Charbel-Raphaël Segerie

🤖 Deep Dive Workshop D: Collaboration Between Agents – Techniques for Multi-Agent LLMs and Their Operational Impact Join Donatien Chedom Fotso, Ph.D., AI & ML Team Lead at Deutsche Bank, at Generative AI EUROPE 2024 for an in-depth exploration of how multi-agent LLM systems can revolutionize business operations. Key highlights: *Advanced techniques for improving AI agent collaboration and interaction *Real-world case studies showcasing the operational impact of multi-agent systems *Interactive group discussions and scenario-based activities to explore potential applications *Ideal for AI researchers, developers, and tech strategists looking to push the boundaries of AI capabilities. 🚀 👉 Register now and enhance your AI solutions with multi-agent systems: https://lnkd.in/e783QTaV #GenAIEUROPE #AIConference

8

⭐️ New video release 📺: Refactoring Large Programs Watch Dr. Kristian Rother guide you through refactoring a large, messy, and undocumented Python program, offering valuable insights and strategies for your next big refactoring challenge. 📺 Watch the video on YouTube: https://lnkd.in/emjn58cb Kristian, a freelance Python trainer, presented a tutorial on refactoring large programs at a recent event. He addressed the challenge of cleaning up complex software with 10,000-100,000 lines of code, particularly in the case of legacy code. The tutorial focused on practical procedures such as code review, writing minimal tests, adding type annotations, and isolating cleanable parts from problematic sections. Kristian emphasized the systematic approach to large-scale refactoring, highlighting that while it is difficult and costly, it can be tackled methodically. The session, aimed at those with fluency in basic Python, included interactive elements and practical exercises for attendees to enhance their refactoring skills. Participants were provided with the messy code and refactoring guidelines through GitHub for hands-on learning.

1

⭐️ New video release 📺: Performant, scientific computation in 🐍Python and 🦀 Rust Watch Stefan Ulbrich as he guides you through building performant scientific computation packages in Python and Rust, offering insights into efficient computation and optimization for machine learning and data science applications. 📺 Watch the video on YouTube: https://lnkd.in/eq-BhHdH Stefan Ulbrich, Senior Data Scientist, offers a tutorial on enhancing scientific computation in Python and Rust. The session explores packaging numerical calculus and algorithms using modern tools. Highlighting the efficiency of vectorization in Python for computational tasks within Machine Learning, Ulbrich presents the integration of Rust to optimize package performance. Rust, recognized for its performance and memory safety, complements Python, particularly for instances prioritizing speed and modern language features. The talk demonstrates creating a scientific Python package, gradually transitioning components to Rust for enhanced performance. Through an example of implementing the "Expectation maximization for Gaussian Mixture Models" algorithm, attendees learn how to leverage modern tools to build, optimize, and extend packages efficiently. Ulbrich emphasizes that while Rust knowledge is not mandatory, the session provides a high-level understanding of its benefits for Python extension and performance improvement. For a hands-on experience, participants are encouraged to clone Ulbrich's repository for swift installation and to optimize session time effectively.

9

1 commentaire

📣 Add the 29th of May 18:30-20:00 CEST to your agenda to attend our online Python workshop NLP projects with spaCy given by 👋 Vincent Warmerdam, who currently works at :probabl. to collaborate on new tools for the scikit-learn ecosystem. Vincent is a senior data professional who worked as an engineer, researcher, team lead, and educator in the past. You might know him from tech talks where he attempts to defend common sense over hype in data science. He is especially interested in understanding algorithmic systems so that one may prevent failure. As such, he's had a preference for simpler solutions that scale, as opposed to the latest and greatest from the hype cycle. 💰 Why is it valuable to join this workshop? What will be covered? During this workshop, Vincent will guide you through a small project where he tried to detect programming languages that appear in transcripts from the popular Talk Python podcast. He will use spaCy for this and he will use this task as a motivating example to help kick the tires. You can ask him anything about the stack and he will also have some freebies to hand out at the end of the session. 📌 How to join: https://lnkd.in/eyAcYxu8 #python #nlp #spaCy #workshop

20

Have you heard of Retrieval Weighting? It might be the technique your RAG app is missing, leading to inaccurate responses. Don't miss out on these weighting techniques, discussed by Rodrigo Nader, that you can apply to your RAG app in a few minutes to gain significant improvements.

12

In this article, Baptiste L. shows how to leverage the power of reinforcement learning for feature selection when faced with very large datasets

64

2 commentaires

🔍 Understanding LLMs (Large Language Models) At eaQbe, we often get asked : " How smart are these AI models, really ? " Think of a Large Language Model ( LLM ) as a super organized helper. It knows a lot—thousands of books, websites, and conversations’ worth of information—but it doesn’t invent brand-new knowledge. It’s like sorting through a giant box of toys ( data ), figuring out what you want, and offering helpful suggestions. However, LLMs aren’t magic. They don’t create new " toys " from scratch. They are great at understanding and organizing what’s already out there - just like a powerful search engine on steroids. When we talk about AI at eaQbe, we emphasize this distinction. AI is incredibly powerful, but it’s not about being "superhuman" - it’s about enhancing how we use the knowledge we already have. Want to dive deeper into what makes AI, data mining, and deep learning tick ? Let's chat ! #AI #LLM #DataScience #UnderstandingAI #DataMining #DeepLearning #eaQbe

5

3 commentaires

New update ! Episode 8 from this summer: Reusing an existing pre-trained model  How to leverage the power of "transfer learning", a widely adopted technique in deep learning ?  https://lnkd.in/eBHX7UvM

4

🎉 𝐄𝐱𝐜𝐢𝐭𝐢𝐧𝐠 𝐁𝐫𝐞𝐚𝐤𝐭𝐡𝐫𝐨𝐮𝐠𝐡 𝐢𝐧 𝐀𝐈 𝐈𝐧𝐭𝐞𝐫𝐩𝐫𝐞𝐭𝐚𝐛𝐢𝐥𝐢𝐭𝐲! Last few weeks have seen a surge in AI Interpretability work: Anthropic researchers work on understanding Claude 3 models [Scaling Monosemanticity in AI](https://lnkd.in/eaYkgSNM) OpenAI released work on understanding GPT-4 [Extracting concepts from GPT-4] (https://lnkd.in/gFzijQwM) What's even more fascinating is their use of 𝘚𝘱𝘢𝘳𝘴𝘦 𝘈𝘶𝘵𝘰𝘌𝘯𝘤𝘰𝘥𝘦𝘳𝘴 (𝘚𝘈𝘌𝘴). Curious why SAEs are so effective and scalable? Our research at Martian provides some compelling insights: https://lnkd.in/gkvN2cXG At Martian, we leverage category theory—a mathematical approach focusing on relationships rather than object internals—to understand why SAEs perform so well. This theory underpins our broader efforts in "model mapping," a series of methods that promise enhanced interpretability without intensive manual analysis. 🔬 𝐌𝐨𝐝𝐞𝐥 𝐌𝐚𝐩𝐩𝐢𝐧𝐠: 𝐀 𝐂𝐨𝐦𝐩𝐮𝐭𝐚𝐭𝐢𝐨𝐧𝐚𝐥 𝐌𝐢𝐜𝐫𝐨𝐬𝐜𝐨𝐩𝐞 𝐟𝐨𝐫 𝐀𝐈 Our model mapping initiative acts like a microscope, revealing how AI models operate without dissecting their inner workings. This scalable approach benefits from increased computational power, paving the way for more efficient AI development. 🚀 𝐉𝐨𝐢𝐧 𝐌𝐚𝐫𝐭𝐢𝐚𝐧 𝐨𝐧 𝐎𝐮𝐫 𝐉𝐨𝐮𝐫𝐧𝐞𝐲! We're pushing the boundaries of AI interpretability and alignment. Learn more and get involved at [withmartian.com](https://withmartian.com).

31

These days I’ve been in Montpellier, France, to present our last research results on combinatorial optimization problems over temporal graphs to the 31st symposium on temporal representation and reasoning (TIME2024). Check out our paper at https://lnkd.in/da66ref4 Here we discuss a new heuristic-based approach for minimum timeline cover over temporal graphs. The paper provides theoretical results to a problem largely adopted in social network analysis as interpretable network summarization framework to find important events. Particularly, we used it to analyze job postings, to find the most required skills by companies and understand how the market is evolving. We are already working on evolutions of this approach with Deep Learning. Stay tuned for next publications. #knowledgegraph #combinatorialoptimization #networksummarization

15

1 commentaire

If you missed the Deep Dive on AI Agents with Olivier Dupuis, the recording is now on YouTube! This project is a great use case for those interested in leveraging AI agents in Dagster. Olivier discusses how to leverage AI agents for research aggregation and processing. Hex notebooks are central to this project due to their ability to integrate with Dagster and rapidly iterate with complex data workflows. Link in comments

17

1 commentaire

Explore our latest blog with our first ever Collaborator, Diogo Camacho : Learn how large language models can accelerate the development of machine learning models and revolutionize ML in research. 👇Click below to read more https://lnkd.in/eF8ycBmh #biostatistics #statistics #data #science #education #career #blog — At biostatistics.ca our mission is to create an information-rich community around biostatistics. Do you have interesting career stories, project successes or technical skills you would like to share with the community? Send us a message and become a certified biostatistics.ca writer!

12

🚀 Day 1 of Mastering LLMs: The Embedding Era! 🌟 Welcome to our brand-new "Mastering LLMs" series, where we explore the fascinating journey of language models, one breakthrough at a time. 🌍 Today, we’re diving into the Embedding Era, a game-changing chapter in NLP history. From the dense magic of Word2Vec and GloVe to the context-aware brilliance of ELMo, this era set the stage for everything that followed. Moreover, if you want to learn about embeddings, check this article: https://lnkd.in/gqXcSPVY 🔑 What you’ll learn: ✨ How embeddings reshaped text representation in 2013-2014. ✨ The genius behind static embeddings like Word2Vec and GloVe. ✨ The leap to contextual embeddings with ELMo. ✨ Why the Embedding Era was critical, despite its limitations. This is just the beginning. 🌟 Join us every day at 6 PM as we unlock the secrets of LLMs and their transformational journey! 💬 What’s your biggest takeaway from the Embedding Era? Share your thoughts in the comments below! #AnalyticsVidhya #MasteringLLMs #LLMs

268

6 commentaires

How to handle private data in ML datasets ? Here is out latest exciting experiment at Hugging Face 🤗 : Many datasets contain undocumented private information, making them difficult for ML practitioners to use. We started to use #Presidio to detect the presence of private data in HF Datasets, in particular Personal Identifiable Information (#PII). We now show a Presidio Report that estimates the amount of PII on dataset pages. This is useful e.g. to help further filter datasets or to take more informed decisions before training a model. More info on Presidio + examples in our latest blog post: https://lnkd.in/e5_mt7j7 We built this with the help of the community (special thanks Omri Mendels for the help with Presidio, Sara Hooker for the feedback on multilingual datasets, Margaret Mitchell and team for the help on the blog post !) and from our work with CNIL - Commission Nationale de l'Informatique et des Libertés and their guidance.

51

📣We’re proud to share that the 2024 Conference on Empirical Methods in Natural Language Processing (EMNLP) has accepted our scientific paper “Self-AMPLIFY: Improving Small Language Models with Self Post Hoc Explanations.” The accepted paper was written by Milan Bhan, PhD student and Domain Specialist at Ekimetrics, Nicolas Chesneau, Head of Innovation at Ekimetrics, Jean-Noël Vittaut, and Marie-Jeanne Lesot, both​ from LIP6 - Laboratoire d'informatique at Sorbonne Université.. The paper introduces Self-AMPLIFY, our three-step method to improve Small Language Models’ performance by automatically generating rationales from post hoc explanation methods. From November 12 to 16, Milan will be in Florida to present his paper at EMNLP 2024, a leading conference in natural language processing (NLP) and artificial intelligence. EMNLP 2024 accepting our paper for presentation highlights our expertise in NLP, specifically in GenAI, thanks to our research applied to business and our talents, who help us to make significant scientific advances. Want to learn more about our paper and EMNLP 2024? Click the link in the comments 👇 #EMNLP2024 #GenAI 

50

1 commentaire

Last chance to sign up for our movie recommender workshop! ⌛ Join us tomorrow at 6pm CEST for a 2-hour immersive session that will show you how to harness the power of recommendation systems using cutting-edge technology. So what are you waiting for? Register now: https://lnkd.in/d8Tt5f_X #Workshop #TensorFlow #pgvector #RecommendationSystems #DataScience

10

📈 Explore groundbreaking strategies in financial modeling with Antoine Febres-Gagné's latest piece on "𝗨𝘀𝗶𝗻𝗴 𝗞𝗼𝗹𝗺𝗼𝗴𝗼𝗿𝗼𝘃-𝗔𝗿𝗻𝗼𝗹𝗱 𝗡𝗲𝘁𝘄𝗼𝗿𝗸𝘀 (𝗞𝗔𝗡) 𝗮𝗻𝗱 𝗕𝗮𝗰𝗸𝘁𝗲𝘀𝘁𝗶𝗻𝗴 𝘁𝗼 𝗣𝗿𝗲𝗱𝗶𝗰𝘁 𝗦𝘁𝗼𝗰𝗸 𝗣𝗿𝗶𝗰𝗲𝘀." This insightful article dives into how neural networks can revolutionize stock price predictions through advanced backtesting, achieving an impressive 422.69% return over 1004 days. A must-read for enthusiasts and professionals in trading and machine learning. 🔗 #MachineLearning #Finance #StockMarket #TheDeepHub #DeepInData

6