Is the AI boom finally starting to slow down? https://t.co/230p5q58Xs

AI mania or AI bubble? Nvidia’s earnings—and its China exposure—will decide how Wall Street answers that question https://t.co/mUxhv5LfC5 @fortunemagazine

A picture now is worth more than a thousand words in genAI; it can be turned into a full 3D world! And you can stroll in this garden endlessly long, it will still be there. https://t.co/barrcW05z0

Or stroll in this garden, another permanently consistent 3D world created from a single image no matter how long you hang out in it. https://t.co/t5ZVW8W7El

image2world https://t.co/3dTrg6v79R

image2world https://t.co/3dTrg6v79R

The WSJ Editorial Board slams Trump over the “vindictive” raid on John Bolton’s home: “This is the kind of gratuitous viciousness that has increasingly defined Mr. Trump’s return to office… The real offender here is a President who seems to think he can use the powers of his office to run vendettas. We said this was one of the risks of a second Trump term, and it’s turning out to be worse than we imagined.”

John Bolton: “Trump will have a very unique place in American history as our worst president ever.” (May) https://t.co/rxJ4ggzAyG

FUCK HEADS TRUMP & EPSTEIN ARE SEX TRAFFICKERS!!! I’m gonna post this every day so nobody forgets exactly who Pedophile Trump is and why the Epstein files have suddenly disappeared. https://t.co/gotWpgjoOJ

KEEP SCREAMING FOR JUSTICE!!! Don't let Trump's FIXER Todd Blanche @DAGToddBlanche whitewash the Epstein scandal by manufacturing fake exoneration via Maxwell's glowing misrepresentation of Donald. #TrumpEpsteinPedoCoverUp

San Francisco has hit its lowest murder rate in 60 years — 35 murders last year vs Houston’s 322. Anyone think Fox News will cover that? https://t.co/ejSEyFP1yY

Trump says he’s respected around the world. The reality is he is regarded as the global village idiot. I can’t think of a better way to describe his global reputation. He’s not even a politician, really. He’s just a bloke who’s filling his boots with bribes and strip mining America to the benefit of his uber rich buddies. Malcolm Turnbull hits the nail on the head. 🎥 TikTok - https://t.co/TRGDd7dadR

And after 1 week of work, here is zml/llmd running transparently on TPU with full prefill/decode paged attention. No code change, single flag, as it should be. https://t.co/UBS2nDFt6N

🪜Introducing: StepWiser🦉 📝: https://t.co/cocHKmWGMM - Reframes stepwise reward modeling as a reasoning task: outputs CoT + judgment. - Trained by RL using relative outcomes of rollouts. Results: (1) SOTA performance on ProcessBench! (2) Improves policy at train time. (3) Improves inference-time search. 🧵1/5

UNITED STATES: Chances of Zohran Mamdani winning the New York City mayoral election are now at 82%, according to prediction market @Kalshi. https://t.co/8fz2hJ63DH

The current H1B visa system is a scam that lets foreign workers fill American job opportunities. Hiring American workers should be the priority of all great American businesses. Now is the time to hire American. https://t.co/l27HEhF7C3

This is one of the most broadly applicable books I’ve read recently, highly recommend for managers The challenge for interns, analysts, and younger talent is the most important skills come from taking notes and observing how someone more senior works An important part of deal making is the delivery itself, which needs to be observed

I reference this AMA with Keith Rabois with LPs because at seed you’re primarily assessing the caliber of the founders Sometimes there are strong signals — repeat founder, acquisition by credible company, clear proof of work Sometimes there are weaker signals — bold idea, charismatic, unique ability to hire proven operators

Never a dull moment with this one… https://t.co/gzfK7Pq6jy

Will Smith is getting absolutely destroyed by fans for using an AI-generated crowd in his new music video. https://t.co/o7JfXB4nCd

The comments https://t.co/GOzi7xtEU7

@pitdesi it’s the robot ring bearer for me… https://t.co/TLogSAK9m4

It blows my mind that people are still outsourcing the most important decision you’ll ever make to apps with no meaningful innovation for over a decade. https://t.co/zNKukJEX1f

People wrote me saying that Jan is not free because you have to pay Serper API. Come on guys, install SearXNG locally, activate MCP et voilà: free, local and with total privacy. https://t.co/SULK4GHdcn

🔥 DeepCode has been trending on GitHub for 2 consecutive days! 🚀 Almost hitting 2k GitHub Stars! 🌟 Fully Open Source: https://t.co/OM1WbgRJiw ✨ All-in-One Agentic Coding Framework ✨ • 📄 Paper2Code - Research to Implementation • 🌐 Text2Web - Natural Language to Frontend • ⚙️ Text2Backend - Automated Backend Generation 🎯 Key Highlights: • Code Generation with Multi-Agent Systems • Powered by MCP (Model Context Protocol) • Multi-modal document processing (PDF, DOC, PPT, URLs, etc) • Fully open-sourced under MIT license 🤝 Join the DeepCode community and advance the future of AI-powered coding together! #DeepCode #vibecoding #agenticcoding

Pretty cool that they open sourced the actual full-sized production model. Here’s the Grok 2.5 architecture overview next to a roughly similarly sized Qwen3 model. The MoE residual is quite interesting. Kind of like a shared expert. I don't think I've seen this setup before. https://t.co/3t3c2buu3A

Tons of AI guides have dropped lately. Most repeat the same ideas, but a few are genuinely worth your time. Here are 8 must-read guides from Google, Anthropic, OpenAI, and Kaggle 🧵 ↓ https://t.co/I8vBWllLWp

After MCP, A2A, & AG-UI, there's another Agent protocol. It's fully open-source and launched by IBM Research. Here's a complete breakdown (with code): https://t.co/urRn4flrO3

💼 Finally a solid 57-page report on AI's effect on job-market from Stanford University. THE SHIFT HAS STARTED. Entry‑level workers in the most AI‑exposed jobs are seeing clear employment drops, while older peers and less‑exposed roles keep growing. Though overall employment continues to grow, employment growth for young workers in particular has been stagnant. The drop shows up mainly as fewer hires and headcount, not lower pay, and it is sharpest where AI usage looks like automation rather than collaboration. 22‑25 year olds in the most exposed jobs show a 13% relative employment decline after controls. ⚙️ The paper tracked millions of workers and boils recent AI labor effects into 6 concrete facts The headline being entry‑level contraction in AI‑exposed occupations and muted wage movement. AI replacing codified knowledge that juniors supply more of, than tacit knowledge that seniors accumulate. 🧵 Read on 👇

Lol i was working on a @DSPyOSS program builder UI as well @dbreunig . Started building it because the pipelines were getting complex. Will make it easier for customers to conceptualize what is happening. End goal is to get generate all the DSPy flavors (py, ts, rust, go, ruby, etc).

Since everyone is talking about RL Environments and GRPO now but no one knows how it works we thought it would be cool to make an explainer video + code you can run: This is an example of using GRPO to train Qwen 2.5 to play 2048 (code in thread) 🧵: https://t.co/Yag6izHmas

NVIDIA research just made LLMs 53x faster. 🤯 Imagine slashing your AI inference budget by 98%. This breakthrough doesn't require training a new model from scratch; it upgrades your existing ones for hyper-speed while matching or beating SOTA accuracy. Here's how it works: The technique is called Post Neural Architecture Search (PostNAS). It's a revolutionary process for retrofitting pre-trained models. Freeze the Knowledge: It starts with a powerful model (like Qwen2.5) and locks down its core MLP layers, preserving its intelligence. Surgical Replacement: It then uses a hardware-aware search to replace most of the slow, O(n²) full-attention layers with a new, hyper-efficient linear attention design called JetBlock. Optimize for Throughput: The search keeps a few key full-attention layers in the exact positions needed for complex reasoning, creating a hybrid model optimized for speed on H100 GPUs. The result is Jet-Nemotron: an AI delivering 2,885 tokens per second with top-tier model performance and a 47x smaller KV cache. Why this matters to your AI strategy: - Business Leaders: A 53x speedup translates to a ~98% cost reduction for inference at scale. This fundamentally changes the ROI calculation for deploying high-performance AI. - Practitioners: This isn't just for data centers. The massive efficiency gains and tiny memory footprint (154MB cache) make it possible to deploy SOTA-level models on memory-constrained and edge hardware. - Researchers: PostNAS offers a new, capital-efficient paradigm. Instead of spending millions on pre-training, you can now innovate on architecture by modifying existing models, dramatically lowering the barrier to entry for creating novel, efficient LMs.