Amine Saboni retweeted
Your pipeline has more access than your CTO and gets less review than a typo fix.
Meet Plumber: open source, one command, 30 seconds.
Amine Saboni retweeted
It started with a DM with @roxannevarza after I launched @looties_io in April.
Now it is The Tech Merch Museum with @joinstationf
July 28 to August 22, 100+ objects: hoodies, stickers, mascots, pins, rare stuff that would usually live in a box forever.
Come see it in Paris!
Amine Saboni retweeted
On might feel that the AI coding era diminishes the value of a library like @scikit_learn, as coding is easy.
It's the reverse: one needs reliable building blocks, and fighting with Claude for eg reliable convergence of LogisicRegression takes energy, expertise, and tokens
Amine Saboni retweeted
Excited to launch the accompanying free RLHF Course for my book. To kick it off, I've released:
- Welcome video
- Lecture 1: Overview of RLHF & Post-training
- Lecture 2: IFT, Reward Models, Rejection Sampling
- Lecture 3: RL Math
- Lecture 4: RL Implementation
I'm going to add question & answer videos throughout the lecture to go deeper on topics that need it, and potentially cover some topics that are too recent and in flux to go in print. I expect 10-15 videos in total over the next few months.
At the same time, development around the code for the book is picking up. It's a great time to build the foundation for post-training methods.
YT playlist and course landing page below.
Amine Saboni retweeted
I just published a post on why I think fakes are more powerful than both mocks and Testcontainers
pierrezemb.fr/posts/why-fake…
It’s official: llm-d has joined the @CNCF! 🚀
Our mission to evolve Kubernetes into SOTA AI infrastructure just got a massive boost. This milestone belongs to our amazing community. Thank you for building this with us. 💜
We’re just getting started!
🔗 cncf.io/blog/2026/03/24/welc…
Amine Saboni retweeted
The persisting importance of prompt engineering -- and now harness engineering -- is one of the best indicators of how far we are from AGI. A general system doesn't need a task-specific harness. And when provided with instructions, it is robust to phrasing variations.
Amine Saboni retweeted
We've open-sourced many algorithms! Check them out!
🚀 Release 0.3.2: We’ve added some new algorithms to Pruna OSS!
- Recoverers
- Compilers
- Decoders
- And many, many more…
📚 Check out the docs: dub.sh/Ht2ryVg
⭐️ Star our OSS GitHub: dub.sh/wqToxlX
💜 Sign up for our events: dub.sh/nRlyHuU
Okay Junyang is not Qwen, and Qwen is not Junyang. Behind every great model is a team grinding through data pipelines, training runs, and sleepless launches. But he was the voice, the bridge, the person who made the global AI dev community feel like Qwen was theirs too. That kind of developer trust takes years to earn and seconds to lose.
What bothers me is the pattern. Big companies talk about valuing people holistically, then punish the very qualities that made those people valuable. A first-principles thinker inside a non-first-principles system doesn't just overburnt - they suffocate. And the system never sees it as its own failure.
In big corp, the hardest part was never the tech- they have talents - It was watching talented people get squeezed between what they know is right and what the org will allow. That gap is where you lose your best.
Wherever he lands next, the community will follow. That should tell Alibaba everything.
Enfin un interlocuteur sérieux autour de la table... !
Patrick Chaize désigné rapporteur de la PPL sur l’implantation des centres de données l.contexte.com/uQL
Amine Saboni retweeted
Replying to @SebastienBubeck
Best research environment?
Except for the whole "don't tell anyone about your research" part.
Research in secret is not research.
To celebrate together the release of FLUX.2 [flex] in collaboration with @bfl_ml. We’re launching the FLUX.2 [flex] Design Challenge! 🎨
𝗧𝗵𝗲𝗺𝗲
𝗠𝗮𝗸𝗲 𝗮𝗻 𝗶𝗻𝗳𝗼𝗴𝗿𝗮𝗽𝗵𝗶𝗰 𝗼𝗳 𝗵𝗼𝘄 𝘁𝗼 𝗴𝗿𝗼𝘄 𝗮 𝗯𝗹𝗮𝗰𝗸 𝗳𝗼𝗿𝗲𝘀𝘁 𝗼𝗳 𝗽𝗹𝘂𝗺𝘀
𝗛𝗼𝘄 𝘁𝗼 𝗽𝗮𝗿𝘁𝗶𝗰𝗶𝗽𝗮𝘁𝗲
1️⃣ Answer with a single creative infographic using FLUX.2 [flex] and the screenshot showcasing it. You can try it here: bfl.ai/models/flux-2
2️⃣ Follow the @PrunaAI and @bfl_ml account in X.
3️⃣ Mention us and add the hashtag #flux2flex.
𝗣𝗿𝗶𝘇𝗲
Popularity + Judge’s evaluation
🥇 €150 🥈€100 🥉 €50
𝗗𝗮𝘁𝗲𝘀
Until February 6th (23:59 CET)
𝗝𝗼𝗶𝗻 𝘁𝗵𝗲 𝗰𝗵𝗮𝗹𝗹𝗲𝗻𝗴𝗲, 𝘃𝗼𝘁𝗲 𝗳𝗼𝗿 𝘆𝗼𝘂𝗿 𝗳𝗮𝘃𝗼𝗿𝗶𝘁𝗲𝘀, 𝗮𝗻𝗱 𝗶𝗻𝘀𝗽𝗶𝗿𝗲 𝘁𝗵𝗲 𝗰𝗼𝗺𝗺𝘂𝗻𝗶𝘁𝘆!
Read the rules in detail: pruna.ai/blog/flux2flex-chal…
Amine Saboni retweeted
🇫🇷 Atos, OVH... L’Etat vient d’établir une première sélection de candidats en vue de désigner un hébergeur souverain au « Health data hub », afin de remplacer Microsoft. @Reesmarc & @empaquette dévoilent la liste des candidats.
▶️ l.linforme.com/yckn3vsz
Amine Saboni retweeted
Our best model for text and typography, FLUX.2 [flex], is now up to 3x faster.
FLUX.2 [flex] delivers highly accurate text rendering, custom typography, richer colors, and smoother visuals. We worked with @PrunaAI to make it up to 3x faster.
Examples in the thread:
Amine Saboni retweeted
Hi, I’m Minette 👋 Since many of you don’t know me yet, I’m starting a “Story of Me” series.
Every Monday, I’ll share how I moved to Paris and became a DevRel — the ups and the downs.
Stay tuned 👀
Amine Saboni retweeted
29.7 million downloads in 6 months.
That's 165,000 downloads per day.
6,875 downloads per hour.
114 downloads per minute.
Every minute, someone, somewhere, pulls an OpenMed model.
Full story: 👇
Amine Saboni retweeted
🚀 Qwen Layered just got faster! Big thanks to @PrunaAI
🎨 Qwen-Image-Layered is LIVE — native image decomposition, fully open-sourced!
✨ Why it stands out
✅ Photoshop-grade layering
Physically isolated RGBA layers with true native editability
✅ Prompt-controlled structure
Explicitly specify 3–10 layers — from coarse layouts to fine-grained details
✅ Infinite decomposition
Keep drilling down: layers within layers, to any depth of detail
🤗 Hugging Face: huggingface.co/Qwen/Qwen-Ima…
🧩 ModelScope: modelscope.cn/models/Qwen/Qw…
💻 GitHub: github.com/QwenLM/Qwen-Image…
📝 Blog: qwen.ai/blog?id=qwen-image-l…
📄 Technical Report: arxiv.org/abs/2512.15603
🚀 Demo (HF): huggingface.co/spaces/Qwen/Q…
🚀 Demo (ModelScope): modelscope.cn/studios/Qwen/Q…

































