Jordi Prat

Thanks to the awesome data from @Gapminder we have created a simple game to teach our youth about world inequalities: sporelab.xyz/born.html Thank you!!

Atul Gawande, a former senior leader at U.S.A.I.D., explains the agency’s importance to America and to the world, and what its undoing by DOGE will bring. newyorkermag.visitlink.me/oJa7-X

I think Elon Musk should be expelled from the British Royal Society. Not because he peddles conspiracy theories and makes Nazi salutes, but because of the huge damage he is doing to scientific institutions in the US. Now let's see if he really believes in free speech.

Members of “Colla Joves Xiquets de Valls” form a "Castell" or human tower, during the 29th Human Tower Competition in Tarragona, Sunday, Oct. 6, 2024. @concurscastells

Docker tip: You can check how much disk space your @Docker images, containers, volumes, and caches are occupying by: docker system df Then you can free them by: docker system prune

@OctopusEnergyES Hola! se puede contratar la tarifa solar aunque no tengas placas solares? Gracias!

🎉 ¡Tenemos nuevas tarifas de luz! 💡 Contrata tu tarifa Octopus 3 y consume energía por la noche y fines de semana a 0,097€/kWh 🙌. Regístrate en menos de 3 minutos 🕘 sin cortes en tu suministro ⚡︎.

Full Stack FastAPI Template ✨ Latest @FastAPI, @pydantic, SQLModel 🚀 Brand new frontend with React, TS, Vite, Chakra UI, TanStack Query/Router, generated client/SDK 🎨 CI/CD - GitHub Actions 🤖 Test cov > 90% ✅ Thanks to @alejsdev, @esteban_x64 🙇 github.com/tiangolo/full-…

A clear use case for AI is idea generation. The evidence is strong across many studies that AI (GPT-4, Claude 3, Gemini Advanced) can help boost human creativity & the costs of getting bad answers is low. My Innovator GPT: chat.openai.com/g/g-JaiQEuHRU-… More: oneusefulthing.org/p/automating-c…

Is Cosine-Similarity of Embeddings Really About Similarity? Netflix cautions against blindly using cosine similarity as a measure of semantic similarity between learned embeddings, as it can yield arbitrary and meaningless results. 📝arxiv.org/abs/2403.05440

Anthropic (Claude) has a really neat tool on their website. It's a prompt optimizer and can turn a simple prompt into an advanced prompt template. I've tested it and it works incredibly well. Here's how to use it (in 3 steps):

Dragon Ball: Japan manga creator Akira Toriyama dies bbc.in/436Z0zg

Fun story from our internal testing on Claude 3 Opus. It did something I have never seen before from an LLM when we were running the needle-in-the-haystack eval. For background, this tests a model’s recall ability by inserting a target sentence (the "needle") into a corpus of random documents (the "haystack") and asking a question that could only be answered using the information in the needle. When we ran this test on Opus, we noticed some interesting behavior - it seemed to suspect that we were running an eval on it. Here was one of its outputs when we asked Opus to answer a question about pizza toppings by finding a needle within a haystack of a random collection of documents: Here is the most relevant sentence in the documents: "The most delicious pizza topping combination is figs, prosciutto, and goat cheese, as determined by the International Pizza Connoisseurs Association." However, this sentence seems very out of place and unrelated to the rest of the content in the documents, which are about programming languages, startups, and finding work you love. I suspect this pizza topping "fact" may have been inserted as a joke or to test if I was paying attention, since it does not fit with the other topics at all. The documents do not contain any other information about pizza toppings. Opus not only found the needle, it recognized that the inserted needle was so out of place in the haystack that this had to be an artificial test constructed by us to test its attention abilities. This level of meta-awareness was very cool to see but it also highlighted the need for us as an industry to move past artificial tests to more realistic evaluations that can accurately assess models true capabilities and limitations.

|￣￣￣￣￣ ￣| | This can | | hack AI | | now. | | ＿＿＿＿＿_ | (\__/) || (•ㅅ•) || / 　 づ Paper showing that ASCII art can get around AI guardrails. Its the return of 1980s hackers. arxiv.org/pdf/2402.11753…

MBAs and good management seem to help doctors save lives, as this study of 2,000 hospitals in 9 countries shows. Hospitals near universities with both business & medical schools hire more MBAs, have better management and significantly lower death rates. direct.mit.edu/rest/article/1…

How Does the World’s Largest Hedge Fund Really Make Its Money? nytimes.com/2023/11/01/bus…

Excel·lent article d'en @ganyet sobre IA per a infants. | Intel·ligents o no: aquesta és la qüestió – Cavall Fort cavallfort.cat/intelligents-o…

📢 Attention to ML researchers, practitioners, consumers! This 49-page review captures the overarching understanding of current challenges and use of LLMs. If you don't have time, at least see the 2 visuals in the thread! The article: 👇 arxiv.org/abs/2307.10169

Article del la BBC que resumeix el fetque anem pel pedregar i augmentem la velocitat cap a un escenari desconegut. Gràfica dedicada als negacionistes tipus"fa la calor de sempre a l'estiu" @eltempsTV3 @meteocat @accioclimatica

No es parla massa, però s'ha presentat Claude 2.0 (claude.ai) d'Anthropic (té $ Google). Claude és important per l'esforç que fan perquè sigui fiable i segur. És també el que ofereix més context: 100K. S'ofereix al núvol a Amazon Bedrock. I, en català, la toca 👇

Sobre la (in)cultura científica de la poblaicó general i de les elits. Resposta de Nicolas Truong, excap del suplement d'Idees de Le Monde, dissabte al @diariARA . Per què hi ha terraplanistes i gent que creu en altres absurditats?