SonyAI’s Post

Introducing Pi AI Weekly Trends 12! Our weekly feature brings you the latest AI tech to keep you ahead. This week, we highlight: 🔹 AdalFlow: an LLM application library for building optimised LLM task pipelines. The library's design is quite similar to PyTorch. 🔗 https://pischool.link/0xy 🔹 Mistral 2- 123B: the newest Mitral Model that gives tough competition to the largest LLama 3.1 405B model on several baselines. This model is released under the Mistral Research License, allowing usage/modification for non-commercial or research purposes. 🔗 https://pischool.link/3rj 🔹 MINT-1: A 1-trillion token Multimodal dataset collected and curated by Salesforce AI Research 🔗 https://pischool.link/qp5 Our Machine Learning Scientist, Marcello Politi, selected these links for you. Was this helpful? Let us know by liking and sharing! #artificialintelligence #largelanguagemodels #MachineLearning #DeepLearning #AI #LLMs #PiAIWeeklyTrends

Has the hype and recent surge in popularity of #GenerativeAI overshadowed federated learning? Deep dive with one of our AI experts, Holly Toner, and learn about how federated learning could still make a comeback. Learn about the pros and cons, use cases and much more in the article below 👇 https://kainos.pub/3QB7dqj #FederatedLearning #AIExperts #TechTrends #MachineLearning #ArtificialIntelligence #TechInnovation #AICommunity #FutureTech

Absolutely thrilled to share that I have earned the Applied AI Bronze Badge! The course was incredibly valuable and practical, elevating my understanding of neural networks and their application in business. Big shoutout to my fantastic colleagues for their support and for putting up with my constant AI chatter. Now, onwards and upwards with our ServiceMindAI project, equipping our customers with an AI-powered knowledge automation toolset. 🚀 #AI #Innovation #CustomerExperience #ServiceAutomation #ContinuousLearning

🚀 Boosting Machine Learning Performance with Data Enrichment 🌍 Incorporating enriched data can significantly enhance the accuracy of machine learning models. Check out this quick demo on how to extract useful features: 1️⃣ Well Formation Tops to Logs: Well tops and zonation, when used as features, can dramatically improve the model's performance. 2️⃣ Location to Distance Logs: Incorporating location-based data points helps the model understand spatial dependencies, leading to more accurate results. 3️⃣ Log Encoding: Leverage discrete logs to create new encoded features for training your model more effectively. Enrich your data, elevate your models! 🔍💡 #MachineLearning #Geoscience #GeoXMate #DataScience #FeatureEngineering #AI #Petroleum

Hey there !! As im interested in the field of AI , I have successfully completed the "Introduction to Artificial Intelligence" course offered by Great Learning Academy in the month of November 2023. #GreatLearningAcademy #greatlearning #glacertificate

Who thought Next-Gen AI would be Gen-AI? In the post-deep learning era, there has been a significant shift in the landscape of machine learning systems. According to the epochai dataset, available at https://lnkd.in/dCzaPnDu, this year witnessed the addition of 38 new systems, with language models dominating the leaderboard by a staggering six times more than the next best multi-modal systems. An astonishing number worth pondering: 8.00000000001e 25, representing the training compute (FLOPS) utilized by Gemini Ultra, securing the top spot in training compute usage across time. Below, I've outlined the best-fit slopes after the deep learning era, illustrating the growth trends in notable AI systems across various domains, indicating where significant investments are being made: Language: 7.3x Vision: 3.7x Games: 3.5x Multimodal: 2.7x Image Generation: 2.5x What are your thoughts now on the future of AI, would it be #AGI? Feel free to share your comments below! For more intriguing insights, delve into Stanford University's comprehensive report, available at https://lnkd.in/d6V3iR8j. And don't forget to hit that follow button for more updates 😉 #genai #AI #MachineLearning #DeepLearning #GenAI #EpochAI #StanfordUniversity #Research

Sept. 12, join Abhinav Kimothi, Co-founder/VP of AI Yarnit and author of the upcoming Manning Publications Co. book, "A Simple Guide to Retrieval Augmented Generation," for the #ACMTechTalk "Introduction to Retrieval Augmented Generation: What Is It and Why Should You Pay Attention." Marlene Mhangami, Developer Advocate @ Microsoft & Vice-Chair of the ACM Practitioner Board, will moderate the talk. This introductory talk is aimed at explaining why RAG is becoming a critical component of the AI toolkit. If you want to build AI systems or are simply interested in the latest trends in generative AI, this session is designed to provide an introduction, insights and practical knowledge on RAG. Register free to attend: https://bit.ly/3ZaqN1F #RAG #GAI #RetrievalAugmentedGeneration #LLMs

We’ve made it easy to visualize dozens of computer vision datasets locally in the open source FiftyOne toolset with just a few simple commands. https://lnkd.in/gUBk3FGj Datasets include ActivityNet, BDD100K, Caltech, CIFAR, Cityscapes, COCO, Families in the Wild, Kinetics, KITTI, MNIST, Open Images, UCF101 and many more! #computervision #ai #artificialintelligence #machinevision #machinelearning #datascience

We’ve made it easy to visualize dozens of computer vision datasets locally in the open source FiftyOne toolset with just a few simple commands. https://lnkd.in/gUBk3FGj Datasets include ActivityNet, BDD100K, Caltech, CIFAR, Cityscapes, COCO, Families in the Wild, Kinetics, KITTI, MNIST, Open Images, UCF101 and many more! #computervision #ai #artificialintelligence #machinevision #machinelearning #datascience

💥 Now at #ECCV2024 Lorenzo Baraldi and Federico Cocchi are presenting our paper "Contrasting Deepfakes Diffusion via Contrastive Learning and Global-Local Similarities". 🔍 In this paper, we propose a new embedding space specifically designed for deepfake detection. Our embedding space is trained via contrastive learning by enforcing global-local similarities, on a novel dataset of 9.2 million images produced by using four different generators. ✨ Step by at our poster #3 for more details and discussion! Read the full paper at: https://lnkd.in/dcs_xmEt. Check out the code and dataset on GitHub: https://lnkd.in/dmCi725A. This is a joint work with Leonardo AI Labs, within the ELSA (European Lighthouse on Secure and Safe AI) European Project. Alessandro Nicolosi Marcella Cornia Rita Cucchiara #DeepFakes #TrustworthyAI #ComputerVision #AI