Your curated collection of saved posts and media
@halexam Yea check out this video: https://t.co/D41UhRabSZ
.@HuggingFace did a nice explainer video of Hermes Agent's architecture, check it out here: https://t.co/wqWargH39p
A #Vivatech, le chercheur français @ylecun Yann Le Cun défend le droit des États à être maîtres de leur destinée technologique 👉"Ils veulent tous assurer leur #souveraineté en matière d'IA, et je pense qu'ils ont raison" https://t.co/f2XgviSE0G
Yann Le Cun porte un constat juste : les IA génératives, comme ChatGPT, Claude ou Gemini et autres, sont extrêmement habiles pour manier les langages, mais elles ne sont pas connectées directement à la réalité physique. Les IA génératives n’ont accès que partiellement à une dimension essentielle de notre univers. Le projet de Yann Le Cun, de créer des world models, avec sa nouvelle start-up, se révèle particulièrement intelligent. Ces modèles inédits seraient en capacité de cerner et modéliser la réalité de notre monde, avec davantage de précision et de discernement. Il en résulterait une meilleure faculté de traiter l’information et surtout une lecture immédiate des dynamiques entropiques, qui façonnent et structurent notre réalité. Pour une fois, nous pouvons engager une révolution technologique. L’Europe peut s’imposer dans le domaine prometteur des world models. Le succès sera au rendez-vous si l’Union européenne se montre enfin capable de mobiliser de puissants investissements.
🔴🗣️A quoi ressemblera notre vie dans 10 ans ? "Tout le monde sera une sorte de patron d'une équipe de personnes virtuelles qui nous aiderons dans notre vie de tous les jours", selon Yann Le Cun , fondateur de la start-up AMI Labs. #JT20h https://t.co/f8CVwNGFSM
You asked, now supported. Thanks @Teknium! https://t.co/rVZPs2CWo0
@0xb3llatrix @waefrebeorn https://t.co/xjuQGhVNwI
MOSS-TTS Local Transformer v1.5 is here. Clone any voice. Speak any language. Hear every detail. 30+ languages, 48 kHz clarity. https://t.co/BDUvgj6ZNK
how it started (2 years ago) vs. how it's going https://t.co/gg0x7k8psk
Announcing a new division of Midjourney called "Midjourney Medical" https://t.co/c14YcO6yaU
Sam Altman walked into YC, offering founders $1M in free tokens for a slice of equity. Jason's advice: run. OpenAI is watching your token usage, ranking the winners, then shipping the top 5 as native features. Get off the frontier models, use open source, and own your data. @Jason
@alexellisuk I use pi for the harness. Haven't observed looping - I think the key is to limit the reasoning budget. I saw your blog post today. I would recommend to try to load the MTP model to the side in Q4_0 - should give you extra ~15% perf. See my config for that: https://t.co/byftAF17un
Many young graduates are not worried about learning AI. They are worried about competing with it. That helps explain why some commencement speeches celebrating AI are being met with skepticism rather than applause. Technology may move fast, but leaders still need to understand how change feels to the people experiencing it.
https://t.co/SRqBhYtLHD
𝕏 is the #1 news app in the UK. This is exactly why the BBC, The Telegraph and other UK legacy media keep pushing negative fake attacks on Elon Musk. They are getting destroyed in the rankings. The British public has rejected them. Legacy media is finished. They are lashing out in pure panic because they lost control. 𝕏 is winning because it gave people what they actually wanted - truth.
Excited to share that my talk, “Memory Harnesses for Long-Running Research Agents,” was accepted for @aiDotEngineer SF. I’ll be talking about making long-running research agents @SakanaAILabs more reliable, inspectable, and useful over time. The event had a <5% acceptance rate, so I’m especially grateful for the chance to present. Thank you @swyx & team!
Congrats @midjourney and @DavidSHolz. It has been a while since a new product introduction has gotten this kind of attention without paying any influencers. Look at X’s feed tonight for me. And why? Well the product will save many lives, yes. But that only explains a small bit why we are all excited tonight. For two it was kept pretty secret. One of the tightest kept secrets I have ever seen. Yeah one guy here on X figured it out. But no one in the audience knew. I asked many before it began. And David told me before he started he didn’t know the venue, which was an art museum, had strict rules regarding using mobile phones. That forced the audience to really pay attention. And not to capture it. Steve Jobs forced the same rules. But everyone knew they were gonna see a new phone when he introduced the iPhone. This caused a massively different vibe. Plus Midjourney had long been a leader in the AI space and his company before that. During his presentation he revealed how much AI that had that I didn’t even know and I was one of the few to cover it in his early days (Leap Motion). When we walked in David was sitting on stage. Several said hi. I did too and we caught up a bit. His accessibility is not to be missed. Every week he does spaces here on X and he answers everyone’s questions for hours. Everywhere I looked I saw VIPs from San Francisco and Silicon Valley. Met so many interesting people tonight including a couple building brain computer interface companies. They were there because of the mystery. His companies have been doing bleeding edge things with technology for decades. He didn’t seem scripted. Just talking about his life in a comfortable, humble way, that just gets you to listen more. Add it all up, and the beautiful videos and reveals, which applies a level of taste and intention that AI will struggle to match for years. One last thing. It is a company that makes humans better. It isn’t looking to force change. It made it so beautiful that you want to change. The spa is brilliant. I have had a full body scan at @prenuvo in an MRI machine. It was like going to an office building. Cold and lifeless with danger signs on the door. This is me getting that scan. The experience couldn’t be more different. The AI industry should study David deeply. There are many lessons to learn from what Midjourney announced tonight that I hope everyone takes to heart.
Grok Imagine 1.5 Grumpy cat talks https://t.co/oiZLOnJ2pd
"My favorite enterprise AI service is Cursor. It is an AI coder & every one of our engineers, 100% is now assisted by AI coders & our productivity has gone up incredibly. These companies are some of the fastest-growing companies in world & they address enterprise" — Jensen Huang https://t.co/qfjnU4BOjW
@crystalwizard https://t.co/KAn3clhlsE
This recent paper is super interesting... Using ultrasound imaging to generate images of the entire in vivo human cross-section in the abdomen and thighs Basically trying to do CT/MRI-like imaging but using ultrasound! Ultrasound imaging is highly underrated :) https://t.co/c
@gdb thanks! https://t.co/UKSVCHa3f3
@gdb Hey boss can you reply to my slack message about the rust foundation
never leaving this app https://t.co/S3JIT3fcI1
I don't often walk out of a product introduction thinking, "Wow, I just saw the world change for the better." I have one of the few pictures of @DavidSHolz doing the presentation. I'm very honored to be able to capture this image right before I got yelled at by security guards, because they don't allow photography at all in the space we were in. Forced everyone to turn off their phones. Breaking rules is necessary when you are witnessing history happening in front of you. Met @joshelman who was heavily involved in rebuilding Siri at Apple just after the event ended and we both were so excited by what we had both seen. Giddy even that we had seen it up close. A day that started with sadness and ended with a product launch unlike any other. It will be studied by business students for decades.
My best friend Brandon Wirtz, a brilliant AI pioneer, died at the age of 40 from colon cancer. The doctors missed it. This would catch it. I asked the first question because of Brandon. I love @DavidSHolz who just announced this new unloading scanner and spa that could save m
BREAKING: California Birth Rate Falls Below Replacement Level • California’s fertility rate hit a record low • It fell from 2.21 children per woman in 2007 to 1.48 in 2023 • That’s far below the 2.1 replacement level • By 2038, deaths could outnumber births in California • Public school enrollment could drop by about 630,000 students
a demo of the system. try it out here! https://t.co/kNkJjky02G https://t.co/VtCTYJ9b0u
a demo of the system. try it out here! https://t.co/kNkJjky02G https://t.co/VtCTYJ9b0u
The Anthropic shutdown exposed an uncomfortable reality. If critical AI systems can disappear overnight because of decisions made in another country, then access to AI is becoming a sovereignty issue, not just a technology issue. The question for many nations is no longer whether to use AI. It is how much control they want over the infrastructure that powers it.
https://t.co/rNLXJ59rme
Top trending paper on https://t.co/nYgeLha6ut today is "VibeThinker-3B: Exploring the Frontier of Verifiable Reasoning in Small Language Models" 🔥 A 3B model placed in the same performance tier as models like DeepSeek V3.2, GLM-5, and Gemini 3 Pro (!) https://t.co/tiNPW2DZcm
WHAT THE HELL is happening in AI? A 3B parameter model just put up coding benchmark scores in the same league as Claude Opus 4.5. 3 BILLION. The weights are on Hugging Face, anyone can test it. I genuinely don't know if this is a breakthrough or if the benchmarks are broken.
Top trending paper on https://t.co/nYgeLha6ut today is "VibeThinker-3B: Exploring the Frontier of Verifiable Reasoning in Small Language Models" 🔥 A 3B model placed in the same performance tier as models like DeepSeek V3.2, GLM-5, and Gemini 3 Pro (!) https://t.co/tiNPW2DZcm
It’s now easier to move local agents to the cloud so they can keep working with your laptop closed. Prompt Cursor from your phone, run many agents in parallel, and get back PRs with demos of their work. https://t.co/vuh5aZbH3e
Grok Build just got new update with new improvements, features, and bug fixes Release Notes: v0.2.55 Features: • Added option to fully disable the hunk tracker via --hunk-tracker-mode, GROK_HUNK_TRACKER, or config. Bug Fixes: • Windows install scripts now run cleanly via irm | iex without spurious BOM errors. • Tables & wide content no longer leave stray characters next to timestamps in the scrollback. • Mermaid diagrams now render node labels cleanly without HTML tags or raw markdown syntax. • MCP servers using HTTP now recover automatically after temporary connection drops! • Very long sessions can now scroll all the way to the bottom of the conversation history
Grok Build just got new update. The team is shipping at an absolutely relentless pace Release Notes: v0.2.54 Features: • Rewind now works end-to-end across conversation and file state with proper CAS handling. Bug Fixes: • Git branch icons now render correctly on Windows witho
