Tanat Tonguthaisri, CISSP®’s Post

Exciting news! Our latest blog post introduces AutoSurvey, a groundbreaking methodology for automating the creation of comprehensive literature surveys in rapidly evolving fields like artificial intelligence. This innovative approach leverages large language models (LLMs) and addresses key challenges faced in traditional survey paper creation. Learn more about AutoSurvey's systematic approach and experimental validation in our new post: https://bit.ly/4cqbtBK #ArtificialIntelligence #AutoSurvey

Curious about the inner workings of Large Language Models (LLMs)? At the heart of LLMs lies the Transformer architecture—an innovative yet complex technology that's not always easy to grasp. That's where TRANSFORMER EXPLAINER comes in. This interactive visualization tool is specifically designed to make understanding Transformers more accessible to non-experts. By running a live GPT-2 instance directly in your browser, the tool allows you to experiment with your own inputs and see, in real-time, how the model's internal components and parameters collaborate to predict the next tokens. Dive in and explore the mechanics behind one of the most transformative technologies in AI!

🚀 Excited to share my latest article: Quantization – Making Large Language Models Lighter and Faster! 🤖💡 In this piece, I dive deep into quantization, a game-changing technique that allows us to reduce the size of large language models (LLMs) like GPT and LLaMA, making them more efficient and accessible for everyday devices without compromising performance. Key takeaways include: Simplifying data types (from FP32 to FP16, INT8, etc.) How it improves memory efficiency, speed, and energy consumption The trade-offs and challenges involved in this optimization This is perfect for anyone looking to deploy powerful models on edge devices or explore innovative AI solutions. I hope it helps you unlock new possibilities in your work! Read the full article here 👉 https://lnkd.in/gqVrH5UM #AI #MachineLearning #LLMs #Quantization #TechInnovation #EdgeComputing

Discover 6 AI prompt libraries to enhance your use of language models and image generation tools.

Retrieval-augmented generation (RAG) combines information retrieval from proprietary data sources with text generation to improve the accuracy of large language models. Learn strategies for implementing a RAG system capable of transforming vast amounts of data in this upcoming Tech Decoded webinar. Register now: https://intel.ly/3XMHTly #ArtificialIntelligence #GenerativeAI #RetrievalAugmentedGeneration

🌟 Excited to share a breakthrough in Large Language Models (LLMs) efficiency: SUBLLM. This innovation integrates subsampling, upsampling, and bypass modules, resulting in remarkable enhancements in both training and inference speeds as well as memory usage when compared to LLaMA. Find out more about this novel architecture and its impact on LLMs here: https://bit.ly/4en487x #LanguageModels #AI #Innovation

🌟 Exploring Meta's Revolutionary LCM Architecture 🌟 Excited to share my latest article, where I dive deep into Meta's groundbreaking shift from Large Language Models (LLMs) to Large Context Models (LCMs). 🚀 In this article, I discuss the fundamental differences between LLMs and LCMs, highlighting how LCMs excel in handling larger contexts, improving scalability, and optimizing efficiency in real-world applications. This shift marks a significant advancement in the AI landscape, with the potential to transform how we think about language processing and contextual understanding. 🔗 Read the full article here: https://lnkd.in/dwPdsDxy #AI #Meta #LLM #LCM #ArtificialIntelligence #Innovation

🎉 Excited to share that our paper "Process Modeling with Large Language Models", co-authored with Alessandro Berti, Daniel Schuster, and Wil van der Aalst, has been accepted at the international working conference on Exploring Modeling Methods for Systems Analysis and Development (EMMSAD2024)! In this paper, we propose a framework that leverages large language models for the automated generation and iterative refinement of process models starting from textual descriptions in natural language. Our framework involves innovative prompting strategies for an effective utilization of large language models, along with a secure model generation protocol and an error-handling mechanism. Preliminary results demonstrate the framework's ability to streamline process modeling tasks, underscoring the transformative potential of generative AI in the business process management field. Check out the preprint at: https://lnkd.in/ew4gKw2v #BusinessProcessManagement #ProcessModeling #GenerativeAI #LargeLanguageModels

🚀 Day 8 of ANAIS: Exploring Large Language Models (LLMs)! Today's session was all about diving deep into the transformative world of Large Language Models (LLMs) and how they are reshaping AI. Here's what we covered: Overview of LLM Architecture Training Techniques and Scaling Challenges Real-World Applications of LLMs Fine-Tuning Models for Specific Tasks The potential of LLMs to enhance language understanding and AI interaction is truly exciting. I can’t wait to see how these models continue to evolve and influence the AI landscape! 🌟 #LLM #LargeLanguageModels #ArtificialIntelligence #MachineLearning #GeometricDeepLearning #AITraining #ANAIS #Innovation

#SignLLM is a new large language model that generates sign language gestures from text inputs or prompts. The model was trained on the Prompt2Sign dataset, which includes American Sign Language (#ASL) and seven other sign languages. In the video below, you can see the result of style-transferring the stick-figure-like SignLLM videos to a more general AI video generation model. This is the work of Sen Fang et al, Rutgers University, The Australian National University (ANU), CSIRO's Data61, Carnegie Mellon University, The University of Texas at Dallas, and University of Central Florida.