Rodrigo Rivera

@MaJorNashTW @DotCSV @grok Uy te ghosteo

@DotCSV Che @grok buscale un modelo preentrenado actual para obtener embeddings de animaciones de manos

Hace no tanto ChatGPT hubiera preferido alucinar una variedad de opciones inexistentes o cortarse los cables de alimentación antes que darme una respuesta negativa como primera respuesta.

@MoureDev Cool! Aunque unos pixeles más en la imagen habrían estado mejor 😁

El artículo oficial donde OpenAI nos dice para qué utilizar cada uno de sus modelos, y el mapa mental creado por MindBranches para entenderlos aún mejor. → help.openai.com/en/articles/11…

@vikhyatk @robmarkcole Ah yes, because open-source maintainers are just sitting around waiting for your personal emergencies. Why should they waste time sleeping, working, or living when they could be immediately fixing bugs for free? Maybe next time they'll hire a 24/7 hotline just for you!

@robmarkcole normally i'd agree, but not when the bug is deleting files on my computer. i'll simply choose not to use it

@marpedev @AbelOscopri @5eniorDeveloper Me suena más a problema de autoestima que otra cosa 😒

Entras a una reunión con 10 personas, de las cuales, 9 hablan español y 1 habla inglés... y por esa persona, las otras 9 deben hablar inglés. Así funciona el mundo!

Wow, diffusion models (used in AI image generation) are also game engines - a type of world simulation. By predicting the next frame of the classic shooter DOOM, you get a playable game at 20 fps without any underlying real game engine. This video is from the diffusion model.

@svpino eMacs is the way !

Lots of people tell me that Cursor + Claude is the best dev setup right now. I stopped using Cursor a while back. It was too overwhelming. Trying it again. Looking impressive so far.

@svpino @hridayakuhara Probably stating clearly in the first line that the content is promoted ? We don't know how many companies contact you so this 10% tell us nothing of the kind of selection you made.

Hey man, any time I promote something, I'll disclose it in the post: "This company is collaborating with me." I accept about 10% of the companies that contact me to review their services. I try to ensure I always post something that's useful for my audience. I talk about Zerve because I've tried it, and I genuinely think it's useful and super cool. It's also free. How can I improve these posts so you find them more useful?

If something could kill Jupyter notebooks, this would be it. Zerve is a web-based IDE. It's block-based, like a notebook, but it follows a different philosophy, and the experience is very different. And they just released a killer feature! You can now drag and drop any model from HuggingFace, a dataset, and fine-tune it using your infrastructure and GPUs. No privacy concerns, no data leaks, and no need to manage any infrastructure. Fine-tune your model with your data and host it in your environment without leaving the tool. A lot is happening in the attached screenshot, but that's an end-to-end process to fine-tune Gemma 2B. This is huge, but that's not all: 1. Zerve is web-based, so you can access your work anywhere 2. Collaboration is a first-class citizen 3. You can deploy your code directly 4. It promotes code modularity by using blocks 5. You can use multiple languages together 6. You can run code blocks in parallel Zerve is new, but it has the potential to change how data scientists and machine learning practitioners write code. I'm rooting for them. The tool is free. You can start right now by visiting. Thanks to their team for collaborating with me on this post.

I recorded a new YouTube video to teach you how to evaluate a RAG application. And we'll do it step by step. Starting from scratch. 8 out of 10 people I talk to are evaluating their LLM-powered systems manually. This is wild! They try a few samples and deploy the system if the answers look good. It reminds me of people testing the UI of an application by just "looking at it" from time to time. Please, don't do this. In this video, I'll show you how you can build automated tests for a simple RAG system that answers questions from a website. It's a 50-minute video. My goal is not to show you the code but to help you understand everything that's happening. Here is the link to the video: youtu.be/ZPX3W77h_1E I'm using @langchain and @giskard_ai to implement the evaluation process. Giskard is an open-source library that will help you with the following: 1. Generate test cases automatically. Each test case consists of a question, a ground-truth answer, and a reference context. 2. It will run every test case and point out problematic topics and RAG components that need improvement. 3. It will show recommendations to improve the system. It's a great library! Star their repository here: github.com/Giskard-AI/gis… Hope you enjoy the video!

Reinforcement learning explained twitter.com/i/status/17310…

This new open-source library is magic! It helps you detect any of the following issues with Large Language Models: • Hallucinations • Misinformation • Harmfulness • Stereotypes • Disclosures • Prompt injections The library will help you find problems with your model you didn't know you had. If you try to do this by hand, it'll take you weeks. Here is how it works: The library is Giskard. It's fully open-source. Star the repository here: github.com/Giskard-AI/gis…. Here is an example where we detect issues on a Retrieval Augmented Generation (RAG) task using Langchain and OpenAI GPT-3.5. We load a PDF document, ask questions about it, and run a scan to find any vulnerabilities. colab.research.google.com/github/giskard… Giskard works with any Python model, including PyTorch, TensorFlow, HuggingFace, Scikit-Learn, XGBoost, and LangChain. They also released integrations with HuggingFace Spaces, Weights & Biases, and MLFlow. Seeing more and more open-source libraries tackling really complex problems is fantastic! Thanks to the @giskard_ai team for partnering with me on this post. Are there any other open-source libraries you'd recommend for machine learning? (Obviously, something different from the usual suspects.)

Live tracking of the 8th Balloon during the first AIRCORE campaign in South America. Flying over Toledo, Bolivia @LSCE @IPSL_outreach @ird_fr @INSU_CNRS @obs4clim

First AIRCORE measurements in South America performed at Oruro (Bolivia) to track CO2,CH4, CO, H2O up to 34 km in the stratosphere. @lsce @umsa @ovsq @IPSL_outreach @ird_fr @obs4clim

@freddier Esos son temas bastante amplios, crees que es posible convertirse en experto a un nivel profundo abarcando ese amplio espectro de temas ? O si en lugar de eso te estás enfocando en puntos específicos de cada tema, cuales son?

⭐️ La gente más exitosa que conozco quiere ser experta en uno o varios temas (no muchos). Yo me enfoqué en software, Latam, educación y tech. ¿Qué tema(s) estás en proceso de dominar a un nivel profundo y experto?

This new library will 2x your Machine Learning workflow. It's open-source, and in 30 seconds, it'll find problems with your model you didn't know you had. I've tried doing this by hand. It takes weeks. Read on for the details:

I prefer: MANGA 💕💖✨

🌱 This research paves the way for more accessible & affordable CH4 monitoring solutions, crucial for reducing emissions & protecting our planet. 🌍💚 (10/10) #MethaneMonitoring #LowCostSensors

🎉 Our findings demonstrate that low-cost sensors can effectively monitor CH4 peaks, offering a promising alternative to expensive research-level instruments! 🙌 (9/10)

📢 Our new paper was published yesterday, examining the potential of low-cost sensors in reconstructing high-frequency methane atmospheric concentration peaks! 🌍📊🤔 Want to know our goal, methodology, and results?👇 (1/10)