Ganapathy Shankar

𝗪𝗶𝗹𝗹 𝗔𝗜 𝗛𝗶𝘁 𝗮 𝗗𝗮𝘁𝗮 𝗪𝗮𝗹𝗹 𝗶𝗻 𝟮𝟬𝟯𝟬? 🆘 Very interesting analysis from epoch.ai about data corpus available for training large language models and whether we will hit the ceiling soon. https://lnkd.in/gD78f9DW The authors estimate that all available data will be used by ~2030, but they're also the first to admit that by then we will have ways to address "the shortage" (such as synthetic data, data efficiency improvements, use of other modalities). What I found more interesting though is why the latest predictions differ from their 2022 paper, where they speculated that all the data will be used by 2024: ☑ Web data proves more valuable than expected (on par with scientific papers) ☑ Models can "reuse" datasets multiple times 2030 is right around the corner, so we will see a lot of innovation on this front and creative ideas on how to 'recycle' and make best use of the data we already have. These estimates also don't take into account proprietary data owned by individual companies. In this No priors episode (https://lnkd.in/g2BkTJBz) there is a mention of JPMorgan owning 150PB of proprietary data, which is at least 150x of what was used to train GPT-4. JPMorgan is monetizing this data today (https://lnkd.in/gXjE92qx), but it gets exponentially more valuable in a world where data is a finite (and extremely sought after) resource.

12

Large Language Models for Knowledge Graph Completion Class membership relations are critical components of knowledge graphs (KGs) that assign entities to given classes, and help KGs represent classification schemes which have significant effects on social policy and scientific consensus. Evaluation of these relations is necessary, but expensive as it is manually performed. This article evaluates the abilities of several large language models (LLMs) to evaluate class membership relations in a zero-shot manner.   Article link: https://lnkd.in/gxH4pttf 🔹To evaluate KG relations, knowledge engineers must translate natural language from domain experts into symbolic representations, then check if the knowledge is properly reflected in a KG. LLMs can use natural language directly when given descriptions of an entity and a KG class, to predict if the entity is an instance of the class. 🔹The authors determined LLMs’ abilities to do this effectively by studying if LLMs could detect missing or incorrect relations. They constructed evaluation datasets from the Wikidata and CaLiGraph KGs and evaluated seven LLMs, including OpenAI’s GPT-4 and Meta’s Llama-2. 🔹It was found that LLMs and KGs were well-aligned, meaning KGs could effectively address knowledge gaps in LLMs and LLMs could similarly do so for KGs. The LLMs that were evaluated also displayed abilities to detect missing or incorrect KG relations. Moreover, when analysing the LLMs’ classification errors, it was found that 40.9% of the errors were attributed to the KGs themselves, which had missing or incorrect relations. A further 29.1% of errors were due to insufficient data in entity descriptions. 🔹LLMs have demonstrated good performance across many tasks, and now display their capabilities to assist knowledge engineers in KG evaluation and refinement as well. However, they still face some challenges like slow speed of processing tasks and costly API calls, which future efforts could target by using locally deployed LLMs and sampling approaches. Add additional points if needed, for instance, about limitations or problems. 📑 Allen, B. P., & Groth, P. T. (2024). Evaluating Class Membership Relations in Knowledge Graphs using Large Language Models (Version 1). arXiv. DOI: 10.48550/ARXIV.2404.17000 📌 Ensuring the completeness and correctness of KGs has been a limiting factor in the widespread employment of KGs across applications. LLMs have previously been used to optimise KG construction, and now, they show their abilities to perform KG refinement as well, thus easing the burden of knowledge engineers.

2

📝𝐀𝐖𝐒 𝐫𝐞:𝐈𝐧𝐯𝐞𝐧𝐭 𝟐𝟎𝟐𝟒 𝐇𝐢𝐠𝐡𝐥𝐢𝐠𝐡𝐭𝐬: 𝐒𝐰𝐚𝐦𝐢'𝐬 𝐍𝐨𝐭𝐞 Have you ever wondered what the future of AI and data looks like? 🤔 𝐃𝐫. 𝐒𝐰𝐚𝐦𝐢 𝐒𝐢𝐯𝐚𝐬𝐮𝐛𝐫𝐚𝐦𝐚𝐧𝐢𝐚𝐧’𝐬 𝐤𝐞𝐲𝐧𝐨𝐭𝐞 𝐚𝐭 𝐀𝐖𝐒 𝐫𝐞:𝐈𝐧𝐯𝐞𝐧𝐭 𝐝𝐞𝐥𝐢𝐯𝐞𝐫𝐬 𝐚 𝐜𝐫𝐲𝐬𝐭𝐚𝐥-𝐜𝐥𝐞𝐚𝐫 𝐯𝐢𝐬𝐢𝐨𝐧! Packed with insights and actionable takeaways, it’s a must-watch for anyone in the tech and data space. But don’t worry if you missed it—here’s everything you need to know in just a few minutes! ⏱️ 𝟭. 𝐀𝐈 𝐚𝐧𝐝 𝐌𝐋: 𝐁𝐫𝐞𝐚𝐤𝐢𝐧𝐠 𝐁𝐚𝐫𝐫𝐢𝐞𝐫𝐬 AI adoption is now easier than ever! 🔹 AWS democratizes machine learning with tools that are user-friendly and scalable. 🔹 Businesses of all sizes can now leverage AI to streamline operations and make smarter decisions. 𝟮. 𝐓𝐡𝐞 𝐏𝐨𝐰𝐞𝐫 𝐨𝐟 𝐃𝐚𝐭𝐚 Data is the fuel for innovation, and AWS is unlocking its potential: ✔️ Real-time processing capabilities and seamless integration with advanced analytics. ✔️ Key innovations include enhanced data lakes, storage, and analytics tools. 𝟯. 𝐒𝐜𝐚𝐥𝐚𝐛𝐢𝐥𝐢𝐭𝐲 𝐰𝐢𝐭𝐡𝐨𝐮𝐭 𝐂𝐨𝐦𝐩𝐥𝐞𝐱𝐢𝐭𝐲 Scaling machine learning is no longer daunting. AWS’s latest MLOps tools simplify workflows, enabling faster deployments and results. 💡 𝐖𝐡𝐲 𝐢𝐭 𝐦𝐚𝐭𝐭𝐞𝐫𝐬: The future of AI and data is here. AWS is equipping businesses with the tools to innovate, scale, and thrive. 📖 𝐑𝐞𝐚𝐝 𝐦𝐨𝐫𝐞: Check out the full article https://lnkd.in/gqNyMzDn 💬 𝐋𝐞𝐭’𝐬 𝐝𝐢𝐬𝐜𝐮𝐬𝐬: How are you using AI, ML, and data in your organization? Share your thoughts! #𝐀𝐖𝐒 #𝐀𝐈 #𝐌𝐚𝐜𝐡𝐢𝐧𝐞𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠 #𝐃𝐚𝐭𝐚𝐒𝐜𝐢𝐞𝐧𝐜𝐞 #𝐓𝐞𝐜𝐡𝐈𝐧𝐧𝐨𝐯𝐚𝐭𝐢𝐨𝐧 #𝐫𝐞𝐈𝐧𝐯𝐞𝐧𝐭

4

Reason we need a solid framework that can validate LLM response before it is presented to end customer Till such time GenAI in enterprise will be mostly focussed on human in loop use cases with only some of it direct customer facing #generativeai #artificialintelligence

51

24 Comments

If you are still thinking that AI is just a hype, you should listen the conversation between Mukesh Ambani and NVIDIA CEO Jensen Huang. Why ? Mukesh Ambani bullishly talks about Generative AI models, GPUs, need of open source models etc. He takes up the challenge to surprise world with AI capabilities in coming years. He is not someone who has been leading IT/Software companies day in day out like Narayan Murthy/ Shiv Nadar/Bill Gates etc. What it means? AI has already arrived and every big firm, irrespective of their traditional business, wants to build business and create impact using this. AI is not just another cool tech anymore. It is the industrial revolution changing every facet of our life. Embrace it or get ready for classic "Kodak Case study" in your career and business. Watch here: https://lnkd.in/ejvnCMHZ

10

💡A super interesting blogpost by Siddharth Jindal on Analytics India Magazine, talking about the uniqueness and potential advantages of Gemma over the LLaMA models when it comes fine-tuning LLMs for Indic languages. Some highlights: 📌This is specifically due to the sentencepiece tokenizer used by Gemma which is able to encode strong representation for scripts like Devanagari. 📌The vocab size of Gemma is 256K whereas the LLaMA-3 has only 128K, this seems to indicate that the coverage is much larger for Gemma! 📌The Gemma model is able to generalize to other languages without the explicit need for continuous pre-training as opposed to LLaMA models where the tokenizer needs to be extended to accomodate the language/s in question. 📌Of course, this dosen't mean that Gemma is the silver bullet, since it does face some stability issues in training. This article definitely was an eye-opener for me and also exciting in terms of working with Gemma for fine-tuning Indic LLMs. Check it out here: https://lnkd.in/dRCEtF9Q Also, I'd love to hear more thoughts from my audience here, about their experiences with Gemma and multilingual LLMs! #artificialintelligence #airesearch #largelanguagemodels

9

😎 Big challenge alert 🚀: How to disrupt the GPUs market! Researchers and tech enthusiasts are exploring innovative ways to simplify Attention and steer clear of matrix multiplication complexities. From classical hashing to ternary matrices and now clipping, strategies are evolving to transform multiplication into addition with impressive results. If you're passionate about deep tech or considering a startup, this is a prime problem to tackle. Dive into the details here: 🌟 #TechInnovation #GPUs #ResearchOpportunity

40

2 Comments

𝐓𝐡𝐞 𝐏𝐨𝐰𝐞𝐫 𝐂𝐨𝐦𝐛𝐨 𝐨𝐟 𝐀𝐈 𝐀𝐠𝐞𝐧𝐭𝐬 𝐚𝐧𝐝 𝐭𝐡𝐞 𝐌𝐨𝐝𝐮𝐥𝐚𝐫 𝐃𝐚𝐭𝐚 𝐒𝐭𝐚𝐜𝐤: 𝐀𝐈 𝐭𝐡𝐚𝐭 𝐑𝐞𝐚𝐬𝐨𝐧𝐬 In this latest edition, Animesh Kumar, Brij Mohan Singh, and Ritwika C. explore the evolution of AI from LLMs (Large Language Models) to more sophisticated AI agents capable of reasoning, autonomy, and adaptive problem-solving. 𝐊𝐞𝐲 𝐓𝐚𝐤𝐞𝐚𝐰𝐚𝐲𝐬: 💠 𝐀𝐈 𝐀𝐠𝐞𝐧𝐭𝐬: 𝐓𝐫𝐚𝐧𝐬𝐟𝐨𝐫𝐦𝐢𝐧𝐠 𝐭𝐡𝐞 𝐑𝐨𝐥𝐞 𝐨𝐟 𝐀𝐈 𝐢𝐧 𝐄𝐧𝐭𝐞𝐫𝐩𝐫𝐢𝐬𝐞𝐬 AI Agents are the next frontier in enterprise AI, transforming from simple large language models (LLMs) to sophisticated, reason-driven systems. These agents don't just provide answers but deliver purpose-driven actions that align with business goals, making AI a more valuable strategic partner in enterprise innovation. 💠 𝐖𝐨𝐫𝐤𝐟𝐥𝐨𝐰 𝐨𝐟 𝐀𝐈 𝐀𝐠𝐞𝐧𝐭𝐬: 𝐄𝐟𝐟𝐢𝐜𝐢𝐞𝐧𝐜𝐲 𝐓𝐡𝐫𝐨𝐮𝐠𝐡 𝐌𝐨𝐝𝐮𝐥𝐚𝐫 𝐒𝐨𝐥𝐮𝐭𝐢𝐨𝐧𝐬 AI Agents are built on a modular architecture where each component, such as an LLM, handles a specific task within a larger workflow. This setup makes them exceptionally good at breaking down complex problems into manageable parts. 💠 𝐂𝐨𝐧𝐬𝐢𝐬𝐭𝐞𝐧𝐜𝐲 𝐢𝐧 𝐃𝐞𝐬𝐢𝐠𝐧: 𝐀𝐥𝐢𝐠𝐧𝐢𝐧𝐠 𝐀𝐈 𝐚𝐧𝐝 𝐃𝐚𝐭𝐚 𝐒𝐭𝐚𝐜𝐤𝐬 The power of design in AI agents comes from a consistent, modular approach that spans both AI and data stacks. Just as Apple's minimalist design philosophy permeates every layer of its ecosystem, aligning the data and AI layers creates a cohesive structure that supports transparency, efficiency, and a high degree of adaptability in AI systems. 💠 𝐋𝐋𝐌 𝐋𝐢𝐦𝐢𝐭𝐚𝐭𝐢𝐨𝐧𝐬: 𝐓𝐡𝐞 𝐍𝐞𝐞𝐝 𝐟𝐨𝐫 𝐀𝐠𝐞𝐧𝐜𝐲 𝐚𝐧𝐝 𝐑𝐞𝐚𝐥-𝐓𝐢𝐦𝐞 𝐃𝐞𝐜𝐢𝐬𝐢𝐨𝐧-𝐌𝐚𝐤𝐢𝐧𝐠 Traditional LLMs often fall short due to limited interactivity with real-world systems, resulting in generic or incomplete solutions. AI agents fill this gap by taking proactive, context-driven actions, allowing real-time decision-making that traditional LLMs cannot achieve, ensuring a more accurate and timely output for critical tasks. 💠 𝐓𝐲𝐩𝐞𝐬 𝐨𝐟 𝐀𝐈 𝐀𝐠𝐞𝐧𝐭𝐬: 𝐅𝐫𝐨𝐦 𝐑𝐞𝐟𝐥𝐞𝐱 𝐭𝐨 𝐔𝐭𝐢𝐥𝐢𝐭𝐲-𝐁𝐚𝐬𝐞𝐝 𝐃𝐞𝐜𝐢𝐬𝐢𝐨𝐧-𝐌𝐚𝐤𝐢𝐧𝐠 AI agents come in various forms, each suited to different types of tasks. Simple reflex agents provide basic reactions, while utility-based and goal-based agents excel in complex environments by prioritizing goals or optimizing outcomes. 𝐑𝐞𝐚𝐝 𝐭𝐡𝐞 𝐞𝐧𝐭𝐢𝐫𝐞 𝐛𝐥𝐨𝐠 𝐡𝐞𝐫𝐞: https://lnkd.in/dw3Htg8a 📢 Do share what you feel about this take! 🗣️ 𝐒𝐡𝐨𝐮𝐭 𝐨𝐮𝐭: At MD101, we actively collaborate with data experts to bring the best resources to a blooming community of data practitioners. If you are willing to share an insightful perspective on anything about data, we’re all ears!! Note: All submissions are vetted for quality & relevance. 🔔 Follow 𝐌𝐨𝐝𝐞𝐫𝐧 𝐃𝐚𝐭𝐚 101 and stay updated with our weekly highlights from the modern data space.

20

1 Comment

Lifting the Veil on AI's decision making, Vinay Kumar Sankarapu from Arya.ai will share about AI transparency that goes beyond the usual "black box" discussions. In today’s world of complex deep learning models, understanding why a model makes certain decisions is crucial—especially in mission-critical applications. He'll explore DL-Backtrace, a new model-agnostic technique designed to bring transparency to any deep learning model, whether it’s an LLM, computer vision, NLP, or tabular data model. What You’ll Learn: - The importance of explainability for mission-critical deep learning applications - How DL-Backtrace compare to existing techniques like SHAP, LIME, and GradCAM for models including LLMs (Llama 3.2), NLP (Bert), and CV (ResNet) - Real-world benchmarking results showing DL-Backtrace’s performance across varied deep learning architectures - Insights into the current limitations and future possibilities in scaling explainability for large models #ResponsibleAI MLOps World Generative AI World

7

How do you get to a quantized model when you start out from a standard model? Arun Nanda looks into the nitty-gritty details of several quantization approaches for neural network models.

6

Looking forward to this panel where Divakar Tantravahi, Meduri Ravi Kumar, and Cosimo Spera will talk about the latest AI technologies that are changing product and customer support, along with their real world examples and practical strategies. If you have not registered, use this link - https://lnkd.in/gdTMgryA #ai #businessoutcomes #product #customer

22

Read the detailed blog on my recent post about Vector DB vs Graph DB for LLM applications. #graphdb #vectordb #chroma #pinecore #llms #chatgpt4 #chatgpt4o #performance

15

🎯 Open Source Data Platforms: From Complexity to Clarity An excellent presentation by Priyanka Naik on building robust data platforms using open source tools. Key insights shared: 📊 Transitioning from batch to real-time streaming 🔄 Modular architecture for better maintainability 🎯 Data catalogs for non-technical user engagement ⚡ Apache Camel connectors for optimized Kafka workflows With 17 years of experience, Priyanka offered valuable perspectives on data democratization and quality management in modern architectures. Watch here: https://lnkd.in/gPcm8sZs #DataEngineering #OpenSource #Kafka #DataPlatform #Engineering

1

"𝐔𝐥𝐭𝐢𝐦𝐚𝐭𝐞 𝐌𝐚𝐜𝐡𝐢𝐧𝐞 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠 𝐰𝐢𝐭𝐡 𝐒𝐜𝐢𝐤𝐢𝐭-𝐋𝐞𝐚𝐫𝐧"🎉 Authored by- Parag Saxena Technically reviewed by - Dr. Alok Tiwari (Ph.D, IIT-BHU) ✳ 𝐁𝐔𝐘 𝐍𝐎𝐖! ✳ 👉 AVA™: [https://lnkd.in/gECvrWh2] 👉 Amazon International: [https://rb.gy/wpyaja] 👉 Amazon India: [https://rb.gy/sfdj1e] #MachineLearning #ScikitLearn #DataScience #Python #AI #ML #DataAnalysis #PredictiveModeling #DataMining #ArtificialIntelligence #Programming #Technology #Coding #DataAnalytics #DataPreprocessing #FeatureEngineering #ModelTraining #ModelEvaluation #Deployment #Algorithm #DataInsights #PatternRecognition #DataVisualization #AVA #Orangeava #Publishers #itandsoftware #Sneakpeek

51

AI Hallucinations & What is Retrieval Augmented Generation (RAG)? The retrieval augmented generation or RAG, it does not rely on information coded during initial training. Instead, it augments a user’s prompts by gathering information from a custom database and then a large language model (LLM) generates an answer based on that data At Blu Parrot Analytics, our commitment to precision in data analytics helps us deliver actionable and trustworthy insights for our clients.We leverage Retrieval-Augmented Generation (RAG) to enhance the accuracy and reliability of our AI models. By integrating RAG, we effectively reduce AI hallucinations, ensuring that our data insights and predictions are grounded in factual and relevant information.  reach out at [email protected] to learn more about us.. https://lnkd.in/gjxmFDpY Aarushi Malhotra Akshat Prasad Chirag Thakur Nirmitk Tripathii Namita Goyal Muskan Vashistha Harpreet Singh Tushar Agarwal Tanishq Kalra VIRENDRA MITTAL SESH SUKHDEO Rishu Patel Priyanshi Prajapati Divyanshi Gatik Gupta Prerna Arya yogesh soni Madhu Gupta Aditi Kalra Blu Parrot Jessica Mehra

18

Totally agree with you, Anthony. The name of the game is to get the right context as tight as possible to the question or task on hand. Graphs are one good way to do this, but you should use all the tools in your belt. With our solution, we use graphs, tables, key value pairs, geospatial information, regular search, vectors, and so on. When you combine all these things, the probability of getting the right context as much higher.

5

🎥 Discover how a purpose-built database for multimodal AI can save you time and money in this insightful talk by Vishakha Gupta, Founder & CEO of ApertureData. The replay will be available in the comment section below! 👇 Key Takeaways from the Talk: ✅ Building the right data tools for multimodal AI ✅ Challenges faced in visual AI and chat support use cases ✅ Importance of indexing high-dimensional embeddings and metadata filtering ✅ ApertureDB's unified backend for multimodal data management ✅ Real-life examples of increased productivity and cost savings with ApertureDB Have you experienced the pain of dealing with complex data workflows in AI? What solutions have you tried? Share your thoughts in the comments! 👇 #AI #MultimodalAI #Database #AItools #ApertureDB

5

3 Comments

I believe a solution to this is currently being developed.

5

Revised Evaluation of LLM Compression Using Distance Metrics Researchers from Microsoft Research, India, propose an advanced evaluation framework incorporating distance metrics like KL-Divergence and percentage flips alongside traditional accuracy metrics. https://is.gd/MDzsSN #ai #aitechnology #artificialintelligence #llm #llmcompression #machinelearning

1

1 Comment