Dimitri Charles

Ghostty is leaving GitHub. I'm GitHub user 1299, joined Feb 2008. I've visited GitHub almost every single day for over 18 years. It's never been a question for me where I'd put my projects: always GitHub. I'm super sad to say this, but its time to go. mitchellh.com/writing/ghostt…

Bonjour @DevoxxFR , où peut-on trouver la musique qui parle de Devoxx que vous passez avant chaque conférence ?

Happy to be back at @DevoxxFR this year.

Yann LeCun was right the entire time. And generative AI might be a dead end. For the last three years, the entire industry has been obsessed with building bigger LLMs. Trillions of parameters. Billions in compute. The theory was simple: if you make the model big enough, it will eventually understand how the world works. Yann LeCun said that was stupid. He argued that generative AI is fundamentally inefficient. When an AI predicts the next word, or generates the next pixel, it wastes massive amounts of compute on surface-level details. It memorizes patterns instead of learning the actual physics of reality. He proposed a different path: JEPA (Joint-Embedding Predictive Architecture). Instead of forcing the AI to paint the world pixel by pixel, JEPA forces it to predict abstract concepts. It predicts what happens next in a compressed "thought space." But for years, JEPA had a fatal flaw. It suffered from "representation collapse." Because the AI was allowed to simplify reality, it would cheat. It would simplify everything so much that a dog, a car, and a human all looked identical. It learned nothing. To fix it, engineers had to use insanely complex hacks, frozen encoders, and massive compute overheads. Until today. Researchers just dropped a paper called "LeWorldModel" (LeWM). They completely solved the collapse problem. They replaced the complex engineering hacks with a single, elegant mathematical regularizer. It forces the AI's internal "thoughts" into a perfect Gaussian distribution. The AI can no longer cheat. It is forced to understand the physical structure of reality to make its predictions. The results completely rewrite the economics of AI. LeWM didn't need a massive, centralized supercomputer. It has just 15 million parameters. It trains on a single, standard GPU in a few hours. Yet it plans 48x faster than massive foundation world models. It intrinsically understands physics. It instantly detects impossible events. We spent billions trying to force massive server farms to memorize the internet. Now, a tiny model running locally on a single graphics card is actually learning how the real world works.

@Ryo_Saeba_3 Moi j'ai bien aimé

Vu HUMINT, le dernier Ryoo Seung-Wan, sur Netflix. On sent qu'il essaye de faire quelque chose de plus maîtrisé niveau mélodrame mais c'est assez bancal. Par contre niveau action c'est un super hommage au Heroic Bloodshed et John Woo, c'est très efficace !

@IamMechano Bonjour, c'est dans tout le département ?

Brève coupure de courant (quelques secondes) à l'instant dans le 94.

AI Pro Tip 🧠💡 Using Claude Code with local models (e.g. Ollama)? Set CLAUDE_CODE_ATTRIBUTION_HEADER=0. It disables the x-anthropic-billing-header, preventing full prompt processing on every request. Result: 3×+ faster local inference 🚀 Any other tips except those below?

@_smontlouis C'est maintenant que tu vois ça. Claude m'a même dit d'aller dormir après plusieurs heures de coding.

c'est l'idée de qui ça ? un moment café ? seriously ? 🫥

The Paxos vs Raft conversation is interesting.

@benjilegnard Ça, ça va encore. J'ai vu pire tkt.

Je suis un peu saoulé, j'espère que ça ne se voit pas trop...

@benjilegnard Moi l'astuce que j'avais trouvé c'est d'installer sdkman et de faire pointer la config jdtls (si tu l'utilises) sur le lien symbolique sur la version configurée de Java par sdkman.

Faut que je prenne du temps pour bien configurer java/mvn dans neovim, mais pour l'instant ça a été ma pire expérience de config. Et je blame java, ses != versions de SDK, et ses annotations processors de partout. Les autres langages sont OK direct avec LSP présents 🤷

Le nombre de fois où je dois cliquer sur "reload from disk" dans Intellij IDEA est beaucoup trop élevé. damn-bitch-you-live-like-this.png

@LLCoolChris_ @benjilegnard Si tu as de la patience pour configurer ton jdtls, les keymap et les plugins associés oui il est possible de faire du Java dans nvim.

@tpierrain J'attends la sortie de la 2eme édition avec impatience. Très bon livre.

Essential reading for software engineers passionate about systems design and fundamentals 🤩

ironically the first thing ai killed was no-code everyone is now coding

my productivity skyrocketed when I stopped setup agents and started to write code again like a normal fucking human being

@Shirleyyych Un peu de ressemblance aussi avec Tomer Sisley

En France c’est donc Georges-Alain de la Star Academy 😂

@therealdanvega Rust

It's the weekend 🤩 If you have any free time after spending time with friends and family (and hopefully getting outside and soaking in the sun) what is one thing you're going to learn about?

Wrote up about my personal journey from AI skeptic to someone who finds a lot of value in it daily. My goal is to share a more measured approach to finding value in AI rather than the typical overly dramatic, hyped bait out there. mitchellh.com/writing/my-ai-…

Anytime someone asks me why I haven't changed my Toyota Camry.

Defending my Spring Boot Java app that uses 64GB RAM to return { "status": "ok" }