Santosh Sawant

Exciting developments in AI-driven enterprise solutions! 🚀 Dheeraj Pandey, CEO and Founder of DevRev, has shared groundbreaking insights on how their advanced knowledge graph is revolutionizing enterprise operations. Here's what caught my attention: • Seamless Integration: Devrev's platform integrates data from legacy systems into a real-time, interconnected network of customer, product, employee, and work data. .• AI-Native Advantage: The robust knowledge graph powers an AI-native platform, enabling low-latency search, analytics, and workflows - crucial for AI to outpace human input • Five-Layer Architecture: Unlike retrofitted legacy systems, Devrev boasts a purpose-built, five-layer knowledge graph that fuels cutting-edge AI applications for customer service, software development and CRM. .• Flexibility for Businesses: Companies can start small with simple integrations or fully embrace Devrev's AI-native suite to enhance productivity. This innovative approach promises to transform how enterprises leverage AI to work smarter, not harder. It's fascinating to see how knowledge graphs are becoming the backbone of next-gen AI solutions in business operations. What are your thoughts on this AI-driven approach to enterprise management? How do you see it impacting various industries? #AI #EnterpriseAI #KnowledgeGraph #Innovation #BusinessTechnology https://lnkd.in/eFjfW7k2

26

LLMs - Function Calling In one of my previous posts i have talked about function calling with LLMs. I have attached the code snippet to illustrate function calling using a mistral model. The gist of the code is i have asked about the current stock price of a company to a LLM, which it can't answer. Now with the help of function calling leveraging yahoo finance, the LLM can provide the stock price. Invoking mistral model needs an api key which can be obtained by having a mistral subscription.

2

💥💥💥 Align-SLM: Textless Spoken Language Models with Reinforcement Learning from AI Feedback Guan-Ting Lin, Prashanth Gurunath Shivakumar, Aditya Gourav, Yile Gu, Ankur Gandhe, Hung-yi Lee, Ivan Bulyko Abstract While textless Spoken Language Models (SLMs) have shown potential in end-to-end speech-to-speech modeling, they still lag behind text-based Large Language Models (LLMs) in terms of semantic coherence and relevance. This work introduces the Align-SLM framework, which leverages preference optimization inspired by Reinforcement Learning with AI Feedback (RLAIF) to enhance the semantic understanding of SLMs. Our approach generates multiple speech continuations from a given prompt and uses semantic metrics to create preference data for Direct Preference Optimization (DPO). We evaluate the framework using ZeroSpeech 2021 benchmarks for lexical and syntactic modeling, the spoken version of the StoryCloze dataset for semantic coherence, and other speech generation metrics, including the GPT4-o score and human evaluation. Experimental results show that our method achieves state-of-the-art performance for SLMs on most benchmarks, highlighting the importance of preference optimization to improve the semantics of SLMs. 👉 https://lnkd.in/dXBzaibk #machinelearning

4

👉🏼 Analyzing the Readability and User Adaptability of AtlasGPT in Neurosurgical Education 🤓 Raj Swaroop Lavadi 👇🏻 https://lnkd.in/ejB7mXJQ 🔍 Focus on data insights: - 📊 Readability scores varied significantly across user profiles, indicating the effectiveness of tailoring content to specific users. - 🧠 Medical student responses showed better reading ease than those intended for surgeons or residents, suggesting a need for reevaluation of complexity in content delivery. - 📈 The mean grade levels for patient responses exceeded recommended readability standards, raising concerns about accessibility for non-professionals. 💡 Main outcomes and implications: - 🔍 The study highlights the necessity for generative tools like AtlasGPT to adapt content comprehensibility according to user expertise. - ⚖️ Findings call for improved strategies in presenting information to patients, ensuring clarity and understanding of medical concepts. - 📚 A need for continuous evaluation and modification of educational resources within neurosurgery to meet diverse audience needs. 📚 Field significance: - 🌐 Enhances understanding of how AI-generated content can bridge knowledge gaps in medical education. - 🏥 Promotes a shift towards greater inclusivity and accessibility of medical information for patients. - 🔗 Sets a precedent for future studies on AI tools in professional education and communication within healthcare. 🗄️: [#AtlasGPT #Neurosurgery #Readability #MedicalEducation #AIInHealthcare #PatientCommunication #UserAdaptability #ContentAccessibility]

2

Kolmogorov–Arnold Networks seem like a paradigm shift in ML architectures. Curious to see how they evolve. "While MLPs have fixed activation functions on nodes ("neurons"), KANs have learnable activation functions on edges ("weights"). KANs have no linear weights at all -- every weight parameter is replaced by a univariate function parametrized as a spline. " https://lnkd.in/gdPk_qCQ

28

I came across an excellent GitHub repository packed with hands-on Jupyter notebooks for anyone eager to explore LLMs. It covers: 🔹 Introduction to LLMs – Understanding the basics. 🔹 Tokens & Embeddings – The foundation of language modeling. 🔹 Model Internals – Peeking under the hood of LLMs. 🔹 Text Applications – Classification, clustering, and topic modeling. 🔹 Prompt Engineering – Crafting prompts for better performance. 🔹 Advanced Text Generation – Tools and techniques. 🔹 Semantic Search – Intelligent information retrieval. 🔹 Multimodal Models – Combining text and image understanding. 🔹 Customization – Building embeddings, fine-tuning BERT, and generative models. Big thanks to the repo creator for sharing such a valuable resource. Here is the repo: https://lnkd.in/gUKg7Jp7

77

2 Comments

Anyone dabbling in AI generation?! I have been building my own models to understand how deep learning models operate and how to improve them, but I have stumbled across an amazing resource. I am not the one to gatekeep You don't always need to build your model from scratch, you can leverage other models that solve similar problems to yours, it's called transfer learning and you can review latest research in which model performs best on which task based on numerous benchmarks right here: https://lnkd.in/dCSndjsM #AIGeneration #DeepLearning #MachineLearning #TransferLearning #AIResearch #PapersWithCode #TechInnovation #AICommunity #ModelBuilding #DataScience #ArtificialIntelligence #MLModels #TechResources #AIInsights #DeepLearningModels

2

India to Shape the Future of AI: Insights from Meta's Yann LeCun and Nandan Nilekani 🇮🇳🤖 Exciting times ahead for India in the field of Artificial Intelligence! At the Subra Suresh Distinguished Lecture Series hosted by IIT Madras, Yann LeCun, VP & Chief AI Scientist at Meta, highlighted the country's immense potential in driving AI innovation, not just in product development but also in cutting-edge research. 🌍💡 With a rich talent pool and vibrant tech ecosystem, India is making significant strides in areas like natural language understanding, crucial for serving diverse populations. LeCun emphasized Meta’s commitment to open-source AI, which fosters global collaboration and creativity, enabling AI systems that empower individuals and transform societies. Nandan Nilekani, Non-Executive Chairman at Infosys, also shared his vision for India’s AI future. He stressed that multilingual Large Language AI Models will emerge in the next 12 to 18 months, accelerating AI adoption and contributing to solving the nation’s diverse challenges. Meta’s open-source approach, including its game-changing Llama AI models, will provide developers with the tools to address key issues across sectors like agriculture, healthcare, citizen services, and governance. 🌱🏥💼 Exciting opportunities ahead for India to lead on the global AI stage! Let’s harness the power of AI for societal transformation. 🚀 #AI #ArtificialIntelligence #Meta #NandanNilekani #OpenSource #AIInnovation #IndiaTech #DigitalTransformation #AIForGood

5

1 Comment

🚀 Cache-Augmented Generation (CAG) vs. Retrieval-Augmented Generation (RAG): The Future of Efficiency in AI In the world of AI, both CAG and RAG play a pivotal role in enhancing language model workflows, but their approaches differ significantly: 💡 CAG: A game-changer that eliminates real-time retrieval by integrating extended knowledge contexts into a large language model (LLM) upfront. With precomputed inference states stored in a key-value cache, CAG ensures ultra-fast, retrieval-free knowledge tasks—ideal for low-latency applications where efficiency and simplicity are key. 🔍 RAG: A powerhouse for real-time knowledge retrieval, dynamically fetching relevant documents to enhance model responses. While highly effective for applications requiring fresh or dynamic data, RAG involves additional retrieval steps, which can impact response speed and system complexity. Why Choose CAG Over RAG? ✅ Faster Response Times: No retrieval delays—perfect for real-time use cases. ✅ Cost-Effective: Reduced infrastructure demands by reusing precomputed states. ✅ Simpler Architecture: No need for external databases or retrieval systems. ✅ Consistency: Cached results ensure reliable, repeatable outputs. CAG shines in static or semi-static data scenarios, such as customer support, FAQs, and domain-specific bots. RAG remains unmatched for dynamic, real-time knowledge updates, but CAG is the future for speed and scalability. What’s your take on these evolving AI paradigms? #AI #MachineLearning #CAGvsRAG #LLMs #Innovation

16

𝗟𝗶𝗴𝗵𝘁 𝗻𝗼𝘁𝗲𝘀 𝗳𝗼𝗿 𝗮 𝗠𝗼𝗻𝗱𝗮𝘆 𝗲𝘃𝗲𝗻𝗶𝗻𝗴: I had been reading through Advanced RAG techniques and GraphRAGs late evening yesterday in perparation of my next blogs on LinkedIn. Some points that caught my attention: In order to improve the retrieval accuracy in the Retrieval-Augmented Generation (RAG) approach, we may use: ⭕Re-Ranking of Context: we use LLMs like GPT-4 or specialised models like ColBERT or Cohere to re-rank the context and then pass only the top most relevant contetxts to the LLM to get the final answer. 𝗧𝗵𝘂𝘀, 𝘄𝗲 𝘂𝘀𝗲 𝗮𝗻 𝗟𝗟𝗠 𝘁𝗼 𝗿𝗲-𝗿𝗮𝗻𝗸. 𝗧𝗵𝗲 𝗶𝗱𝗲𝗮 𝗯𝗲𝗶𝗻𝗴 𝘁𝗼 𝗶𝗺𝗽𝗿𝗼𝘃𝗲 𝘁𝗵𝗲 𝗰𝗼𝗻𝘁𝗲𝘅𝘁 𝗽𝗮𝘀𝘀𝗲𝗱 𝘁𝗼 𝘁𝗵𝗲 𝗟𝗟𝗠 ⭕GraphRAGS: Going from Community Summaries -> Community Answers ->Global Answers 📌 Fundamentally in a GraphRAG , we get chunks out of the source documents as in the standard RAG process 📌 We then carry out Entity Extraction and Relationship Capture from these chunks. This is essentially building up of Nodes and Edges of the Knowledge Graph. 📌 We carry out Element Summmaries: This is the summarization of all the nodes and relationships extracted. 📌 The community summaries are randomnly shuffled and the reason for shuffling being that they shouldn't be concentrated in a single context window. 📌 We make multiple calls to the LLM to answer the query against each of the chunk of the community summaries. 𝗔𝗻𝗱 𝗳𝗼𝗿 𝗲𝗮𝗰𝗵 𝗰𝗵𝘂𝗻𝗸 𝘄𝗲 𝗴𝗲𝗻𝗲𝗿𝗮𝘁𝗲 𝗶𝗻𝘁𝗲𝗿𝗺𝗲𝗱𝗶𝗮𝘁𝗲 𝗮𝗻𝘀𝘄𝗲𝗿𝘀 𝗮𝗻𝗱 𝗲𝗮𝗰𝗵 𝗮𝗻𝘀𝘄𝗲𝗿 𝗶𝘀 𝘀𝗰𝗼𝗿𝗲𝗱 𝗮𝗻𝗱 𝗿𝗮𝗻𝗸𝗲𝗱. 📌𝗔𝗴𝗮𝗶𝗻 𝗮𝗻 𝗟𝗟𝗠 𝗶𝘀 𝘂𝘀𝗲𝗱 𝘁𝗼 𝘀𝗰𝗼𝗿𝗲 𝘁𝗵𝗲 𝗮𝗻𝘀𝘄𝗲𝗿 𝗯𝗲𝘁𝘄𝗲𝗲𝗻 𝟬 𝗮𝗻𝗱 𝟭𝟬𝟬 𝗮𝗻𝗱 𝗳𝗶𝗻𝗮𝗹𝗹𝘆 𝘁𝗵𝗲 𝘁𝗼𝗽 𝗿𝗮𝗻𝗸𝗲𝗱 𝗮𝗻𝘀𝘄𝗲𝗿𝘀 𝗮𝗿𝗲 𝘁𝗮𝗸𝗲𝗻 𝗮𝗻𝗱 𝘂𝘀𝗲𝗱 𝘁𝗼 𝗴𝗲𝗻𝗲𝗿𝗮𝘁𝗲 𝗮 𝗳𝗶𝗻𝗮𝗹 𝗴𝗹𝗼𝗯𝗮𝗹 𝗮𝗻𝘀𝘄𝗲𝗿. The point I'm trying to make is: The above remined me of ChatGPT training process [https://lnkd.in/dWpuyCpV] . Here, we had human labellers rank the ChatGPT responses based on “Likert Scale” – the ranking being based on the quality of the responses. This was part of the Step 2 of ChatGPT training process. 𝗔𝗻𝗱 𝘁𝗵𝗲 𝘀𝘁𝗼𝗿𝘆 𝗮𝗳𝘁𝗲𝗿 < 𝟮 𝘆𝗲𝗮𝗿𝘀: the human ranking is already replaced by LLM ranking! Coming Soon: ⭕RAG Beyond Basics: Part 15 of my Graph Series of Blogs ⭕The Marriage of Retrieval Augmented Generation with Knowledge Graphs: Part 16 of my Graph Series of Blogs Watch this space! #graphs #deeplearning #rags #llms My Tech Blogs Library 1) https://lnkd.in/ejyWacXn 2) https://lnkd.in/bfrkRjq Follow Ajay Taneja for insights on Data Engg, Data Science, Finite Element Analysis, Non Linear Finite Element Analysis, Fatigue, Fracture Mechanics Image credits: Claudia Sonnet 3.5!

12