Docling

?? Multimodal RAG with Docling and IBM's Granite 3.2! ?? We're thrilled to announce a comprehensive tutorial that guides you through building an AI-powered multimodal Retrieval-Augmented Generation (RAG) system using IBM's latest Granite 3.2 model and Docling. Why This Matters: The Granite 3.2 model introduces enhanced reasoning capabilities, enabling more sophisticated understanding and generation of human-like text. When combined with Docling's robust document parsing and conversion features, you can create AI systems that seamlessly process and comprehend diverse data types, including text and images, leading to more accurate and insightful responses. Key Highlights: 1. Granite 3.2 Model: Experience the advanced reasoning and multimodal processing capabilities of IBM's latest language model. 2. Docling Integration: Efficiently handle and transform documents from various sources using this open-source toolkit. 3. LangChain for Workflow Orchestration: Streamline and automate document processing and retrieval workflows, ensuring seamless interaction between different system components. This tutorial is ideal for AI developers, researchers, and enthusiasts aiming to deepen their understanding of document management and advanced natural language processing techniques. Embark on this journey to harness the power of multimodal AI and elevate your projects to new heights! - tutorial: https://lnkd.in/eq2YMYf2 - github: https://lnkd.in/e5MMBdQb #AI #MultimodalAI #RAG #IBMGranite #Docling #LangChain #MachineLearning #DataScience #ArtificialIntelligence #IBM #IBMResearch

Six months ago IBM gave all of us a delightful gift. It’s called Docling, and it converts PDFs to text. You know this, I know this: a knowledge goldmine is tucked away in company PDFs. But getting information out of them (even just copy-pasting!) is a pain. Because the PDF format is this weird middle ground between a document and picture…looks lovely, but doesn't play nicely. Recently, there’s been buzz on how AI from Google or OpenAI will let you upload a document, have the model pull the information and answer questions about its contents. But there are problems right away: ?? Upload company documents just like that? ?? What if you need the text to save for later? ?? With lots of documents, what, you feed them one by one? So these huge, powerful models aren’t really a solution at all here. This is a key point: your big productivity gains from AI are not necessarily going to come out of a chat box... Meanwhile, in August 2024 – to basically ZERO fanfare – IBM just releases a free tool, compact enough to work on a regular work laptop. It can crunch a PDF and give you back the text. FinText ran Docling on tens of thousands of PDFs: reports, papers, commentaries. It runs well, and it does a good job. To learn more, I’ve written up a case study about it, right here ?? https://lnkd.in/ehpGTBzx

?? Just Discovered an AI-Powered Document Whisperer – Meet Docling! Brought to you by IBM,?Ever wished you had a magical AI assistant that could read, understand, and neatly organize all your documents without breaking a sweat? Well, Docling does just that! ?? Think of it as ChatGPT for all your files – whether it’s PDFs, Word docs, spreadsheets, web pages, or even scanned images. It parses, structures, and transforms messy documents into clean, organized formats. ?? Why Docling is a Game-Changer? ? Multi-format Parsing – Handles PDFs, DOCX, XLSX, HTML, images, and more ? Smart PDF Understanding – Reads layouts, tables, formulas, code, and images like a pro ? Unified Document Representation – Converts everything into a structured, AI-friendly format ? Secure & Local Execution – No cloud? No problem. Keep your data private ? AI Integrations Ready – Works seamlessly with LangChain, LlamaIndex, Crew AI, and Haystack ? OCR Superpowers – Extracts text from scanned documents effortlessly ? Simple CLI for Developers – Get started in minutes ?? And the best part? It's open-source! Big things are on the way—metadata extraction, chart interpretation, chemistry models, and more. ???? If you're into AI, research, or automation, Docling is a must-try! ?? Check it out on GitHub: https://lnkd.in/e9ksQXJS ?? What’s your favorite AI tool for document handling? Drop your thoughts below! ?? ?? Share ?? React ?? Save & follow for everything related to AI ! Be curious, stay hungry. Ask it all, learn it all, and pass it on! ????? #AI #OpenSource #Docling #Productivity #Automation #LLMs #DocumentProcessing ??

This is a classic mistake when processing PDF documents: it is a perfect showcase of the capabilities of Docling. ?? See in the comments the Docling output for original 1959 paper. ? Don't know yet about Docling? Docling is an open-source framework that makes document parsing effortless. It supports multiple formats—PDF, DOCX, XLSX, HTML, images, and more—offering deep document understanding and seamless integration with the Gen AI ecosystem. Key Features: ? Multi-format parsing: PDFs, Word, Excel, HTML, images, and more ? Advanced PDF understanding: Page layout, reading order, tables, formulas, and even code! ? Unified DoclingDocument format: An expressive, structured way to work with document data ? Multiple export options: Markdown, HTML, and lossless JSON ? Local execution support: Keep sensitive data secure, even in air-gapped environments ? Plug-and-play integrations: LangChain, LlamaIndex, Crew AI, and more for AI workflows ? OCR support: Extract text from scanned PDFs and images ? Simple CLI: Process documents with ease Give it a try at https://lnkd.in/dZCfQx92! #Docling #AI #OpenSource #DocumentProcessing

This is a classic mistake when processing PDF documents: it is a perfect showcase of the capabilities of Docling. ?? See in the comments the Docling output for original 1959 paper. ? Don't know yet about Docling? Docling is an open-source framework that makes document parsing effortless. It supports multiple formats—PDF, DOCX, XLSX, HTML, images, and more—offering deep document understanding and seamless integration with the Gen AI ecosystem. Key Features: ? Multi-format parsing: PDFs, Word, Excel, HTML, images, and more ? Advanced PDF understanding: Page layout, reading order, tables, formulas, and even code! ? Unified DoclingDocument format: An expressive, structured way to work with document data ? Multiple export options: Markdown, HTML, and lossless JSON ? Local execution support: Keep sensitive data secure, even in air-gapped environments ? Plug-and-play integrations: LangChain, LlamaIndex, Crew AI, and more for AI workflows ? OCR support: Extract text from scanned PDFs and images ? Simple CLI: Process documents with ease Give it a try at https://lnkd.in/dZCfQx92! #Docling #AI #OpenSource #DocumentProcessing

Getting started with Docling? This new tutorial shows you how to "unlock" your documents and build a sample AI application! ? Shoutout to Dave Ebbelaar for putting this together! ?? Video: https://lnkd.in/eBrEDaFa

The first open-source multimodal model built using datasets created with Docling!

?? Nice?talk by Sujee Maniyam at The AI Alliance Office hours! Sujee gave a great introduction of Docling showcasing many key features as well as its integration in the #DataPrepKit. ?? Well done and kudos for the great work including live Q&A session. ?? Watch on Youtube https://lnkd.in/gkxF_fcw ?? Check it out on GitHub https://lnkd.in/dZCfQx92 ?? Try the DPK integration https://lnkd.in/d9cGHwrG #AI #OpenSource #Docling #DPK #DocumentsAI #GenAI #IBM #IBMResearch

Agentic AI with docling!

Nice application of docling in Azure!