All entries

// Self-play with a pinch of human data // Really cool paper combining human demonstrations and self-play RL. 30 minutes of human data, 2500x less than imitation learning, is enough to make self-play

Since 2008, 15B taxpayer dollars have been invested in California High Speed Rail. There is no working track and zero passengers so far. In the same time frame 12B in private funds were invested in Sp

So true Many people think any given ML project is 99% training. In reality, it’s 50% evaluation, 40% data cleaning, 8% integration, and 2% training. The first two set the noise floor for learning. No

Some interesting findings suggesting that, by leveling performance, AI also commoditizes contract labor. New research: Is AI making employers view labor as more of a commodity? In a large online labor

Elon Musk expressed an opinion on X (formerly Twitter) that SpaceX possesses superior aesthetic design compared to other companies. The statement reflects Musk's perspective on the company's visual id

Speculation Is All You Need. In this blog post, we announce the co-release (w/ Z Lab) of six more state-of-the-art DFlash speculators for @Alibaba_Qwen 3.x. Over 1k output tps for 3.5 122B-A10B on a B

Elon Musk announced the launch or availability of Starlink satellite internet service in Lesotho, a landlocked country in southern Africa. This expansion represents Starlink's growing global coverage,

Terrifying sentence from Goldman Sachs: “Hyperscalers will need financing from across markets, structures, and currencies as they potentially bump up against saturation considerations in liquid credit

Atlantic reporter Alex Reisner recently uncovered four datasets of music being used to train AI models and made them fully searchable for the public. Two of the sets are absolutely enormous at 12 mill

The NSF helped build the internet, funded the observatory that detected gravitational waves, laid foundations for AI, advanced MRI technology, and trained generations of scientists. It is on track for

The UK Home Office is testing facial age estimation technology throughout 2026, with plans to implement it at the border in 2027 to help determine whether undocumented asylum seekers are adults or chi

There is no way the hyperscalers are going to make back their 5.3 trillion — unless they take it out of the taxpayers, via massive government subsidies. Which is *exactly* what they are going to try t

Emad Mostaque predicts that open-source AI models will soon reach a capability level ('fable-level') where they can run efficiently on consumer-grade MacBook Air/Mini computers without specialized har

This post likely discusses economic policy errors and systemic vulnerabilities that could transform a stock market correction into a severe, prolonged economic depression similar to the Great Depressi

this is so not true. anthropic’s own Claude Code uses harnesses, symbolic tools, regular expressions and 500,000 lines of symbolic code. it’s not scale alone and it *is* specialized. Anthropic's Jack

A Japanese restaurant's logo cleverly incorporates dual imagery, where the design of a traditional Japanese crane simultaneously forms the shape of a hand gripping chopsticks. This ambigram-style desi

Time to return to sanity The left just normalized so much insanity that basic self-preservation now gets called “far right extremism.” Decades of open borders and failed assimilation delivered paralle

Blank Slate mode is a feature in Hermes that can be enabled by running the 'hermes setup' command and selecting it from the available options. This mode is documented in the Hermes Quickstart Guide, w

Traded my Range Rover for a @Tesla Model Y and picked it up two weeks ago. Turned on FSD today after the 14.3.3 update. If you don’t believe in magic, I don’t know what to tell you. I’ve always been f

TSAuditor is a contamination auditing framework for time series foundation models that uses probe adaptation dynamics to detect when evaluation datasets may have been exposed during pretraining, valid

Vibecon was a two-day event held in New York where participants used Replit's coding platform to build projects in a playground-style environment, demonstrating code as a creative medium. The event br

A social media post expressing gratitude to David Hoang and Swyx for helping purchase a conference or event ticket, with the author confirming their attendance.

Very cool work from @jit_infinity: 🔥Leve: filesystem-first, durable agent framework built on LangGraph. You describe an agent as a directory of files. Leve compiles that directory into an agent and ru

We, along with @zhijianliu_ and @jianchen1799, are excited to share these models on @huggingface for the entire community to use. We look forward to getting speculative decoding speedups into the hand

when Vincent superimposed last year's floor plan ontop of this year's we were all quiet for a little bit just noticed AIE 2026 is at SF Moscone Center this year, huge upgrade from Marriott Marquis. we

This post documents a Wordle game completion (puzzle #1,826) solved in 3 guesses, with the final answer being a five-letter word with all letters correctly placed (indicated by the five green squares)

This appears to be a call for community input from DAIR.AI's Omar Saro regarding hands-on educational material development, likely soliciting topic requests and feedback from followers on what technic

This is a humorous tweet by Elon Musk making a joke about hypothetical taxation policies, using the scientific concept of antimatter as a comedic device. The tweet appears to be satirical commentary,

Yeah I agree. The big winner is going to be Ollama: I've offloaded all my supervisory, code review, and ontology learning agents to my $20 a month Ollama subscription, because I keep getting weight li

Fireworks AI reports that GLM 5.2, their language model, demonstrates performance comparable to GPT 5.5 across various evaluation benchmarks and real-world use cases. This claim suggests GLM 5.2 is co

Anthropic is going to IPO at $2T A bit of news: After nearly 9 years, I have decided to leave Google DeepMind and join Anthropic (after taking some time to recharge). I am incredibly grateful for my t

This post likely compares the pricing of major AI model APIs between the United States and China, highlighting cost differences that reflect different market conditions, regulatory environments, and c

C'est un grand honneur d’avoir reçu cette semaine l'insigne d'officier de l'Ordre national du Québec. Merci @CFrechette pour la remise de cette distinction. Il reste beaucoup de chemin à parcourir pou

Clients of SemiAnalysis Memory and Accelerator model knew about Rubin Ultra 16 Hi to 12 Hi cut in March :) LMAO Rubin Ultra's HBM literally got downgraded to 12 Hi, and they are not even using HB yet

This article demonstrates a practical application of Claude's code interpretation capabilities by using it to help decipher Linear A, an ancient undeciphered writing system from Minoan Crete dating ba

Deepagents code is sick cuz you can just use the best model as it comes out it is indeed quite good! don't try it in claude code/codex - those harnesses are overly tuned for their proprietary models d

dot-com bubble vs. a possible AI bubble. From the famous 'Dean of Valuation', Professor Aswath Damodaran, of NYU Stern School of Business, “And that’s the real big difference between the dot-com boom

Eight reasons to slow down hyperscaling: • What GenAI has already done to society • What AI slop is doing to the internet • What data center overbuild is likely to do the economy • What data center ov

Elastic NV has reportedly acquired Deductive AI Inc., a startup with a platform that helps enterprises fix errors in their technology infrastructure. TechCrunch on Thursday cited a source as saying th

GLM 5.2 quantized GGUFs: https://huggingface.co/unsloth/GLM-5.2-GGUF GLM 5.2 on Open Router: https://openrouter.ai/z-ai/glm-5.2 Minion coding agent: https://github.com/Sentdex/minion A decent consolid

Hermes Agent v0.17.0, released by Nous Research, is a major update referred to as 'The Reach Release' that introduces new capabilities and improvements to the Hermes Agent framework. The changelog det

LTX-2.3-22b-IC-LoRA-Cross-Eyed is a LoRA (Low-Rank Adaptation) model developed by Lightricks for the LTX-2.3-22b video generation model, created by Reddit user urabewe to address cross-eye artifacts i

I suspect that companies underestimate the value of using higher intelligence for tasks where weaker AIs seem to be good enough to hit KPIs at a lower price. At least build architectures where you can

'If Bores wins the primary, it has the potential to clarify and perhaps even transform the federal debate about regulating artificial intelligence. On the other hand, if he loses, Bores and others fea

If your AI is still being sycophantic, here’s how to solve it 👀 ✅ Option 1: “here is my research paper - be critical of it please” ✅ Option 2: “here is my research paper - spin up subagents to review

Elon Musk speculated on X about the astronomical costs of producing antimatter for interstellar propulsion systems, suggesting that future civilizations might spend an unimaginably large sum (a trilli