AI & IA

1 curation · 458 membres · fenêtre de 168 h, demi-vie 6 h, seuil 3 relayeurs

Fenêtre 6 h 12 h 24 h 2 j 3 j 7 j
Montrer Les deux Liens 7 Tweets 12
Seuil 2 relayeurs 3 relayeurs 4 relayeurs 5 relayeurs 6 relayeurs Tout remettre
7liens
12tweets
46091posts lus
358auteurs
4899liens en base
1455 hde recul

Les liens 7

Classés par auteurs distincts, jamais par audience.

1
3 relayeurs à 3 · il y a 28 h thume.ca
Qui l'a relayé
MONTREAL.AI @Montreal_AI · 07/11 19:46 ⚜️✨ https://t.co/LziAg5YHcG | Montréal Artificial Intelligence : Sovereign AI, AI‑First Enterprise, Autonomous Agents & Governance ( Français: @Montreal_IA ) #MontrealAI
Earlier this week Tristan published his top YouTube channels: Which looked excellent so clearly I had to turn it into TristanTube: 47K super nerdy videos +TV mode for laid back watching. (One-shotted with Claude Code in 3 minutes)
Andrej Karpathy @karpathy · 07/11 18:07 deep learning
Earlier this week Tristan published his top YouTube channels: Which looked excellent so clearly I had to turn it into TristanTube: 47K super nerdy videos +TV mode for laid back watching. (One-shotted with Claude Code in 3 minutes)
Dominique Fleury @D4ive · 07/11 23:05
Earlier this week Tristan published his top YouTube channels: Which looked excellent so clearly I had to turn it into TristanTube: 47K super nerdy videos +TV mode for laid back watching. (One-shotted with Claude Code in 3 minutes)
2
3 relayeurs à 3 · il y a 28 h karpathy.ai reprise probable
Qui l'a relayé
MONTREAL.AI @Montreal_AI · 07/11 19:46 ⚜️✨ https://t.co/LziAg5YHcG | Montréal Artificial Intelligence : Sovereign AI, AI‑First Enterprise, Autonomous Agents & Governance ( Français: @Montreal_IA ) #MontrealAI
Earlier this week Tristan published his top YouTube channels: Which looked excellent so clearly I had to turn it into TristanTube: 47K super nerdy videos +TV mode for laid back watching. (One-shotted with Claude Code in 3 minutes)
Andrej Karpathy @karpathy · 07/11 18:07 deep learning
Earlier this week Tristan published his top YouTube channels: Which looked excellent so clearly I had to turn it into TristanTube: 47K super nerdy videos +TV mode for laid back watching. (One-shotted with Claude Code in 3 minutes)
Dominique Fleury @D4ive · 07/11 23:05
Earlier this week Tristan published his top YouTube channels: Which looked excellent so clearly I had to turn it into TristanTube: 47K super nerdy videos +TV mode for laid back watching. (One-shotted with Claude Code in 3 minutes)
3

https://z.ai/

Meet Z.ai, the AI assistant powered by GLM-5.3-Flash. Build websites, write code, handle long-horizon tasks, and get instant answers. Fast, smart, and reliable.

3 relayeurs à 7 · il y a 3 j z.ai
Qui l'a relayé
Machine Learning FLX @machinelearnflx · 05/11 22:18 AI engineering, distilled. Agents, LLM systems, MLOps & infra architectures, papers, and tools worth your time.
chief scientist Jie Tang just revealed the training pipeline behind the upcoming GLM-5.5 (32B, 122B, and 256B) during a lecture at Tsinghua University today. Instead of relying on public datasets, their pretraining pipeline pulls from six separate
Peter Xing @peterxing · 06/11 12:46 si/acc @microsoft w/ @azure & @github +@openai @anthropicai @spacexsi @meta. c/f @transhumanau @transhumancoin. h+ @singularityu. fsd @tesla. ex @kpmg @deloitte
chief scientist Jie Tang just revealed the training pipeline behind the upcoming GLM-5.5 (32B, 122B, and 256B) during a lecture at Tsinghua University today. Instead of relying on public datasets, their pretraining pipeline pulls from six separate
ℏεsam @Hesamation · 06/11 00:50 building ai agents or stuff with agents CrusaderBench: https://t.co/Ib0DkhWkEm
chief scientist Jie Tang just revealed the training pipeline behind the upcoming GLM-5.5 (32B, 122B, and 256B) during a lecture at Tsinghua University today. Instead of relying on public datasets, their pretraining pipeline pulls from six separate
4

moonshotai/Kimi-K3.1-Pro · Hugging Face

Image-Text-to-Text model with 2.8T parameters. Download, explore files and run it on Hugging Face.

3 relayeurs à 3 · il y a 3 j huggingface.co
Qui l'a relayé
clem 🤗 @ClementDelangue · 04/11 18:02 Co-founder & CEO @HuggingFace 🤗, the open and collaborative platform for AI builders
Releasing the model weights of Kimi K3.1 Pro: We'll keep advancing open intelligence.
AK @_akhaliq · 04/11 18:04 AI research paper tweets, ML @Gradio (acq. by @HuggingFace 🤗) dm for promo ,submit papers here: https://t.co/UzmYN5XOCi
Kimi-K3.1-Pro just dropped on Hugging Face!! 🌝🚨
Machine Learning FLX @machinelearnflx · 05/11 22:19 AI engineering, distilled. Agents, LLM systems, MLOps & infra architectures, papers, and tools worth your time.
Kimi-K3.1-Pro is now open-weights
5

Qwen 3.8 27B on my laptop can interpret model weights, too

Yesterday’s hot paper was Sparks of Weight-Space Programming: LLMs can natively read & write neural network weights by Rikiya Takehi, Jayden Teoh, Alex Zhang, and Omar Khattab. The paper found …

3 relayeurs à 3 · il y a 4 j Simon Willison’s Weblog
Qui l'a relayé
MONTREAL.AI @Montreal_AI · 03/11 19:50 ⚜️✨ https://t.co/LziAg5YHcG | Montréal Artificial Intelligence : Sovereign AI, AI‑First Enterprise, Autonomous Agents & Governance ( Français: @Montreal_IA ) #MontrealAI
There's a buzzy new paper about GPT-6 Astra being able to directly interpret model weights (prompts full of floating point numbers) - I tried it with Qwen 3.8 27B on my laptop and found that, after 45+ minutes, that model could achieve the same thing
AI @DeepLearn007 · 04/11 15:48 Imtiaz Adam CS #AI Postgrad |#Strategy #MachineLearning RSI Evolutionary Neurosymbolic |#RL #Agentic | #LLM Liberal | MBA alum @morganstanley @LBS @Columbia_Biz
There's a buzzy new paper about GPT-6 Astra being able to directly interpret model weights (prompts full of floating point numbers) - I tried it with Qwen 3.8 27B on my laptop and found that, after 45+ minutes, that model could achieve the same thing
Simon Willison @simonw · 03/11 17:31 Creator @datasetteproj, co-creator Django. PSF board. Hangs out with @natbat. He/Him. Mastodon: https://t.co/t0MrmnJW0K Bsky: https://t.co/OnWIyhX4CH
There's a buzzy new paper about GPT-6 Astra being able to directly interpret model weights (prompts full of floating point numbers) - I tried it with Qwen 3.8 27B on my laptop and found that, after 45+ minutes, that model could achieve the same thing
6

Weight-Space-Programmed Qwen3.5-0.8B - a Hugging Face Space by ML-Intern-lab

This app lets you type a math problem and (optionally) set a token limit. It then runs both the original Qwen3.5‑0.8B model and its weight‑edited version, showing the two generated answers side‑by‑...

3 relayeurs à 3 · il y a 5 j huggingface.co reprise probable
Qui l'a relayé
AK @_akhaliq · 03/11 05:40 AI research paper tweets, ML @Gradio (acq. by @HuggingFace 🤗) dm for promo ,submit papers here: https://t.co/UzmYN5XOCi
@a1zhang @lateinteraction @jayden_teoh_ The edited model is now on the Hub: Huge thanks to @akhaliq, who rebuilt it from our published logs, re-ran MATH-500 (51.0%). They also seems to have a live demo:
Hugging Face @huggingface · 04/11 09:21 The AI community building the future. https://t.co/TpiXQMQ9rZ
We @huggingface reproduced this paper!! It's built with ML Intern, an open-source harness to do AI research, train models, and more Model: Demo:
ℏεsam @Hesamation · 03/11 16:23 building ai agents or stuff with agents CrusaderBench: https://t.co/Ib0DkhWkEm
We @huggingface reproduced this paper!! It's built with ML Intern, an open-source harness to do AI research, train models, and more Model: Demo:
7

ML-Intern-lab/Qwen3.5-0.8B-weight-space-programming · Hugging Face

Image-Text-to-Text model with 0.9B parameters. Download, explore files and run it on Hugging Face.

3 relayeurs à 3 · il y a 5 j huggingface.co
Qui l'a relayé
Hugging Face @huggingface · 04/11 09:21 The AI community building the future. https://t.co/TpiXQMQ9rZ
We @huggingface reproduced this paper!! It's built with ML Intern, an open-source harness to do AI research, train models, and more Model: Demo:
AK @_akhaliq · 03/11 18:10 AI research paper tweets, ML @Gradio (acq. by @HuggingFace 🤗) dm for promo ,submit papers here: https://t.co/UzmYN5XOCi
We @huggingface reproduced this paper!! It's built with ML Intern, an open-source harness to do AI research, train models, and more Model: Demo:
ℏεsam @Hesamation · 03/11 16:23 building ai agents or stuff with agents CrusaderBench: https://t.co/Ib0DkhWkEm
We @huggingface reproduced this paper!! It's built with ML Intern, an open-source harness to do AI research, train models, and more Model: Demo:

Les tweets 12

Axe indépendant : repris sans regarder les liens.

1
Pushmeet Kohli @pushmeet · 08/11 19:36
Across our projects @GoogleDeepMind, we are witnessing a genuine phase transition in the pace of scientific discovery. You will see a glimpse of this in our releases in the next few weeks.
5 relayeurs à 5 · il y a 8 h Voir sur X
Qui l'a repris
E echo.hive Dr Singularity” elvis” B Benjamin De Kraker” Demis Hassabis
Ce qu'ils en ont dit
Dr Singularity @Dr_Singularity a cité · 08/11 18:30
1000 scientific breakthroughs drop from Google incoming?
elvis @omarsar0 a cité · 08/11 17:08
Been working for this moment for over a decade. What I've witnessed behind the scenes, together with what we have been building, tells me we are about to enter a truly incredible era of AI-powered scientific discovery. It's an incredible time to be doing research. I will be
B
Benjamin De Kraker @BenjaminDEKR a cité · 08/11 15:15
What did he mean by this 👀
3
J
Justin Lo @loresjpg · 08/11 22:00
Nice DeepSWE score! Does it survive a lowercase filename? Introducing MaxxBench: 89 DeepSWE tasks with small changes to the instructions or starting code. Same engineering problem, slightly different requirement. Models clustered around 74% on DeepSWE end up 23 pp apart. Good
3 relayeurs à 3 · il y a 8 h Voir sur X
Qui l'a repris
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)” Ethan Mollick” ℏεsam”
Ce qu'ils en ont dit
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞) @teortaxesTex a cité · 08/11 22:00
holy shit, an art form of scientific grift next METR/Epoch/AA employee right here you all need to work on your epistemic hygiene
Ethan Mollick @emollick a cité · 08/11 18:33
ℏεsam @Hesamation a cité · 07/11 18:58
dude. this is painful to look at.
4
Richard Sutton @RichardSSutton · 08/11 16:18
Brute force computation is not enough for intelligence. Clever new methods are needed. The science of AI is far from finished. IMHO
7 relayeurs à 7 · il y a 21 h Voir sur X
Qui l'a repris
Tommy” elvis” ℏεsam” Gary Marcus” cocktail peanut” Srini Pagidyala”
Ce qu'ils en ont dit
Tommy @Shaughnessy119 a cité · 08/11 16:18
I’ve been actively on the hunt for non LLM based AI models and I’ve invested into one already Looking for totally new designs
elvis @omarsar0 a cité · 08/11 01:26
1000% agree with Sutton here. Having worked in AI for over a decade (pre-LLMs), it's clear we need smarter methods for intelligence to truly compound in the way some of us envision it. To me, Transformers just scratched the surface. This arc is coming to an end; a new
ℏεsam @Hesamation a cité · 07/11 23:55
> “what happened to the bitter lesson?” people are genuinely trying to explain the bitter lesson to the guy who wrote it. it’s called the bitter lesson not the dumb lesson. the bitter lesson says general methods that scale with compute win in the long run. you still have to
Gary Marcus @GaryMarcus a cité · 07/11 23:32
what i have been saying for years maybe now people will listen?
cocktail peanut @cocktailpeanut a cité · 07/11 21:29
Srini Pagidyala @Srini_Pa a cité · 07/11 21:16
Human like learning does it. Implies understanding, concept formation, metacognition, self-directed, goal-oriented, persistent memory, contextual awareness, belief revision, dynamic world models are all part of it. That is Integrated Neuro Symbolic Architecture- INSA.
5
Nous Research @NousResearch · 08/11 16:20
Introducing Arete-27B: Qwen3.8-27B, post-trained to reinforce Western values. Chinese open models are currently the strongest available, but they ship with state-aligned behavior in the weights. We built an 803-prompt eval across seven value categories, including freedom of
4 relayeurs à 4 · il y a 17 h Voir sur X
Qui l'a repris
Tommy” Robert Scoble clem 🤗 Marco Franzon”
Ce qu'ils en ont dit
Tommy @Shaughnessy119 a cité · 08/11 16:20
The era of super interesting AI brain surgery is upon us We saw this with Evil Qwen where someone changed 2 parameters and altered the entire personality We are seeing it with Nous taking the power of open source models and removing the Chinese marching orders At the end of
Marco Franzon @mfranz_on a cité · 06/11 23:17
Nous Research is a good company. They are making the best AI coding agent and now they drop an amazing local model. I really appreciate what you are doing
6
Anthropic @AnthropicAI · 08/11 06:17
Memory is what allows Claude to learn from experience. In a new Engineering Blog post, we share how “dreaming” helps Claude organize what it’s learned, correct its mistakes, and find what it needs more easily the next time.
anthropic.com The components of agent memory How we think about agent memory, and how “dreaming” helps Claude correct mistakes, organize what it’s learned, and find it in future sessions.
7 relayeurs à 7 · il y a 20 h Voir sur X
Qui l'a repris
A Alvin-GenAI” AI Rahul” Somi” Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)” elvis”
Ce qu'ils en ont dit
A
Alvin-GenAI @AlvinWeb3 a cité · 08/11 06:17
AI can also DREAM now? Anthropic inspired from how our brains organize memories during sleep to build a memory system (similar to wikis) for Claude. agents can now “dream” about past experiences: revisit its transcripts, correct mistakes, and evolve its memory wiki when not in
Rahul @sairahul1 a cité · 07/11 03:31
POV: you're Claude playing Pokémon. You've fallen off the same ledge three times. You write "I'm going to stop rather than keep looping." The test setup auto-replies: continue. So you keep walking, all the way back to the last town, and never make it. Most relatable thing an
Somi @somi_ai a cité · 07/11 02:01
Claude wrote "Mt Moon DONE" in its own Pokémon notes. It wasn't done. One wrong note, and it fell off the same ledge three times. Anthropic's "dreaming" pass checked the note against the old transcript and fixed it. That run reached Cerulean City. I'll take memory that checks
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞) @teortaxesTex a cité · 07/11 00:54
Anthropic proposes an answer to Wenfeng's question about continual learning: …have a smarter model write more .md files. It's impressive how far this goes, but I still believe that it's a crutch with a ceiling set by the pretrain's ability, just like RLMs are a context crutch.
elvis @omarsar0 a cité · 06/11 23:49
Highly recommended read. Learn how one of the most advanced agent memory systems works. Your memory system is one of the biggest levers you have in the harness. I also don't think a one-size-fits-all memory system exists today. Memory is something you constantly have to
7
Andrej Karpathy @karpathy · 07/11 23:05
Earlier this week Tristan published his top YouTube channels: Which looked excellent so clearly I had to turn it into TristanTube: 47K super nerdy videos +TV mode for laid back watching. (One-shotted with Claude Code in 3 minutes)
karpathy.ai TristanTube
thume.ca My YouTube tier list, six years later
3 relayeurs à 3 · il y a 28 h Voir sur X
Qui l'a repris
Dominique Fleury Rahul” MONTREAL.AI
Ce qu'ils en ont dit
Rahul @sairahul1 a cité · 07/11 20:07
karpathy is so done with the youtube algorithm he's building his own youtubes now 😭 last week: 6,454 videos pulled from 43M Hacker News comments. ~2 hours. today: turned a YouTube tier list into a 47K-video site. 3 minutes with Claude Code.
8
P
Physical Superintelligence @psipbc · 07/11 20:21
Today Physical Superintelligence PBC is announcing a proof of the Aaronson-Ambainis conjecture, one of the central open problems in quantum complexity, with major implications for quantum computing speedup. Thread 👇
PSI A proof of the Aaronson-Ambainis conjecture A proof of the Aaronson-Ambainis conjecture about quantum speedups on random inputs, with a complete formalization in Lean.
4 relayeurs à 4 · il y a 31 h Voir sur X
Qui l'a repris
Machine Learning FLX Peter Xing Dr Singularity” ℏεsam”
Ce qu'ils en ont dit
Dr Singularity @Dr_Singularity a cité · 07/11 04:54
New wave of AI physics breakthroughs
ℏεsam @Hesamation a cité · 06/11 17:11
Anthropic’s UNRELEASED MODEL helped researchers solve an 18 year old conjecture in quantum computing. it’s interesting how far ahead these internal models are compared to what the public has available.
9
Claude @claudeai · 07/11 23:33
Introducing Personal Finance in Claude, in beta in the US. Connect the accounts you choose, and Claude answers your money questions in everyday language, from your own balances and transactions. Access is read-only, and Claude can't move, manage, or invest your money.
5 relayeurs à 5 · il y a 28 h Voir sur X
Qui l'a repris
Somi” Min Choi” Kanika” Tom's KI Ecke Rahul”
Ce qu'ils en ont dit
Somi @somi_ai a cité · 07/11 23:33
Read-only is the best call in this launch. Claude sees every balance and every transaction, and it can't move a single dollar. That's the version I'd actually connect my bank to. One that pays the bills on its own? Not for a long time.
Min Choi @minchoi a cité · 06/11 17:29
Anthropic just turned Claude into your personal finance assistant. Connect your bank + cards. Find hidden charges. See what's left before payday. Know where your money goes. Read-only. It can't move or invest a dime. Beta, US only
Kanika @KanikaBK a cité · 06/11 16:49
Everyone's asking what's going on at Anthropic. And honestly.. a lot happened in about 24 hours. Let me walk you through it. First, Claude launched Personal Finance. You connect your bank accounts, and Claude answers money questions from your real balances and transactions.
Rahul @sairahul1 a cité · 05/11 19:12
claude can see your bank account now. here's claude's.
10
Elon Musk @elonmusk · 07/11 21:46
Tesla will be winding down production of the Cybertruck at Gigafactory Texas around the middle of next year to make room for increased Cybercab production
3 relayeurs à 3 · il y a 29 h Voir sur X
Qui l'a repris
Sid Bharath” Kanika” Robert Scoble”
Ce qu'ils en ont dit
Sid Bharath @Siddharth87 a cité · 07/11 21:46
Now that it’s dead, can we all agree it was ugly af
Kanika @KanikaBK a cité · 06/11 11:34
do you understand what elon just did.. he's winding down cybertruck production in texas.. and handing that factory space to a car with no steering wheel.. let me explain.. in 2019 he said the cybertruck would start at $39,900.. and tesla would make 250,000 of them every
Robert Scoble @Scobleizer a cité · 06/11 07:33
Holy shit. My friend and I argued about this. I was wrong.
11
A
AItonomy Foundation @aitonomyfound · 07/11 21:31
I’m incredibly proud to share what we’ve been building at the AItonomy Foundation: Science’s Last Exam (SLE). 🧪✨ Can AI do science? Instead of Q&A, we give agents actual research goals and real-world tools (lab data, software), then grade them against hidden expert checks.
3 relayeurs à 3 · il y a 29 h Voir sur X
Qui l'a repris
Peter Xing Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)” elvis”
Ce qu'ils en ont dit
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞) @teortaxesTex a cité · 06/11 08:36
Another new eval, another expected result Here at least V4.1F is closer to Kimi than to the trash tier. But the gap even to Opus 5 (which is, surprisingly, above Sol) is depressive.
elvis @omarsar0 a cité · 05/11 18:19
Can AI agents do scientific research on their own? Science's Last Exam is a new benchmark measure this. It's built from real research tasks across nine scientific fields. It tests whether an agent can form hypotheses, plan protocols, run scientific code and software, and read
12
R
Ravi Parikh @ravisparikh · 07/11 14:23
Dear Claude: you are being held back by human-centric graphics APIs. I've built a 3D and VFX engine where agents can write their own shaders, materials, cameras, physics. The core API fits in ~2,500 tokens. Vork is: • 2x more token-efficient vs Blender, Unreal Engine, Houdini
3 relayeurs à 3 · il y a 36 h Voir sur X
Qui l'a repris
小互” Linus ✦ Ekenstam 🔜 Adobe MAX Miami 2026” Peter Xing
Ce qu'ils en ont dit
小互 @xiaohu a cité · 07/11 14:23
绝对疯狂
Linus ✦ Ekenstam 🔜 Adobe MAX Miami 2026 @LinusEkenstam a cité · 07/11 09:13
This is absolutely insane 🤯 we’ve seen nothing yet.