Keith McNulty’s Post

The most intriguing aspect of LLMs is not just their capabilities, but the startups leveraging them to elevate human potential and organizational missions. These ventures aim to prove that LLMs can 'serve' humans better than humans, even at the high cost of scarce resources like water. As highlighted in the paper 'How Much Data is Needed to Train a Model?', efficient use of data and resources is crucial. These startups are pushing AI boundaries while balancing technological advancement and resource sustainability, striving for value-driven missions that benefit society.

AI Use Case: UAP expert, based on 29 video transcripts from the Sol Foundation Symposium at Stanford, and individual opening remarks from Congressional UAP hearings. System Prompt cribbed largely from Professor (Super) Synapse. Bonus: GPT helped me write Python to download video transcripts based on a playlist. I can use it to quickly build a custom bot that is an expert in [playlist topic], such as all the lectures from a convention that you didn't attend. It allows you to consider a whole body of information in context. What types of files do you upload for your GPT's knowledgebase/library? #AI #UAP

xLSTM introducing the Extended Long Short-Term Memory as a challenger for the ever growing wave of transformer architectures: European innovation introduces an advanced version of the traditional LSTM model by incorporating exponential gating and innovative memory structures to address its limitations. These modifications allow xLSTMs to perform competitively with contemporary Transformer models in language processing tasks, showcasing improved capability in handling complex memory operations and scaling to large model architectures. Paper: https://lnkd.in/ezxFfReb -- If you liked this article you can join 60,000+ practitioners for weekly tutorials, resources, OSS frameworks, and MLOps events across the machine learning ecosystem: https://lnkd.in/eRBQzVcA #ML #MachineLearning #ArtificialIntelligence #AI #MLOps #AIOps #DataOps #augmentedintelligence #deeplearning #privacy #kubernetes #datascience #python #bigdata

Hello, my friends! 👋 In my research, I used to employ supervised AI techniques for data analysis and engineering result prediction. But this holiday season, my curiosity led me down the rabbit hole of AI. It's a wonderland where unsupervised machines not only learn but also perform various tasks. The most interesting part is that they can learn from their mistakes and perform better than before, just like us with a small difference, they can perform thousands of rounds in just a blink of an eye 😅. I challenged myself to create an AI agent (001 😁) that could play Tic Tac Toe strategically. I implemented Q-Learning in combination with DNN to create an agent who can consider the effects of current decisions on future situations. After numerous iterations, 001 agent can now beat a random player function 10,000 vs 2,000 (the result is attached). The code is now available on my GitHub. Dive in, explore, and maybe even challenge the AI! 😁 [https://lnkd.in/gpHTJQYd] This is just the beginning. I'm eager to explore much more complex AI projects, connect with fellow AI enthusiasts, and continue unravelling the mysteries of this captivating field. #AI #TicTacToe #HolidayProject #ArtificialIntelligence #Python #GitHub #Datascience #Dataanalysis

xLSTM introducing the Extended Long Short-Term Memory as a challenger for the ever growing wave of transformer architectures: European innovation introduces an advanced version of the traditional LSTM model by incorporating exponential gating and innovative memory structures to address its limitations. These modifications allow xLSTMs to perform competitively with contemporary Transformer models in language processing tasks, showcasing improved capability in handling complex memory operations and scaling to large model architectures. Paper: https://lnkd.in/gbfpXBYe -- If you liked this article you can join 60,000+ practitioners for weekly tutorials, resources, OSS frameworks, and MLOps events across the machine learning ecosystem: https://lnkd.in/eRBQzVcA #ML #MachineLearning #ArtificialIntelligence #AI #MLOps #AIOps #DataOps #augmentedintelligence #deeplearning #privacy #kubernetes #datascience #python #bigdata

Chip Huyen dives into solutions for the probabilistic undeterminism of Large Language Models through robust sampling methods: In order to address the probabilistic nature of machine learning models in text generation we have to focusing on the challenges and strategies for achieving desired outputs. Sampling methods can be explored to address this, such as top-k, temperature and top-p sampling, highlighting their impact on the balance between creativity and consistency in model responses. The article also discusses test time sampling, a technique for improving model performance by generating multiple outputs and selecting the best one, and the concept of structured outputs, emphasizing the importance of guiding models to produce specific formats, especially in applications requiring precise structures like text-to-SQL. This piece is particularly insightful for ML practitioners looking to optimize text generation models in production environments. Deep dive: https://lnkd.in/ecX4yDAz -- If you liked this article you can join 50,000+ practitioners for weekly tutorials, resources, OSS frameworks, and MLOps events across the machine learning ecosystem: https://lnkd.in/eRBQzVcA #ML #MachineLearning #ArtificialIntelligence #AI #MLOps #AIOps #DataOps #augmentedintelligence #deeplearning #privacy #kubernetes #datascience #python #bigdata

Learning a bit more about Generative AI this week🚀 ,in particular I have been focused on learning about a particular technique called Variational Autoencoders (VAEs) which allows you to generate new images 📷 but can be extended to text, audio and tabular data.💡 Using the 'Deep Learning with Python' book by François Chollet (thanks Joshua Uwaifo for the book recommendation!) I was able to apply it on the classical MNIST digits dataset on the example shown in the image below. 🔢 See project code here: https://lnkd.in/df8mtV5i Furthermore read and summarised 'Variational Lossy Autoencoders' research paper which uses a statistical model used for analysing time series data to improve on normal VAEs: https://lnkd.in/etSPCyFd 🚀 #GenerativeAI #MachineLearning #DataScience #DeepLearning #AIResearch #VariationalAutoencoders #MNIST #5 #SoliDeoGloria #bara #audienceof1