Nick Tarazona, MD’s Post

🚨Are Large Language Models Chameleons⁉️ A very interesting propostion🤓 👉 Do large language models (LLMs) have their own worldviews and personality tendencies? 👉 a paper by Mingmeng Geng Sihong He Roberto Trotta from SISSA UT Arlington College of Engineering Imperial College London 👉 there is no clear cut answer to the question but different LLMs have different simulations outcomes 👉 great to have the question on the agenda 👉 need to see how that topic evolves, but clearly it is a key issue to see whether AI may be able to adjust to real life situations and take different "personalities" #artificialintelligence #future #riskmanagement

The Strategic Capabilities of Large Language Models

This Machine Learning Paper Introduce PISSA: Principal Singular Values and Singular Vectors Adaptation of Large Language Models Fine-tuning large language models

Fine-Tuning Large Language Models 🚀 The following one-hour workshop by Oren Sultan provides an introduction to fine-tuning large language models. The workshop covers topics such as fine-tuning approaches, prompt engineering, RAG methods, and demo fine-tuning process using LLaMA 2-7b-chat LLM. 📽️ https://lnkd.in/g3nhNmAb ⭐️ 𝑱𝒐𝒊𝒏 𝒎𝒚 𝑫𝒂𝒕𝒂 𝑺𝒄𝒊𝒆𝒏𝒄𝒆 𝑪𝒉𝒂𝒏𝒏𝒆𝒍 👉🏼 https://lnkd.in/g_GdP-pf #llm #datascience #rag #ai

Enhancing Text Embeddings in Small Language Models: A Contrastive Fine-Tuning Approach with MiniCPM LLMs excel in natural language understanding but are resource-intensive, limiting their accessibility. Smaller models like MiniCPM offer better scalability but often need targeted optimization to perform. Text embeddings, vector representations that... https://lnkd.in/dsuxnkv2 #AI #ML #Automation

🚀 Part 1/13 of my "Know-All-About-LLMs" Blog Series is Live! 🚀 Excited to share the first post in my new blog series on Large Language Models (LLMs)! In this series, I’ll be breaking down the history, timeline, and types of LLMs, providing both beginners and enthusiasts a deeper understanding of this transformative AI technology. 🔍 In Part 1, we cover: - What Large Language Models are - A brief history of NLP and the rise of deep learning - Key milestones in LLM development, from early models to the latest innovations like GPT-4 - Different types of LLMs (Autoregressive, Masked, Multimodal, and more) 📅 This post serves as the foundation of a 13-part series, where we'll dive into the math, mechanics, and applications that make LLMs the backbone of modern AI. Stay tuned for more exciting content in the coming days! #LLMs #AI #MachineLearning #DeepLearning #NLP #GPT #Transformers #ArtificialIntelligence #TechBlog #Medium #DataScience

Discover how Dual Chunk Attention (DCA) transforms long-context comprehension in Large Language Models (LLMs). This innovative approach optimizes attention mechanisms within and between data chunks, enhancing the efficiency of processing extensive text sequences. Dive into this article to explore how DCA overcomes the limitations of traditional transformers and revolutionizes long-context handling in LLMs. Read the full article here: https://lnkd.in/g-P4UJX9 #generativeai #largelanguagemodels

Just published my latest article on Medium: "Prompt Engineering and Transformation with Llama 3.1"! 🚀 In this article, I dive into the art and science of prompt engineering with advanced language models like Llama 3.1. It covers: Using Retrieval-Augmented Generation (RAG) to enrich model outputs Leveraging Program-Aided Language Models (PALM) for handling complex calculations Practical examples and strategies to get the most out of large language models Whether you're a researcher, developer, or just curious about AI, you'll find something valuable here. Prompt engineering is revolutionizing how we interact with AI, making it more accessible and powerful than ever. I'd love to hear your thoughts or your experiences with prompt engineering. Let's push the boundaries of AI together! 🤖✨ #AI #PromptEngineering #Llama3 #MachineLearning #ArtificialIntelligence #MediumPost #AIInnovation

Improving Factuality and Reasoning in Language Models through Multiagent Debate "...In this paper, we present a complementary approach to improve language responses where multiple language model instances propose and debate their individual responses and reasoning processes over multiple rounds to arrive at a common final answer..." https://lnkd.in/eUWf__st I have used this approach as soon as I got my hands on more than 1 Large Language Model. I now use Intel Jesus as lead (a GPT agent I created, finetuned and customized with prioritized information sourcing https://lnkd.in/ednmU5kU) combined with Gemini Advanced and Claude Haiku as assistants. Human in the loop is required, with little to no intervention and steering, and the final result is, consistently, pretty impressive! 🔥 🔥 🔥 🔥

New research shows large language models have achieved human-level performance on higher-order theory of mind tasks! 🤖🧠 A team of researchers tested five LLMs, including GPT-4 and Flan-PaLM, on a new benchmark called the Multi-Order Theory of Mind Q&A. Theory of mind is the ability to reason about mental states recursively (e.g. I think that you believe that she knows). The results are impressive: - GPT-4 and Flan-PaLM reached adult human-level performance overall - GPT-4 even exceeded human performance on complex 6th-order inferences! This suggests the most advanced LLMs have developed a general capacity for higher-order theory of mind, likely due to a combination of model size and finetuning. Given how crucial theory of mind is for human social interactions, this has major implications as LLMs are deployed in more user-facing applications. Check out the full paper for more details: https://lnkd.in/ezYh-XCv #AI #LanguageModels #TheoryOfMind #GPT4 #Research