What if a VLM could teach itself from zero data? Meet MM-Zero: one base model that self-evolves by playing 3 roles: 🧠 Proposer – imagines visual concepts 💻 Coder – renders them in Python/SVG 🔎 Solver – reasons over the results No seed images. Just RL (GRPO) + execution feedback. @zli12321 @ChengsongH31219

What if a VLM could teach itself from zero data? Meet MM-Zero: one base model that self-evolves by playing 3 roles: 🧠 Proposer – imagines visual concepts 💻 Coder – renders them in Python/SVG 🔎 Solver – reasons over the results No seed images. Just RL (GRPO) + execution feedback. @zli12321 @ChengsongH31219

https://t.co/0muCOoj0BN

Very exciting initiative! AI is only as good as its data. At @SophontAI, we've become acutely familiar with this. Across the three domains we're working on (language, pathology, neuroimaging) we're very quickly observing how important data quality, diversity, coverage, etc. all affect model performance, capabilities, and scaling. There is very clearly a need to carefully study how all of these aspects affect frontier model research. Glad to see that Protege is building a frontier AI data lab to rigorously tackle these research questions!

@Miles_Brundage i'm honored to be following you! i remember you from the days you used to post papers and then someone made a bot to replicate that...good old days lol

Agents? Planetary scale inference? Claws? Surfboards? A stolen Vespa? Check out this Latent Space Podcast episode featuring @NaderLikeLadder and I. We talk agents, NVIDIA, Dynamo, and a whole bunch more! https://t.co/GAu1mlPagW

Replit is the best example of what @paulg calls a “cockroach” - no matter what happens in the outside world, they survive and thrive. Replit is no longer for underage students learning to kode. It’s somehow a more capable and cohesive AI productivity suite than either Google or Msft have put together for the knowledge working. You can do slides! You can do videos! you can use @excalidraw and make real an animation!!! massive respect for this team. they know where the vibes are shifting and always play toward there while somehow making it all look obvious in retrospect.

@SpatialCapStark @amasad @swyx @paulg @Replit Yeah, the ones who can pivot fastest without destroying themselves often win. Agreed that it's impressive.

thanks for sharing our work!

thanks for sharing our work!

@pagan_hoetry just a taste of the regime to come

i’ve decided to be less stupid going forward

free bleachers show at union pool brooklyn in 1 hour first come first serve 🍅🙂‍↔️

a new STI will be born

🇨🇳 Registration is open for #KubeCon + #CloudNativeCon + #OpenInfraSummit + #PyTorchCon China, 8-9 September in Shanghai! 🎟 Register by 7 April + save up to ¥1270: https://t.co/EohHFBgyki. 🎤 Got insights to share? Submit a talk by 3 May: https://t.co/ScyV1Tn8JJ https://t.co/VQT2Ms6zR3

+1 more

AI is rapidly improving at reading medical images. But the real question is not whether it will replace radiologists. Most experts see AI becoming a powerful co-pilot: spotting patterns faster, flagging anomalies earlier and helping radiologists focus on complex diagnoses and patient context. The technology augments expertise rather than eliminating it. The bigger shift may be this: radiologists who use AI could eventually replace those who don’t. https://t.co/eGYsTyC6Tw @ConversationUK @ConversationUS

last week we got 1M views and 100s of death threats for giving Openclaw access to drones, humanoids, quadrupeds, and other physical hardware. Now we’re releasing EVERYTHING open-source. Dimensional gives agents access to the physical world. Join us. Repo 👇🏽👇🏽👇🏽 https://t.co/uyYaWMRehT

The vibes at Perplexity’ Ask conference are immaculate https://t.co/AjSMcPmHr6

iMac: the bicycle for the mind

Interesting stat - our enterprise customers have already done more Devin sessions (and more merged Devin PRs) in 2026 than in all of 2025. Not bad for 2-ish months into the year!

Oracle is confirmed cutting 20,000-30,000 jobs but sources inside are saying the real number is closer to 45,000 I'm hearing this isn't just about AI data center costs Word is they've been running pilot programs with AI agents doing database administration work for 8 months One source told me a team of 47 DBAs in Austin got replaced by 3 senior architects plus automated Oracle Cloud Infrastructure management The agents are handling routine maintenance, performance tuning, backup verification - stuff that used to require armies of L4 and L5 engineers Internal metrics show the AI systems are catching 94% of database issues before human intervention needed But here's the terrifying part: they're not just cutting the obvious roles I'm hearing entire solution engineering teams are getting eliminated - the people who customize implementations for enterprise clients Apparently the new AI workflow can generate custom database schemas and migration plans in 6 hours instead of 6 weeks One insider said they watched a 12-person team that handled Fortune 500 implementations get told their roles were "redundant effective immediately" The severance packages are allegedly massive - 18 months salary plus equity vesting acceleration But that's because Oracle knows these people can't find equivalent work anywhere Every other enterprise software company is running the same playbook One source said it best: "We're not getting laid off, we're getting archived"

I moved from TUIs/IDEs to my own agent orchestrator in 3 months. Coding agents can do more for you, but the wrong UI is going to hold you back. How it looks for me: taskboard, notes, skills, automations, control center,...all tunable by agents. There is a lot more. https://t.co/Mo0vSYnVLd

Good time to follow: https://t.co/jP1onDjnuD

@stellayu Yeah it is missed.