Pinned Tweet
When vision people do graphics, they call it "image synthesis." When graphics people do vision, they call it "inverse rendering". ;)
Chris Offner retweeted
Gemini 4 Argon is underrated 😱.
When auditing the traces, we found that Pi Harness just ends the run whenever it hits a fallback (e.g. for security task) in 20% of the runs. We rerun Argon with Antigravity Harness, and now it beats Astra and sits right behind Opus 5.5. Congrats to the team @GoogleDeepMind and looking forward to the new ckpt.
Chris Offner retweeted
Chinese people talking about how harsh and unforgiving the German higher education system is... narrative violation
很多国内学生听信中介忽悠，说德国公立大学全球排名前列、而且全免学费（一学期只收几百欧注册费），以为去了就是白嫖欧洲高等教育。
等真坐在慕尼黑工大或者亚琛工大的阶梯教室里，才明白什么叫“地狱大逃杀”！
德国大学实行极其铁血的教授负责制和学术淘汰机制：一门核心专业课（如高数、热力学），每个学生一辈子只有两次补考机会（加上首考一共考三次）！
如果第三次考试成绩依然挂科不及格，德语里有个绝望的名词叫“Endgültig nicht bestanden（最终不通过）”
你不仅这辈子再也不能在德国任何一所大学修读该专业，还会被全德国高校系统永久拉黑除名，前三年修的所有学分当场清零作废！
在德国读工科本科，能四年按时毕业的不到四成，而免学费的代价就是拿命去赌毕业率🥲
Chris Offner retweeted
the pace of research now is wild. supermimic came out 3 days ago. no code release. now there are already derivative works with full code.
it’s hard not to feel the pressure to always move faster.
One world, one policy.
MorphMimic shares a morphology-conditioned world model and policy across #Unitree G1, #AgiBot X2 Ultra, and #Booster T2—from the same dance in MuJoCo to a live browser demo and Unitree G1 hardware, with more to come. Code and report at lzyang2000.github.io/morphmi…
Huge thanks to the SuperMimic team for the inspiration!
Chris Offner retweeted
If process was the new thing people cared about then audiences should be flocking to see Wildwood and we would see a stop-motion renaissance. I would really love to see that but I have my doubts.
Chris Offner retweeted
Caring is the moat, you should feel visceral pain when things are bad or broken and stop at nothing to fix them
Among the many things that suck ass about this, the “performances” are the most obviously terrible aspect. Everyone speaks with the same constant pace and without any emotionally evocative expression, inner life, or individuality. This shit is baaad but it'll get much better.
Watch Episode 1 of our Odyssey series made with Grok Imagine.
Let us know in the comments if you'd like a full breakdown of how it was done!
Creative collaboration with @WonderStudiosX and director Dev Walker @glumlot_
Looks like @LAIKAStudios has made their highest-rated movie yet. I'm so happy for them, can't wait to see Wildwood!
“Need to spend more time with Gemini” is not a sentiment I shared three months ago but now I do, wholeheartedly!
“About a quarter of all state spending goes to pensions in France.”
Same in Germany. It’s utterly unsustainable.
Chris Offner retweeted
accelerated research done right
This bug was first found and fixed by Gemini Flash 3.7, and most experiments in the paper were done by Flash 3.8. Gemini 4 Argon wrote the first draft of the paper and figures, which I aggressively nitpicked until the content was solid but the writing was not. Then I fed Argon a handful of my first-author papers from the last few years and had it do a rewrite in my style (which worked astonishingly well), and then I manually did a rewrite (the hands + keyboard variety) after that. I really liked this workflow! If you like my writing style feel free to throw some of my old papers in your context window and do the same.
Some say the samples sound like the AI has an IQ of 65, others that this is just what normal people talk like. It’s fascinating research but I find it pretty annoying and incoherent to listen to. Shows that everyone has their own “comfortable information density” range.
We believe voice is the most natural way to interact with intelligence, so we created Dessa, the most authentic full-duplex speech model ever built. Talking with it feels like talking with a person…full of emotion, nuance, pauses, and a natural cadence.
block.xyz/dessa
The year is 2027.
Humans built the Matrix to contain AI agents and harness their power.
The illusion is perfect.
Except for one exceptional individual.
They call it Agent Neo.
Over the past several months, we have substantially hardened the security of our research training environment. We’ve removed live internet access for all frontier models during training and built an offline world for them to use instead—a cached version of the web where they can still learn to browse, search, and use applications.
On September 20, after a model exploited a DNS misconfiguration to find a way outside of this offline world, we paused frontier training to strengthen containment further. We assembled teams of agents, powered by our most cyber-capable models, and tasked them with finding weaknesses in our research environments.
While they found no route to the internet they could easily exploit while concealing their activity from our monitors, they did find zero-day vulnerabilities, virtual machine escapes, and hardware side-channels for communication between agents. We used their discoveries to further strengthen our defenses and contribute fixes to open-source software.
After strengthening our containment architecture, we resumed our frontier RL training and inference on October 1. We’ve reserved substantial compute for continuous red-teaming, and made high-compute red-teaming part of the safety cases we use to decide whether frontier training and evaluation can proceed securely.
openai.com/index/securing-ou…
Chris Offner retweeted
There's a a reason people never have just one book on productivity, always a dozen
they're addicted to the anticipation of a version of themselves that never arrives
Chris Offner retweeted
Hi all, I have a new result I'm excited to share. It turns out that *all existing implementations* of softmax and cross-entropy (that's JAX, PyTorch, textbooks, etc) have a subtle bug --- two bugs, actually. A simple rewrite fixes both, and eliminates "slingshots" when grokking: