Ashwin Pingali

Common Agentic workflow patterns #llm #llms #agenticworkflow

1

AI next frontier is possibly developing and scaling the reasoning layer, where AI systems possess hard problem-solving and higher order cognitive intelligence. Came across this good article from Sequoia.

20

Very good paper and an excellent post from Aishwarya Naresh Reganti . Will follow up with a review of the paper and how to put some of the learning from the paper into practice.

1

Sharing my latest article on Overconfidence in State of the Art LLMs. Explore the challenges and limitations these models face in critical applications. #AI #MachineLearning #LLMs #ArtificialIntelligence #TechInsights #Innovation #EnterpriseSolutions #AIResearch

23

3 Comments

I've taken a brief detour from our ABCs of AI series to bring you a fresh perspective with "𝐓𝐞𝐜𝐡𝐋𝐢𝐧𝐠𝐨 𝐔𝐧𝐩𝐚𝐜𝐤𝐞𝐝" series. After numerous requests, I finally buckled and dove into the epic showdown of Data Mesh vs. Data Fabric! 🥊 In this data management bout, we explore these titans' strengths, challenges, and potential for teamwork. Don't worry; we'll return to our regularly scheduled ABCs of AI programming soon. Curious about other tech terms? Drop the lingo you'd like demystified in the comments! #DataManagement #TechLingo #DataMesh #DataFabric

12

6 Comments

Jeremy Howard proposes a new standard: a /llms.txt file for websites to provide information optimized for Large Language Models (LLMs). This Markdown file, located at the root of a website, offers a concise overview and links to more detailed Markdown files, making it easier for LLMs to access key information. The proposal aims to improve how LLMs interact with websites, addressing limitations of context windows and the challenges of converting complex HTML. The format includes key sections, a summary, and links to further documentation, allowing both human and LLM readability. The /llms.txt standard is intended to work alongside existing web standards like robots.txt and sitemap.xml. https://lnkd.in/gkRGVM2W

25

3 Comments

Is your team building AI/ML and struggling to get accurate models into production? Have you ever wondered "Are there better tools out there than what i'm using now? Is there a simpler & faster way to do this?" Imagine building chatbots, AI agents, ML models in 1/10 of the time and effort. Dozens of Fortune 500 companies have, and are glad they found Abacus: The world's first AI super-assistant where AI, not humans, build enterprise-scale applied AI systems and agents. If you're not 100% happy with how you're building AI/ML today, Abacus could be for you too. Ping me if you'd like to have a friendly introduction to learn more! hashtag #AI #ArtificialIntelligence #MachineLearning #MLOps #DataScience #Technology #Innovation #AbacusAI #AIAgents

5

🚀 Unlocking the Power of Reciprocal Rank Fusion (RRF) 🚀 In the world of AI-driven search and recommendation systems, accuracy is critical—and that's where Reciprocal Rank Fusion (RRF) shines. Rather than relying on a single algorithm to produce results, RRF leverages multiple ranking systems, merging their outputs to deliver more precise and relevant outcomes. 📊 🔍 Why is this pivotal? In industries that demand high precision, like recommendation engines and information retrieval, RRF optimizes the decision-making process by amplifying the results that rank consistently across different systems. This blend of rankings is particularly useful for improving AI performance, ensuring that top results are not only accurate but also contextually relevant. 💡 How does RRF drive value? RRF assigns higher significance to results ranked consistently well by multiple models, balancing between quick wins and strategic relevance. By synthesizing data from various ranking systems, RRF enhances the reliability of recommendation algorithms, delivering outcomes that help businesses make smarter, data-driven decisions faster. In a nutshell: Enhanced rankings lead to better insights, and better insights drive business success! 💼 #AI #MachineLearning #ReciprocalRankFusion #SearchOptimization #DataScience #RankingAlgorithms #TechLeadership #Innovation

9

Enter Small Language Models or SLMs. These models, typically with up to 20 billion parameters, provide a focused alternative to larger models like GPT-4. They cater to specific business needs such as chat, text analytics, and content generation, balancing capability and control effectively. The decreasing cost of GPT 4.0 mini and platforms like Claude Haiku has made these models more accessible, quicker, and cost-effective. SLMs excel at prompt-based tasks, like generating a 2500-page book for just $0.60. Advantages of SLMs include agile development, reduced risk of hallucinations, lightweight design for edge devices, controlled risks such as bias and toxicity, interpretability for developers, improved latency, and enhanced sustainability by aligning with GreenAI principles. SLMs will help in taking these themes forward:- - Domain specific AI - Distributed AI - Lesser Data Quality/ Privacy issues when compared to Foundation Models - Better AI adoption in enterprises #Technology #AI #GreenAI #SLMs #GPT4 #Innovation #Collaberadigital

50

So what exactly do we do at Ringer Sciences? In short, we help companies build brand reputation intelligently. We do this by tapping the best human analysts and data scientists paired with strategic AI-enabled tools, technologies and techniques. We help uncover actionable intelligence in service of business results and architect next-generation digital experiences with data and insight serving as the foundation. Have a look at our breakdown of services and let's find a way to work together! CC Bob Pearson David Gallagher Ritesh Patel Jeremiah Owyang Yash Gad Tim Hayden

12

1 Comment

As digital and AI transformations accelerate, the potential of hashtag #DataSharing is becoming more evident. Pioneers like Walmart and Procter & Gamble have, over the years, reaped significant benefits from embracing data sharing, yet many organizations remain hesitant. Thanks to new technology, data sharing has become more secure and efficient, easing operational hurdles and regulatory compliance. These advances also redefine the role of trust. In my most recent Fortune column, co-authored with Riccarda Joas, Guillaume Sajust de Bergues, and François Candelon, we explore how these technological advances are impacting the risk-reward calculus. We provide arguments that might encourage executives to reconsider their stance on data sharing. Read more on Fortune: https://lnkd.in/eWP2K6KF

27

1 Comment

WorFBench: A Benchmark for Evaluating Complex Workflow Generation in Large Language Model Agents Large Language Models (LLMs) have shown remarkable potential in solving complex real-world problems, from function calls to embodied planning and code generation. A critical capability for LLM agents is decomposing complex problems into executable subtasks through workflows, which serve as intermediate states to improve debugging and interpretability. While workflows provide prior knowledge to prevent hallucinations, current evaluation benchmarks for workflow generation face significant challenges. Read the full article: https://lnkd.in/eqv7pQCt Paper: https://lnkd.in/eWDxh6Vz

5

What are the key differences between leveraging an LLM-based Agent vs a traditional Pipeline? How do we combine both to create "AI Pipelines?" Great hands-on tutorial leveraging open-source frameworks to create AI Pipelines https://lnkd.in/geAax6NY Thanks Kevin Coyle for the recommendation on CrewAI #genai #agents #pipelines #ai #ml

24

🐬 Dremio users deal with a lot of different data file and table formats. ⚒️ So Dipankar Mazumdar, M.Sc 🥑 of Onehouse and Alex Merced of Dremio worked together to create a new blog post. 🕵️ In the post, they show how you might use Dremio to smoothly do lakehouse analytics on Apache Hudi and Apache Iceberg data using Apache XTable (Incubating). 0️⃣1️⃣ Check it out - and borrow and revise the code as needed. https://lnkd.in/g-jBvJ55 #apachehudi #XTable #datalakehouse #dataengineering #onehouse #opensource #nolockin

28

3 Comments

🚀 Excited to share how Large Language Models (LLMs) are revolutionizing data integration! Discover how they map tabular data to ontologies, making data management smarter & more efficient🔥💡 Read our new blog post: https://shorturl.at/BwkN2 #AI #LLMs #DataScience #Ontologies

9

2 Comments

The discourse around Retrieval Augmented Generation, today, is as vibrant as it was around LLMs and Generative AI back in December 2022. With Graph RAG and Agentic RAG garnering a lot of eyeballs, it is imperative that Gen AI applications will have to use RAG in some shape and form. I have been working on different implementations of RAG at Yarnit to bring contextual intelligence to content marketing. Based on these learnings, I am writing A Simple Guide to Retrieval Augmented Generation with Manning Publications Co. The first three chapters of the book have been released as a part of the Manning Early Access program and I am quite pleased and humbled that the MEAP is amongst the bestsellers at Manning for the third straight week. As a token of appreciation for early supporters we are running a 40% discount on the ebook. You can go to https://mng.bz/jXJ9 and use the promo code watchkimothi40 at the checkout. This book is for technology professionals who want to get introduced to the concept of Retrieval Augmented Generation and build LLM based apps. It will prove to be a handy book for beginners as well as experienced professionals. You'll also get an opportunity to code along in python but the book is intended for non-coders as well. From introducing the technique to building RAG pipelines in production, in 9 chapters the book covers the following - Chapter 1 - Large Language Models and the Need for Retrieval Augmented Generation [First draft Released] Chapter 2 - RAG-enabled systems and their design [First draft Released] Chapter 3 - Indexing Pipeline : Creating a knowledge base for RAG based applications [First draft Released] Chapter 4 - Generation Pipeline: Real time interaction for contextual responses Chapter 5 - RAG Evaluation : Checking accuracy, relevance and faithfulness Chapter 6 - Progression of RAG systems : Naive to Advanced to Modular Chapter 7 - Evolving RAGOps Stack : Technologies that make RAG possible Chapter 8 - Nuances : Comparison with fine-tuning, multimodal and agentic RAG Chapter 9 - Cutting Edge : Best practices and further exploration Why join MEAP? By joining the Manning Early Access Program, you'll get: - Immediate access to the book's current draft and all future updates - A chance to provide feedback and shape the final content - Exclusive discounts and early-bird offers Thank you for your continued support. I hope you find this book as exciting and informative as I have found writing it. Do reach out to me if you have any questions or comments around the content.

96

9 Comments