Olivier Jeunen’s Post

The RL core of #DeepSeek is now replicated in less than $30. The authors implement both GRPO and PPO. The famous Aha moment of the DeepSeek can be observed by any student with a small base model. https://lnkd.in/dtA2AwRk

Just finished taking the RAG++ course from Weitghts&Biases. The course presented a lot of interesting approaches to optimize the data ingestion, evaluation and latency in rag applications. The classes are very short, and the best of all FREE. Strongly recommend if you build RAG systems in your day-to-day. #generativeAI #RAG https://lnkd.in/db2ia-YK

🚀 Exciting Update! 🚀 I’m thrilled to share my latest detailed presentation on K-Nearest Neighbors (KNN)! 📊💡 KNN is one of the simplest and most powerful algorithms in machine learning, and I’ve broken it down to help demystify its concepts. Whether you're new to data science or looking to refresh your knowledge, I hope this presentation makes KNN more accessible and practical for everyone! A special thank you to WeSolutize Technologies and My Research Room for providing me with the opportunity to explore, learn, and create valuable content. Grateful for the constant support and growth environment here. 🙏 Feel free to check out the presentation, and I’d love to hear your thoughts or any feedback! 😊 #MachineLearning #DataScience #KNN #ArtificialIntelligence #TechCommunity #Grateful #LearningJourney #DataDriven

Corporations, how do you make sense of the data you already have? If you have data in linked relational databases, you potentially have a knowledge graph. This knowledge can be ingested by a graph neural network to help you predict patterns in the data, with implications from drug discovery to financial fraud detection, from oil field exploration to agentic large language models. In this video, Prof Jure Leskovec talks about the highlights of last year's Data Science Applications annual meeting, also called the Graph Learning Workshop. 🎥 : Nancy Narimatsu https://dsa.stanford.edu https://lnkd.in/gZgzAKfK

🔰🔰 Happy to share my Second Project! 🔰🔰 I have developed and deployed a User Behavior Analysis and Video Recommendation System from the Machine Learning Use Case. This project performs User behavior analysis and generate video recommendations using K-Means Clusters and LSTM model. This system is designed to analyze user behavior patterns and deliver personalized video recommendations. 🎥✨ This project, which combines machine learning and data science, is now live and interactive on Streamlit. Explore the App Here 🔎: https://lnkd.in/g_wEzHVB Github Repository Link 🔗:  https://lnkd.in/gDqyW3MZ #DataScience #MachineLearning #UserExperience #StreamingData #BehaviorPatterns #VideoPredictions #LSTM #KmeansClustering #LinkedInProjects Take a look in the video below, demonstrating the capabilities and user experience of the video recommendation system.

🚀 Excited to share our latest project: "Classification of Documents using Graph-based Features and KNN"! 📄🔍 🔍 In today's digital age, efficient document classification is paramount for organizing and analyzing vast amounts of data. That's why our team delved into exploring innovative methods to streamline this process. 📊 Leveraging graph-based features and the power of KNN (K-Nearest Neighbors), we've developed a robust framework for document classification that enhances accuracy and scalability. 📈💡 💡 By representing documents as graphs and extracting meaningful features, we've unlocked new insights into document similarity and categorization. Our approach not only improves classification accuracy but also enables more nuanced understanding of document relationships. 👨💻 Huge shoutout to our instructor for their invaluable guidance throughout this journey! And of course, couldn't have done it without the dedication and collaboration of my brilliant group member Amna Zafar! 🌟 🔗 Check out our project for a deeper dive into the intersection of graph theory, machine learning, and document classification. Let's revolutionize how we organize and analyze information! 💪💼 #DocumentClassification #GraphFeatures #KNN #MachineLearning #DataScience #Teamwork

A recent article discusses "Dust: Unified Text-to-SQL for Datalakes, Spreadsheets, and CSVs," a new framework designed to enhance user interaction with diverse data sources through natural language queries. The system effectively translates these queries into SQL, facilitating seamless data retrieval and manipulation across various formats, including datalakes, spreadsheets, and CSV files. This unified approach aims to simplify data access and analysis, making it more user-friendly for individuals who may not possess extensive technical expertise. Such advancements could significantly democratize data interaction, empowering a broader range of users to leverage data for informed decision-making. #DataAccessibility #NaturalLanguageProcessing #SQL #DataManagement ---------------------- Learn more here: https://lnkd.in/eBBqssw2

Last week peer learning session for my ALX Data Science community focused on essential machine learning algorithms, including: 👉 Support Vector Machines (SVM): Finding the optimal boundary for classification tasks. 👉 K-Nearest Neighbors (KNN): Classifying or predicting outcomes based on proximity to data points. 👉 Naive Bayes: A probabilistic classifier based on Bayes' theorem. We also dived into hyperparameter tuning, exploring how selecting the right configurations can significantly improve model performance. One key takeaway from the session was how well my peers grasped complex concepts when broken down into real-world analogies and practical code examples. 📊 Sharing this knowledge reinforced my passion for peer learning and teaching data science concepts in an accessible way. 💡 I've attached the slides I created for the session to give a glimpse into the content covered. If you're passionate about data science or machine learning, I'd love to connect and share ideas! #DataScience #MachineLearning #PeerLearning #ALX

We’ve teamed up with Data Science Dojo for an #LLM bootcamp where Bernease Herman, Senior Data Scientist at WhyLabs, will guide participants through important techniques to build safe and secure LLM applications. This in-person event is a fantastic opportunity to learn from industry leaders on topics like LLM fundamentals, embeddings and vector databases, prompt engineering, observability, evaluation and much more. By the end of the week, you’ll be capable of building LLM-powered applications on any dataset of your choice. 💪 Learn more about the curriculum and secure your spot: https://bit.ly/3syASaH #LLMs #GenerativeAI #ResponsibleAI

#LUD23 Today I learned about a vector store retriever, a tool designed to efficiently retrieve documents using advanced search methods, such as similarity search and Maximal Marginal Relevance (MMR). Key areas I explored: - Retrieval techniques - Integration with language models for improved performance - Real-world applications and use cases I will be implementing this in my projects to enhance information retrieval. By integrating with language models, vector store retrievers enable users to fetch contextually relevant documents that significantly assist in generating insightful responses. Resource: https://lnkd.in/enYsJg2b #LearningUtsav #LearningFestival Learning Utsav