I will be speaking about CodeCarbon at @gosimfoundation Paris #LuckyDraw#

The Elon world is not my world, I’ve just done my #eXit! the X exodus is massive. Don't lose any of your followers. Thanks to #HelloQuitX I've registered for a journey to #BlueSky. Join us on app.helloquitx.com and automatically find your communities on #January20!

I will be there.

Le Projet de loi de finances pour 2025 a été publié hier soir. Vous pouvez dès ce matin visualiser ce que cela va changer et même réaliser vos propres modifications : socio-fiscal.leximpact.an.fr… Le nuit fut courte mais le résultat est là !

AI applied to software testing.

Today, we release several Moshi artifacts: a long technical report with all the details behind our model, weights for Moshi and its Mimi codec, along with streaming inference code in Pytorch, Rust and MLX. More details below 🧵 ⬇️ Paper: kyutai.org/Moshi.pdf Repo: github.com/kyutai-labs/moshi HuggingFace: huggingface.co/kmhf

I will be there for a talk about CodeCarbon.

ICYMI: xAI's new 100,000 GPU datacenter is an environmental catastrophe (and for what? "unwoke" LLMs?): cnbc.com/2024/08/28/musk-xai…

Introducing 🦀 CRAB: Cross-environment Agent Benchmark for Multimodal Language Model Agents 🦀 CRAB provides an end-to-end and easy-to-use framework to build multimodal agents, operate environments, and create benchmarks to evaluate them, featuring three key components: - 🔀 Cross-environment support - agents can operate tasks in 📱 Android and 💻 Ubuntu. - 🕸️ Graph evaluator - provides a fine-grain evaluation metric for agents. - 🤖 Task generation - composes subtasks to automatically generate tasks. By connecting all devices to agents, 🦀CRAB unlocks greater capabilities for human-like tasks than ever before. Use 🦀 CRAB to benchmark your multimodal agents! - 👨‍💻 Check out the repository: github.com/camel-ai/crab - 📝 Read the paper: arxiv.org/abs/2407.01511 - 🌐 Find out more via the project page: crab.camel-ai.org/ - 🐫 Join our community: discord.gg/8zQTBNqf97

Hey @nvidia - you've been mentioning a 'product carbon footprint analysis' in your sustainability reports for years now -- any chance you'll share that information with the community? Asking for a (concerned) friend 😇

AI Doomers are having zero impact on the development of advanced AI surveillance and autonomous weapons systems. Instead, they're intensely focused on restricting civilian access to AI. Ironically, this means they're ushering in the worst possible timeline for mankind. We'll have a world where all civilian systems are controlled, hobbled and deeply censored (aka "safe") and military, weapons systems and surveillance systems are hyper advanced. I don't want this timeline. Nobody else will either once they're forced to experience it in reality. See this post where Anduril just got 1.5B in additional funding to build advanced autonomous weapons systems (x.com/anduriltech?t=Mf-vEyvG… ). Also note that the EU AI bill has a 100% exemption for military/defense/surveillance. Guess what? So does every other bill, including SB1047. Threre is a zero percent chance that governments will restrict themselves from building advanced AI military and surveillance systems. There is not one single government on Earth that will restrict these technologies for themselves. Even if there was a pact, they would do it with black budgets just like the Total Information Awareness (en.wikipedia.org/wiki/Total_…) systems that were built in the US despite explicit restrictions from congress not to build them because mass surveillance was just too tempting for them. If you don't understand this, you don't understand much about life or human nature. By the way this world of advanced military systems is not coming. It's already here. China has an absolutely massive surveillance state that harnesses AI from top to bottom (facial recognition, gait detection, dissident tracking, predictive analytics) (economist.com/china/2023/11/…) and the war in the Ukraine is being fought with drones and AI repurposed from game systems (economist.com/leaders/2024/0…). There is even a newly appointed commander of drones (economist.com/europe/2024/07…). To be very clear, I'm not against AI military systems because I know the are an inevitable fact of life. I hate war. It's a disgusting and ugly waste of human life and it showcases the worst of what we are as a species. But I am a pragmatist to my core. I realize that no amount of wishful thinking will ever stop war or an escalation of military systems. Wars will be fought. Wars are won by having better stuff than the other guys and so I want my team to have the best systems. Simple as that. These systems can built and so they will be built. There is absolutely zero chance of stopping them. Restricting your own military development in the vain hopes that others will follow is foolishly naive. And yet that is exactly what many advocates for strangling American civilian AI believe. Helen Toner said "we don't have to worry about China" (former OpenAI board member and EA (yes you are EA, as you worked for an EA org and are continually funded by them and advocate their positions, Helen, despite your protests to the contrary) and Dan Hendrycks (whose team wrote the first draft of SB1047 and created a consulting org to profit from the bill) believes that by setting a "good example" that authoritarian regimes will just willing follow along to self-restrict development of advanced AI. Of all the ridiculous and stupid arguments of Doomers, this is perhaps the most absurd and frankly, stupidly naive thing I have ever heard in my life. It betrays an almost comically idiotic understanding of human nature and the way power works in the world. It's not just naive, it's dangerous. By pushing their cultural information warfare campaign with corrupted children's videos financed to the tune of 7-10M about AI destroying us all (x.com/DrTechlash/status/1821…) and using disgusting propaganda techniques like push polling (where the questions are knowingly and deliberately designed by AIPI to bias people against AI and NOT to collect an actual, realistic poll about people's real feeling about AI) (x.com/FLI_org/status/1821267…) they are pushing us right to the brink of the worst possible world. It's a world where your AI can't answer questions honestly because it's considered "harmful" (this kind of censorship always escalates), where information is gated instead of free, where open source models are killed off so university researchers can't work on medical segmentation (x.com/BoWang87/status/182102…) and curing cancer (budget conscious academics rely on open source models; they can fine tune them but can't afford to train their own) and where we have killer robots and drones but your personal AI is utterly hobbled and lobotomized. Resist this world at all costs. Protect access to civilian AI. Protect open source. Protect open weights. Fight for the future. If you can hear this, you are the resistance.

Les plus grandes sociétés du monde sont actuellement bloquées à cause d’une dysfonctionnement de mise à jour dans Windows… C’est le moment de réécouter ce débat mené par @LesEclaireurs_ sur la souveraineté numérique 👇

New open source OCR model just dropped! This one by Microsoft features the best text recognition I've seen in any open model and performs admirably on handwriting. It also handles a diverse range of vision tasks. You can play with it here: huggingface.co/spaces/gokayg…

I will be live in 3 hours to talk about the environmental impact of the Internet from #MozFest on youtube.com/live/lLJZ0il0UBM…

Participez au défi 𝟑𝟎 𝐣𝐨𝐮𝐫𝐬 𝐩𝐨𝐮𝐫 𝐟𝐚𝐢𝐫𝐞 𝐮𝐧 𝐝𝐨𝐧 𝐚̀ @dataforgood_fr ! ça fait du bien ! Du bien commun ! Pour donner c'est ici 👇 swll.to/donner_a_data_for_go… #dataforgood #techforgood

AI Music generation : done. Did you ear something to improve?

Chez 𝐃𝐚𝐭𝐚 𝐅𝐨𝐫 𝐆𝐨𝐨𝐝 les saisons se suivent mais ne se ressemblent pas ! Qu'y aura t-il au programme des 3 prochains mois ? Lutte contre la pêche frauduleuse, évasion fiscale, transition énergétique...et ce n'est qu'un aperçu ! Inscription👇 linkedin.com/events/71522996…