AI R&D | Fractional Advisor | Tech Lead | 20+ yrs Software Engineering | 9+ yrs Datascience (AI/ML). Also tweet about politics when democracy is on the line.
The Netherlands
Joined October 2008
- Tweets8.1K
- Following1.1K
- Followers1K
- Likes14.8K
Freddy Snijder retweeted
BREAKING: Revolut moved ~2,000 engineers onto self-hosted GLM for coding and pushed Claude Code and Codex to the exception list.
Per David Abitbol, who leads AI ops there: 56 Nvidia B200s, a blended 30 cents per million tokens, roughly a tenth of what they paid Anthropic and OpenAI. Engineering lead time improved 70%.
Here's the part that matters: the gains showed up when tokens stopped being rationed. Engineers quit saving AI for the hard tickets and pointed it at everything.
Frontier models become the exception, open weights become the default. That's a pricing problem for every lab heading to IPO.
h/t @OliverMolander
Would you move your team off frontier APIs if the bill dropped 90%?
Freddy Snijder retweeted
(wild) predictions after today's MIT Weight-Space Programming work:
1. Model weights become a programming language that AI can understand, but humans can't.
2. Training and inference will merge. Like human brain, AI could continuously rewrite its own weights while thinking.
3. A new path to RSI: AIs will directly edit each other's weights, evolving together without backprop.
Ok, I admit that this is something humans can’t do.
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math.
It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training.
We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵
Blog: rikiyat.github.io/blog/weigh…
Readers added context they thought people might want to know
LLM was given weights explicitly as text in prompt and used code/forward passes (per authors' logs) to ablate Qwen looping direction, a known thinking-mode issue per model card. Primarily fixed repetition in that mode, not general 'natively' improving weights/math.
rikiyat.github.io/blog/weight-sp…
github.com/RikiyaT/lm_und…
huggingface.co/Qwen/Qwen3.5-0…
huggingface.co/RikiyaT/Qwen3.…
Freddy Snijder retweeted
LLMs accumulate a lot of declarative knowledge.
That's how they can answer a large number of questions.
That's also why they require so many parameters and hence so much memory.
But one can replace the accumulation of knowledge by *understanding*
That what JEPA world models do.
That's one reason they can be useful with considerably less memory and training data than LLMs.
This paper merely shows that the underlying similarity structure of the data, whether text or image (e.g. co-occurence of words in text or objects in images) uniquely determines the shape of the representation point cloud, when the encoder is trained trained with a common anti-collapse criterion like SIGReg or VCReg (not by reconstruction).
Freddy Snijder retweeted
Yann LeCun was right & nobody in this industry wants to say it out loud.
109 million parameters beat 72 billion.
He has spent years saying scale is not the thing that makes a model understand the world. Meta bet the other way. He left in November and started his own lab in Paris.
Last Thursday he published the proof.
The theory everyone has been running on since 2024 is that if you train models big enough, on text, on images, on sound, they all end up with the same picture of reality inside them. Size does it. Nobody ever proved that part.
Randall Balestriero and LeCun proved it. And the proof says size was never doing the work.
What does the work is which two things you show a model together.
A photo and the caption of that photo are two views of one thing. Models trained on pairs like that get there fast.
Two paragraphs pulled from the same document are not two views of one thing. Models trained on those get there slowly and expensively.
Next word prediction is the second kind. That is what every chatbot in the world runs on.
In their tests the first method reached the same understanding after 21.6 million examples. Next word prediction needed 900 million.
And a 109 million parameter text model ended up closer to an image model's view of the world than a 72 billion parameter language model was.
660 times smaller.
Freddy Snijder retweeted
Absolutely agree, the current architecture can’t build up long-term memory, and a world model with it. It also mostly still reasons in token space, which is just silly.
When did Big Tech leaders and mentioned VCs do anything for democracy in the US and in the world in the last decade?
“Hinton said that the AI venture capitalists Marc Andreessen and Peter Thiel, and the CEO of the newly rebranded SpaceXSI, Elon Musk, “don’t believe in democracy”. They see themselves as “Übermenschen” who believe they should “control society” and that Trump is a useful instrument in that aim, he said.”
AI godfather Geoffrey Hinton: ‘We’re creating alien beings. I’m willing to be selfish and choose humans’
theguardian.com/us-news/2026…
via Instapaper
Freddy Snijder retweeted
Your LLM agent has to think out loud, token by token, before it can search, run code, or act in the world.
We humans don’t. We ponder in silence and speak when it matters.
Human language was never meant to be the limit of machine thought.
Introducing 🌗Hybrid Reasoning Agents
- Think in latent space, act through text
- Two modes, one unified multi-turn flow
- No black box, fully readable latent thoughts
- Any LLM can become a HybridAgent
Freddy Snijder retweeted
String theory is what happens when an intelligent system is starved of sensory input and starts hallucinating. Intelligence needs grounding.
For this reason, intelligence is practically never the bottleneck to solving impactful problems. The bottleneck is iterated contact with reality.
Freddy Snijder retweeted
Ukraine just hit another Yandex data center inside Russia, and Cloudflare data shows automated traffic coming from Yandex’s network has collapsed to roughly 0.4% of its recent peak.
For some perspective, researchers have previously estimated that bots generated roughly 21–29% of U.S. Twitter content.
With the U.S. midterms just weeks away, the timing is pretty interesting.
Freddy Snijder retweeted
The speed of our current landscape is nuts. Jev was groundbreaking for like 2 weeks.
Cloudflare’s Clef outperforms it, adds vision support and 2x context window.
Oh yeah, and it’s open source.
Cloudflare's Clef beat Jev in just two weeks
How: "frozen" Qwen does one prefill pass, a tiny schema head scores every answer option in parallel, with no text generated at all.
That's why it's 4x faster than Jev at 2x the accuracy on some benchmarks.
And it's open source!
Link to HF:
huggingface.co/Cloudflare/cl…
Freddy Snijder retweeted
openai announces glm-5.3 in codex subscriptions
anthropic announces glm-5.3 is evil and chinese
Freddy Snijder retweeted
Can an LM, starting from random init (!!), learn to generate all of its pretraining data?
Introducing Self-Play Pretraining with Zero Data. Two models start from random initialization: a generator proposes programs for a universal Turing machine and a learner trains on their outputs. We never train on any real data, but see predictable scaling on natural datasets: zero-shot val loss on images, text, audio, and melodies decreases predictably with self-play compute. And the learner develops in-context learning capabilities.
A fun proof-of-concept, co-led with @AdityaCowsik and @KfirDolev and co-authors @gbruno_dl, @ANourya @noahdgoodman, and @YoavLevine.
Freddy Snijder retweeted
Yes, Opus5.5 is a fantastic model, no dense, long explanations without context anymore, such a breath of fresh air! On top of that, so smart and so fast. Incredible to be honest.
