37 years ago today, the Chinese government brutally crushed peaceful protesters in and around Tiananmen Square who were demanding an end to corruption, freedom of speech, and democratic reform. The massacre revealed a truth the world should never forget: the Chinese Communist Party will do whatever it takes to preserve its grip on power. If it did not value the lives of its own citizens, why would it value the lives of others?
Such a scene is often framed as proof the Dutch are superhuman—they’re not.
The real miracle isn't the Dutch cyclist. It's the environment that allows ordinary people to do extraordinary things. Build the infrastructure, and you'll find your city is full of capable cyclists too.
@MarkusQ@ICEgov@MarkusQ I had a Dutch friend try to convert me in middle school, meaning I've been unconvinced for 40 years. I will grant, ketchup is overrated and a good fry requires little other than heat and salt.
Americans are going to find out about me during the World Cup and I’m just going to say it now: I put mayonnaise on my fries.
Not ketchup. Mayo. The Dutch are right and I will not be debating this.
🚨 NVIDIA just pulled off something crazy: making bounding box detection 10x faster by ripping out the exact step the entire industry assumed was mandatory ↓
Every VLM grounding model treats boxes like sentences, predicting them token by token. It’s inherently slow.
Enter LocateAnything (trending #1 on HF, CVPR 2026).
It’s an advanced 3B vision-language model that finds any object, UI target, or text using natural language by asking a simple question:
Why serialize a box at all?
The four corners are coupled.
It predicts the whole box atomically, in one parallel step.
The impact of parallel decoding:
→ 12.7 boxes/sec on a single H100 (10x faster than Qwen3-VL, 2.5× vs Rex-Omni)
→ Accuracy goes up, not down (+3.8% F1 on LVIS, big wins at IoU 0.95)
→ Dense scenes (300 boxes) hit ~25 BPS while sequential falls off a cliff
→ Built-in fallback: reverts to sequential decoding if the output looks wrong
→ Trained on 785M boxes / 138M queries across referring, GUI, and OCR tasks
The breakthrough isn't just speed.
It’s realizing that forcing structured outputs through text-shaped pipes creates artificial bottlenecks.
Boxes were never tokens.
Repo, demo, weights, paper, and other resources in the 🧵 ↓
🏃♂️ I've gamified my own run so I can race my own ghost with the Meta Ray-Ban Display.
I built a web app for the glasses, loaded a previous GPX from Strava, and dropped game mechanics on top.
Pick up coins when you keep pace, sprint zones reward extra points if you push, and a mini leaderboard on the lens shows how you're tracking against your past self in real time.
Best part: it actually works. Seeing your ghost 20 m ahead is a way stronger nudge than any number on a watch. 😅
Madras is breaking ground on a new Latino residential treatment facility, expanding access to culturally specific care for Oregonians working toward recovery. This investment will help more people access support, healing, and treatment closer to home. https://t.co/87ZNNHCRLj
The homelessness crisis has never been about housing.
If housing first works, it would’ve worked.
Make treatment mandatory, reunite families, & stop allowing criminals & our political leaders to profit off of the homeless.
We can no longer wait for the homeless to find the door to treatment.
We need to kick it open and walk them through it.