The Knowledge Graph Conference’s Post

📚 Dive deeper into the world of machine learning with Snorkel AI's ML Whiteboard series. In this episode, our esteemed guest, Manan Shah, and the Snorkel team discuss two influential publications - 'Glean: Structured Extractions from Templatic Documents' and 'Representation Learning for Information Extraction from Form-like Documents'. This episode is packed with insights on model architecture, experimental results, design considerations, and more. The Q&A sessions provide a great opportunity to clarify your doubts and learn from the experts. Check out the full episode here: https://buff.ly/3VM4vla. #snorkelai #machinelearning #datascience

✅ Learn how to use Lighthouz to generate custom benchmarks and evaluate your LlamaIndex LLMs!

How would LLM, like Llama 3 optimized for finance, be integrated with Unstructured , VectorDBs, and Symbolic AI to transform your data analysis workflows? 🔷 Information Extraction: Build entire databases (normal and time series) of accounting data, making data access more affordable and convenient than paying an expensive data provider (#WorkingOnIt #OwnYourData) 🔷 Financial Literacy: Explain financial reports at all levels, from novices to experts, increasing accessibility. (#WorkingOnIt #UnderstandYourData) 🔷 Advanced Reporting: Implement functions for advanced financial analysis, with the LLM and RAG pipelines handling data retrieval and interpretation. (#WorkingOnIt #UseYourData) And more! Open-source LLMs and tooling are making the financial landscape more accessible to everybody 🌍 Excited to see how these tools change the financial landscape, making everything from data extraction to advanced analysis more accessible with just a ticker symbol and a button push! Unstructured: https://unstructured.io/ SymbolicAI: https://lnkd.in/dkXB-d9E And look 👀 people have already started using LLMs in building their own finance-specialized workflows 🚀 #FinancialInnovation #DataScience #MachineLearning #AdvancedAnalytics #SymbolicAI #WorkingOnMyOwnFinanceModel

The inimitable Wenqi Glantz puts our LlamaPacks of 7 pre-baked advanced query strategies through their paces with @lighthouzai for evaluation: ➡️ See how to get a jump-start on advanced querying by bringing in a LlamaPack ➡️ Evaluate the best available query strategy for your dataset https://lnkd.in/dpucB5vb

Who loves a good and up-to-date documentation? Everyone! ✨ Who loves to write and keep the documentation up to date? No one! 😅 At RazorSharp, we're taking a step towards solving that issue! Our upcoming feature lets you store system documentation in markdown, embed topology maps, and leverage AI to analyze logs and draft summaries. Once the basics are in place, we will research methods to detect outdated and contradictory documentation, ensuring your docs are always reliable. Stay tuned for our next release and see how we're making documentation effortless. 🚀 Check out the preview video to learn more.

Do you love entity resolution? Are you interested in the technical details behind performing entity resolution practically, at scale? Or maybe you'd like to hear me talk about math for a couple of hours? Today marks my 11th year with Tamr, and as part of this milestone we've released a five-part video series delving into the challenges of doing entity resolution in real-world contexts, at scale, and the solutions we've developed across years of collaboration with our customers. Here's a blog post summarizing the series:

Want to refresh your RAG skills? Weaviate put together a comprehensive guide to improving indexing, retrieval, and generation in RAG systems, using a medical chatbot as a practical example. TL;DR: 🧠 Indexing: Explore simple, semantic, and language model-based chunking techniques 📊 Metadata: Leverage metadata for improved context and use GLiNER for automatic generation 🔍 Retrieval: Implement hybrid search combining vector and keyword methods ✍️ Query Optimization: Use query rewriting for better database and LM interactions 🎯 Embedding: Fine-tune embedding models on domain-specific data ✂️ Autocut: Remove irrelevant information from retrieved content 🔝 Reranking: Reorder retrieved objects for better relevance 🚀 LLM Fine-tuning: Enhance LLMs with domain-specific data for improved performance 👉 https://lnkd.in/etjXqdAw

Another really useful guide on improving RAG accuracy.

Improving RAG systems is crucial for advancing AI-driven solutions. Weaviate’s guide offers invaluable insights, especially for those in the healthcare tech space. Here’s what caught my attention… #AI #MachineLearning #RetrievalAugmentedGeneration #HealthTech #Innovation”

Slides for my 10th lecture of my course "Mathematics and AI" are now online. In this class we covered regularization and dimension reduction methods. The first half was on ridge regression, lasso (nobody asked me why it's called lasso!). In the second half, we discussed feature engineering via PCA/PCR https://lnkd.in/eq533GvE