local control in the service of a technological revolution designed primarily to steal their jobs.
llama.cpp with MTP support makes local models fast enough to use as daily drivers 🚀 Qwen3.6-27B dense generation (on A10G): From 25 tok/s → 45 tok/s (+78%). Two flags on llama-server: --spec-type draft-mtp --spec-draft-n-max 2 https://t.co/hhslKpLE71
llama.cpp adds MTP for the Qwen3.6 family This is a significant milestone for the local AI ecosystem. The performance jump with these changes is massive and elevates local inference on commodity hardware further. Special thanks to Aman Gupta for leading this development! https
Latest `hf-mem` now breaks down Mixture-of-Experts (MoE) memory estimations into base weights, routed experts, and KV cache. Useful for reasoning about residency footprint and serving trade-offs before picking a parallelism strategy for inference. More in the thread 🧵 https://t.co/PWhqvVvmIZ
Perplexity Computer now has an Artifacts page, with a shortcut in the side panel. You can view and pin apps, documents, etc. that Computer has made for you, plus any that have been shared with you. Helpful for quick access to live finance dashboards that your team creates. https://t.co/wUt7JaQved
We've added an Artifacts tab to Perplexity, your one-stop shop for everything you've created with Computer. Access your apps, PDFs, artifacts shared with you, and anything you've previously viewed, all in one place. Pin your favorites for even faster access. https://t.co/XZ33lqT
This open source multi agent system is great for planning out your viewing of Google IO, starting tomorrow. It planned out my whole week and added all the sessions I cared about to my calendar. Thanks to LobeHub for sponsoring my newsletter: https://t.co/HHwYy7NoAl that ships every Tuesday.
Meet your Chief Agent Operator in LobeHub now. You don't manage agents. Your CAO does. LobeHub hires agents from a 273K-skill marketplace, runs them in the cloud 24/7, and reports through IM apps you already live in. You run the strategy. We run the agents. https://t.co/Fmsc
Good time to walk around with my 'hyperspectral' glasses enjoying the flowers :) here is right eye vs left eye, more in a thread https://t.co/94H24evE0F
Some flower pigments really look blue vs purple https://t.co/wonXwNoMBZ
Harder to capture this one with the phone - to my eyes, on slice of the spectrum showed a plain yellow flower, the other showed strong radial gradients and patterns (Ok probably enough flower spam for now :) https://t.co/fTaAywyOiY
xAI has Released a blog on Skills https://t.co/WyZsvXUYw0
@mjwoo94 @Carles_Reina @Chris_Orlob @austinh___ @akshat_b @karpathy @AndrewYNg @DrJimFan @lilianweng Yes. The best are on this list: https://t.co/wAjs9SAZfe And this site watches 50,000 here on X and picks the best: https://t.co/kiuZ7QXLzb And all 50,000 in the community are on these lists: https://t.co/fasUz7PuHq
I broke my own rule to never post about AI detection as it is fraught in many ways. The problem is that if you use AI a lot, you know AI writing on sight, which makes the difficulty of objectively proving that AI use to others very frustrating https://t.co/YzjUfaAkjJ
Judge Yvonne Gonzalez Rogers basically waved off Elon’s OpenAI appeal, saying the jury’s statute of limitations ruling is tough to overturn. An activist judge letting Sam Altman skate after he hijacked a nonprofit charity — originally pledged to benefit all humanity — and turned it into his $150B+ personal for-profit empire? That’s not justice. Stealing a charity for profit is not OK. You don’t get to rewrite the mission, pocket the upside, and hide behind “time limits” while betraying the public trust. OpenAI’s founding promise was destroyed in broad daylight. This makes zero sense.
This illustrates why the ruling by the terrible activist Oakland judge, who simply used the jury as a fig leaf, creates such a terrible precedent. She just handed out a free license to loot charities if you can keep the looting quiet for a few years!
What day is today? Pets on terminal day! 🐾🐕🐇 https://t.co/Vzw4eijYqy
