Olivier Mertens’ Post

🆕🆕🆕🆕🆕 Google launches Duet AI for Developers, its AI tool for developers: the code assistant is available in various IDEs, such as VS Code and JetBrains (IntelliJ, PyCharm, GoLand, WebStorm), with support for more than 20 languages, including Go, Java, Javascript, Python and SQL. In addition to assisting in code creation, the tool will support automated testing and vulnerability remediation. The system will be available for free until the end of January 2024, when it will cost 19 dollars/month, with a subscription commitment for one year. The information is from the Google Cloud site. #canada #calgary #AI #duetai #bard #grok #openai #chatgpt #jetbrains #intellij #pycharm #goland #webstorm #golangdeveloper #java #javascript #python #sql #developer #softwaredevelopment #softwaredeveloper #softwareengineering #codinglife #it #informationtechnology

🚀 I thrilled to announce the official release of AdaptiveBridge V1.0.0: A Python library for ML engineering that's now ready for production use! (Beta stage so... for seniors only). 📈 In the fast-paced world of machine learning, accurate predictions are paramount. However, dealing with incomplete data can pose significant challenges. AdaptiveBridge is the solution to bridge the gap caused by missing data and features, ensuring your predictions remain reliable and actionable. 🌐 The Power of AdaptiveBridge: Discover how AdaptiveBridge tackles the NaN issue head-on in our latest Medium article, "Navigating the Maze of Features: Stop Wasting Money in Production Blitz." 📊 Benchmark Results: We've included benchmark results in the documentations: in general Minimize more than 50% of data requirements on expense of ~2% accuracy. For us, the geeks, please go directly to Github: https://lnkd.in/eiRR4qKU 🌐🔍#AdaptiveBridge #DataScience #MachineLearning #DataAnalytics #PythonLibrary #DataPrediction #DataCompleteness #Innovation

Unlock the power of #innovation with the Top 10 #AI tools for software #development. From TensorFlow and PyTorch for #ML models to Keras and Scikit-learn for seamless coding, these tools redefine possibilities. Embrace AI-driven solutions for smarter, faster, and more efficient development. #aitools2023 #aitoolsforbusiness #aitools #aidevelopment #aisolutions #aiadvancements #workflow #aiadoption #ai4good #aiforhealth #aiartcommunity #aiapplications #softwaredevelopment #customapp #productinnovation #productivity #itinnovation #itindustry #top10

LangGraph 🤘rocks over LCEL Use LangGraph for building intelligent agents that make decisions! 🚀 Learn how to create dynamic workflows with LLMs. Ever found yourself wishing you could fine-tune your AI workflows for greater control and flexibility? 🤔 Enter LangGraph, a framework built on LangChain that empowers you to build and customize "agent swarms" 🤯. Unlike higher-level frameworks, LangGraph provides a deeper dive into agent orchestration, allowing for granular control over agent behavior and interaction. The magic unfolds with "nodes" and "edges" ✨. Nodes represent functions performing specific tasks within the graph, while edges define the connections and flow between them. Think of it as a dynamic flowchart where data, represented as "state," is passed between nodes during execution. But wait, there's more! 🤩 LangGraph introduces "conditional edges," enabling workflows to adapt based on the input. Imagine a chatbot that seamlessly integrates with a weather API ☀️🌧️. Using conditional edges, the agent can decide whether to fetch weather data based on the user's query. And if you're dealing with tasks requiring iterative actions, LangGraph has you covered with its support for "cycles" 🔄. This sets it apart from traditional expression languages, allowing for more complex and dynamic agent interactions. LangGraph might have a steeper learning curve, but the payoff is immense. It's time to unlock the true potential of your AI agents! 🚀 #LangChain #LLM #AI #Agents #WorkflowAutomation #LangGraph #Python #Coding #AIInnovation #TechTrends #DataScience #MachineLearning #AIEnthusiast #FutureofWork #AICommunity #OpenAI #ChatGPT #AItools #SoftwareDevelopment #TechTalk #LearnAI #AIExperts #TechInnovation #DigitalTransformation #AIRevolution #FutureIsNow #DataDriven #AIForGood

#MLOps is not about specific tools. It is about gluing different components together to ensure traceability and reproducibility. In other words, for every #machinelearning model deployment/ run in any environment it should be possible to look up unambiguously: ➡ Corresponding code/commit on git ➡ Infrastructure used for training and serving ➡ Environment used for training and serving ➡ ML model artifacts ➡ What data was used to train the model To make it possible, my whole career I was building Python packages, code templates, infrastructure integrations, and reusable CI/CD pipelines. I was building "the glue" for the #datascience teams, every time with a different set of tools. ZenML got this philosophy right. Their product is an abstraction layer on top of the user's stack. ZenML has an impressive list of integrations. Here are some of them: ➡ Data annotator: Argilla, Label Studio, Pigeon, Prodigy ➡ Data validator: WhyLabs, Deepchecks, Evidently AI, Great Expectations ➡ Feature store: Feast ➡ Experiment tracker: Comet, mlflow, Neptune.ai, Weights & Biases Check out their components guide: https://lnkd.in/eDVzeKxd And the full list of integrations: https://lnkd.in/ebBtVXy8

Jupyter Notebook is now powered by AI

🚀 Exciting News in the AI World! 🚀 Deepseek has just released Deepseek Coder V2, This model sets new standards, surpassing GPT-4 Turbo, Gemini-1.5Pro, Claude3-Opus, and Codestral in code generation and math tasks. Here are some impressive highlights of Deepseek Coder V2: - 338 Programming Languages Supported - 128k Context Length - 238B Parameters available through Hugging Face and Deepseek API - 16B parameters available on Ollama I’ll be using the 16B parameter version locally with Open WebUI. I’m looking forward to exploring its capabilities and seeing the productivity boost it can bring! I’m grateful that these LLMs are open source, allowing the creativity of the development community to make magic happen. #DeepseekCoderV2 #LLM #AI #MachineLearning #CodeGeneration #TechInnovation #OpenWebUI #OpenSource

🚀 Exploring AI Personalization: Fine-Tuning Gemma-7B with QLoRA for My Personal Copilot! 🧠 I'm excited to share my learning journey of using QLoRA to fine-tune the Gemma-7B model and create a personalized AI copilot. This project leverages the efficiency of QLoRA to optimize the model for my specific tasks, making it a highly effective and resource-efficient assistant. 🔍 Key Highlights: Model: Gemma-7B Technique: QLoRA  Goal: Develop a personal copilot focused on my private repo, optimized for performance while reducing computational resources. Hardware: RTX 4090 By using my private repo as a RAG reference, I'm enabling Gemma to assist me in developing Python Flask applications. If you're concerned about latency or exposing your production code, deploying an open-source LLM locally is the best choice. With this fine-tuning, Gemma focuses on my repo, allowing me to use natural language to generate API endpoints automatically, complete with services, DAOs, and prompts! #MachineLearning #AI #DeepLearning #QLoRA #FineTuning #Kaggle

It's that time again ⏰ Modverse Weekly - Fourth Edition is out now! 💯🚀 Read the latest and greatest on Modular in our Featured Stories 📖, Mojo 🔥 Language changes, Community spotlights 🔦 & more! ⬇️ #modular #modverse #newsletter #mojo #python #pytorch #ai #ml #programming #machinelearning #artificialintelligence #startup #tech

Modverse Weekly - Issue 2 🚀🔥 is out now! Some incredible community highlights this week, and more exciting improvements to our infrastructure. Read it and signup below! ⬇️ 💯 #mojo #modular #modverse #python #technology #innovation #startups #programming #engineering #ai #ml #machinelearning #artificialintelligence