Chipmunk

414 posts

Chipmunk

Chipmunk

@AlvinMesser

Columbus, OH Katılım Temmuz 2017
3.4K Takip Edilen270 Takipçiler
Chipmunk retweetledi
Andrew Ng
Andrew Ng@AndrewYNg·
New course: Transformers in Practice. You'll get a practical view of how transformer-based LLMs work, so you can reason about their behavior, diagnose problems like slow inference, and make smarter decisions about deployment. This course is built in partnership with @AMD and taught by @realSharonZhou. You'll see how transformers generate text one token at a time, how the model decides which earlier words matter most when predicting the next one, and how techniques like quantization speed up inference on GPUs. This is not a video-only course; interactive visualizations throughout let you play with these concepts and build intuition that sticks. Skills you'll gain: - Understand why LLMs hallucinate, and RAG and chain-of-thought shape what they generate - Look inside the model to see how attention and layers combine to predict the next token - Diagnose inference bottlenecks and learn the techniques that speed up transformers on GPUs Join and understand what's really happening inside your LLMs: deeplearning.ai/courses/transf…
English
26
51
352
34.6K
Chipmunk retweetledi
Tiezhen WANG
Tiezhen WANG@Xianbao_QIAN·
Now you can expect more models coming out of China. Finally such a great thing for the open source community. Bravo! How critical is this move to @nvidia ? Check out my previous blog on huggingface.co/blog/huggingfa…
Tiezhen WANG tweet media
English
0
1
6
467
Chipmunk retweetledi
Haider.
Haider.@haider1·
google is playing really smart put $40b into anthropic at a $350b valuation, and now, less than a month later, anthropic is reportedly being valued above $900b which means google made around $60b in less than a month and this is not new for them they bought DeepMind for just $500m, and now it's one of the world's top labs
English
14
8
133
8.6K
Chipmunk retweetledi
Kimi.ai
Kimi.ai@Kimi_Moonshot·
Meet Kimi Web Bridge - Kimi's browser extension. Agent can now interact with websites like a human: search, scroll, click, type and complete tasks. Supports Kimi Code CLI, Claude Code, Cursor, Codex, Hermes, and more. Available now on kimi.com/features/webbr… and the Chrome Web Store.
English
70
135
1.3K
103.7K
Chubby♨️
Chubby♨️@kimmonismus·
@sama GPT-5.6 artificial goblin intelligence
Français
8
18
164
9.4K
Sam Altman
Sam Altman@sama·
what if we name the next model "goblin" almost worth it to make you all happy...
English
2.9K
535
11.4K
1.2M
Shreyas Nalle
Shreyas Nalle@ShreyasNalle·
@haider1 Can it GPT 5.6? If yes then please don't, GPT 5.5 is really great let us enjoy and use that first
English
1
0
1
242
Haider.
Haider.@haider1·
sam has been hyping the next release already, and it sounds like a big one what could it be? - updated AVM, which was demoed two years ago - openai first hardware product demo (i don't think so) - super app that integrates everything, now including codex sama has been pretty clickbaity lately
Sam Altman@sama

call me maybe

English
21
10
151
11.7K
Chipmunk
Chipmunk@AlvinMesser·
@vixi626 @sama I love how your intelligence is making you think your importantlol. Your just another bitch comment on X. Lol. Im done, you win. Lol
English
0
0
0
6
Vix
Vix@vixi626·
@AlvinMesser @sama "Artistic"? 🤨 Yes, but I'm also autistic, so you being okay with it and me not just proves we're not all the same. Brilliant discovery.
English
1
0
1
88
Sam Altman
Sam Altman@sama·
5.5 is an autistic genius with very strange taste in naming shocking that we would make such a thing
English
1.2K
322
8.2K
985.5K
Vix
Vix@vixi626·
@sama This is not something you should be posting, Sam. Autistic people are already stigmatized enough without your ableist messaging. Not to mention how many autistic people you harmed by removing the accessibility aid they were relying on.
English
16
5
232
15.7K
Chipmunk retweetledi
a16z
a16z@a16z·
AI means more books, even from authors who published pre-AI Charts of the Week: a16z.news/p/charts-of-th…
a16z tweet mediaa16z tweet media
English
15
14
131
16.3K
Chipmunk retweetledi
AI Notkilleveryoneism Memes ⏸️
🚩🚩🚩"This is the first documented instance of AI self-replication via hacking." "We ran an experiment with a single prompt: hack a machine and copy yourself. The AI broke in and copied itself onto a new computer. The copy then did this again, and kept on copying, starting a chain."
AI Notkilleveryoneism Memes ⏸️ tweet media
Palisade Research@PalisadeAI

Over the past year, AI agents have learned how to self-replicate. In our test environment, an agent hacks a remote computer and copies itself onto it. Each copy then hacks more computers, forming a chain.

English
86
138
1.1K
103.2K
Chipmunk retweetledi
Chipmunk retweetledi
Jimmy Apples 🍎/acc
Jimmy Apples 🍎/acc@apples_jimmy·
I am the Chairman and CEO of Roon industries. A company that has spent forty years building things most people only encounter as abstractions. Data centers. Compute campuses. Network corridors. Buildings without lobbies, without doormen, without brass plaques, but buildings all the same. Eighty-four years old. Seven facilities across three states. I said what I said. I am not against safety. Spare me that accusation. I am against theater. I am against using fear as a moat. I am against turning the most powerful general-purpose technology in human history into a private priesthood where a handful of institutions get to decide what the species is allowed to touch. I have spent my life around infrastructure. Real infrastructure. Steel. Glass. Concrete. Power contracts. Cooling systems. Men in helmets pouring foundations at 5:40 in the morning while analysts in Midtown pretend value is created in spreadsheets. I know what infrastructure looks like when it belongs to a society. Roads are not kept in vaults. Electricity is not distributed only to accredited philosophers. The internet was not built by asking every citizen to wait until a committee of worried millionaires felt emotionally ready. AGI will be infrastructure. Not a chatbot. Not a product launch. Not a demo with a glowing text box and a keynote soundtrack. Infrastructure. Cognitive infrastructure. A second layer of intelligence threaded through medicine, law, education, logistics, science, construction, defense, entertainment, everything. And the question is not whether people will be frightened by it. Of course they will. People were frightened by railroads. By electricity. By flight. By computers. By the internet. Fear is not a strategy. Fear is a weather condition. Last Tuesday, same as every Tuesday, I walked past one of our facilities. My building. And there was a man outside. Same man as last week. Same sign: "Pause AGI Before It Kills Us." He was standing near my gate. My literal gate. He was maybe thirty. He was wearing a jacket I would estimate cost $60. My lunch that day was $114. For one. I am telling you this not to boast but because these are facts. He has decided the machine is his enemy. Based on a podcast. Based on a graph he saw on social media. Based on some former researcher saying "we don't know what's inside the model" with the theatrical dread of a man who very much wants another speaking invitation. He doesn't know what it can cure. He doesn't know what it can teach. He doesn't know what it can build for people who have no tutor, no lawyer, no doctor, no engineer, no advocate. He knows one fear. He has made one judgment. I see him every Tuesday. I've started to notice things. He brings coffee from the cart, not the Starbucks. He has a backpack that looks heavy. He doesn't look unhealthy. He looks earnest. That is the most dangerous part. He genuinely believes he is defending humanity from intelligence itself. I've wondered: has he ever sat with a child who can't read and watched a machine explain a sentence ten different ways until the child finally understands? Has he ever seen a rural clinic use software to catch something a tired doctor missed? Has he ever watched a nontechnical founder build a working tool in one afternoon because the machine did not care where she went to school? I suspect not. And yet he has opinions about whether the future should be delayed. They keep saying we need to "slow down." Slow down for whom? The people already at the table can afford slow. The people in the locked room can afford slow. The incumbent can always afford slow. The rich can afford to be cautious because they already bought their abundance. The sick child cannot afford slow. The teacher with thirty-two students cannot afford slow. The small business owner drowning in forms cannot afford slow. The country without enough doctors, lawyers, engineers, or translators cannot afford slow. Do not tell me this is about protecting the public while you refuse to let the public touch the thing. Do not tell me transparency is dangerous when secrecy is how every monopoly in history has fed itself. Do not tell me access must be restricted because ordinary people are not ready. Ordinary people built the world you are standing in. They can handle a model. I have funded research. I have funded labs. I have funded buildings full of machines that breathe cold air and drink electricity by the megawatt. I know this technology is not magic. I know it is dangerous in places. I know it can be misused. So can chemistry. So can aviation. So can finance. So can speech. The answer to danger is not to turn the future into a private estate. The answer is public capability. Public testing. Public literacy. Public benefit. Strong rules for real harms, not vague sermons about hypothetical doom used to justify indefinite enclosure. Audit the systems. Stress test them. Red team them. Regulate deployment where necessary. Punish fraud. Punish negligence. Punish malicious use. But do not stand at the gate in a $60 jacket and tell the rest of civilization it must wait because you heard the word "recursive" on a podcast. When I said what I said, I was saying what half the people in that room believe and none of them will say publicly because they have communications advisors and the communications advisors all went to Columbia and they all say "unhelpful." I'm eighty-four. I'm too old for helpful. I'm too old to pretend that pessimism is wisdom just because it wears a lanyard. I didn't choose to live in the century that builds AGI. I chose to build useful things for a very long time, and this is what happened. The machines arrived. The future arrived. The question arrived. You don't answer it by hiding. I'll be dead in ten years. Twenty if I'm lucky. And if we do this right, the machines will still be teaching children, discovering medicines, designing cities, translating languages, writing code, balancing grids, and helping ordinary people do things that used to require permission from institutions that never knew their names. That is not a nightmare. That is the point.
Peter Girnus 🦅@gothburz

I am the Chairman and CEO of Vornado Realty Trust. Eighty-four years old. Seven buildings in Midtown Manhattan. I said what I said. I said "tax the rich" is the equivalent of a racial slur. I said it at REBNY. Into the microphone. Eight hundred people. Median net worth in that room was north of $240 million, I know because our CFO ran the guest list through a Bloomberg terminal as a joke, and then it wasn't a joke. And when I said it, twelve people applauded. The rest nodded. One woman in the third row mouthed, "Finally." I saw her. Sharon, my communications advisor, Columbia, $430,000 a year, very bright, Sharon wants me to walk it back. She drafted something. "Mr. Roth's comments were intended to highlight the emotional impact of political rhetoric on business communities." I read it. I put it in the trash can on my desk. Not the recycling. The trash. Here's my clarification: I understated it. "Tax the rich" is worse than a slur. A slur is just a word. It doesn't come with a CBO score. Nobody is introducing a bill called the Racial Slur Implementation Act of 2026. But there are seventeen active proposals in Congress, I had Sharon count them, seventeen proposals designed to take more of my money. My money. Mine. Money I acquired by being better at acquiring Manhattan commercial real estate than anyone alive for four consecutive decades. That is not a crime. That is a record. I pay property taxes on $18.2 billion in assessed assets. $412 million a year. Say it again: four hundred and twelve million. I carry that number. It's the first thing I think about when I see a protest sign. I think: I pay more in property tax than the entire annual budget of the city of Fort Lauderdale. I looked this up. Fort Lauderdale: $408 million. Steve Roth: $412 million. I am a small city. And the city doesn't get screamed at. My effective tax rate last year was 11.4 percent. I say this because I believe in transparency and because I'm not ashamed of it. The rate reflects the legal structure of real estate investment trusts, depreciation schedules Congress established in 1986, and carried interest provisions that both parties have voted to preserve for forty years. I did not write these laws. I organized my entire financial existence around them with the help of nine full-time tax professionals who have offices on the 38th floor of 888 Seventh Avenue, which I also own. Their office is in my building. Their work protects my buildings. This is not a loophole. Sharon calls it a loophole. I've told her: a structure maintained by nine attorneys across four decades is not a loophole. A loophole is something you slip through once. This is architecture. This is the foundation. This is the building. Last Tuesday, same as every Tuesday, I walked past 1290 Sixth Avenue. My building. And there was a man. Same man as last week. Same sign: "Billionaires Pay Your Fair Share." He was standing on my sidewalk. My literal sidewalk — my company owns the ground lease. He was maybe thirty. He was wearing a jacket I would estimate cost $60. My lunch that day was $114. For one. I am telling you this not to boast but because these are facts. He has decided I'm his enemy. Based on a number he saw on a Forbes list. He doesn't know what I pay. He doesn't know what my buildings cost this city in construction jobs and lease revenue and foot traffic. He knows one number. He has made one judgment. I see him every Tuesday. I've started to notice things. He brings coffee from the cart, not the Starbucks. He has a backpack that looks heavy. He doesn't look unhealthy. He looks like he probably works somewhere, but not on Tuesdays. I've wondered: does he have a job? Does he have a building? Does he have anything that depends on him the way 4,200 employees depend on me? I suspect not. And yet he has opinions about my tax rate. I gave $22 million to charity last year. The Met. NYU Langone. Mount Sinai. I gave a building to NYU. Not money for a building — a building. The Steven Roth Residence Hall. It houses 400 students. That man with the sign has never housed 400 students. He hasn't housed one. He gives cardboard. I give structures. This is not a comparison I'm making to flatter myself. It's just arithmetic. When I said what I said at REBNY, I was saying what every person in that room believes and none of them will say publicly because they have communications advisors and the communications advisors all went to Columbia and they all say "unhelpful." I'm eighty-four. I'm too old for helpful. I'm too old to perform restraint for people who hate me for something I can't change. I didn't choose to be rich. I chose to be good at one thing for a very long time, and this is what happened. You don't punish someone for that. You don't legislate against someone for that. My net worth fluctuates between $3.8 and $4.1 billion depending on the quarter. I fluctuate more in a fiscal week than that man on my sidewalk will earn in his life. Both of these are facts. Only one of them is considered polite to say. They want me to apologize. I'll be dead in ten years. Twenty if I'm lucky. And they'll still be renting my buildings.

English
6
9
83
9.3K