Vanishing Gradients

Ben Taylor, CEO of VEOX Inc., Joe Reis, co-founder of Ternary Data, and Juan Sequeda, Principal Scientist at Data.World, discuss the evolution of AI from prompt engineering to goal engineering. They explore whether generative AI is more akin to an electrifying revolution or a blockchain phase. The panel highlights the importance of tackling the POC-to-production gap, understanding AI's failure modes, and balancing executive enthusiasm with employee workload. They also examine how AI's combinatorial abilities can redefine strategies, paralleling the success of AlphaZero in gaming.

In this conversation with Charles Frye, Developer Advocate at Modal, listeners gain insights into the intricate world of GPUs and their critical role in AI and LLM development. Charles explains the importance of VRAM and how memory can become a bottleneck. They tackle practical strategies for optimizing GPU usage, from fine-tuning to training large models. The discussion also highlights a GPU Glossary that simplifies complex concepts for developers, along with insights on quantization and the economic considerations in using modern hardware for efficient AI workflows.

Hugo chats with Ravin Kumar, a Senior Research Data Scientist at Google Labs, whose career journey includes roles at SpaceX and Sweetgreen. They delve into the balance between technical rigor and practical utility in generative AI. Ravin shares insights on building scalable AI systems, such as using Gemma to optimize bakery operations. He emphasizes the critical role of UX in AI adoption, showcases the Notebook LM tool in action, and explores how AI can aid small businesses—demonstrating the transformative power of accessible technology.

Jason Liu, an independent AI consultant with a background at Meta and Stitch Fix, joins the discussion. He shares insights into structuring valuable consulting contracts and shifting from hourly billing to larger deals. Engaging in a live role-play, Jason coaches the host on effective client interaction and pricing strategies. The conversation also highlights the shift from deterministic to probabilistic AI systems, emphasizing the importance of understanding client motivations and fostering meaningful relationships in the evolving freelance landscape.

Join Sander Schulhoff, a specialist in prompt engineering, Philip Resnik, a computational linguistics professor, and Dennis Peskoff from Princeton as they delve into the cutting-edge world of AI. They explore the security risks of prompt hacking and its implications for military use. Discussion highlights include the evolving role of generative AI across various fields, innovative techniques for improving AI self-criticism, and the pressing need for energy-efficient large language models. Their insights offer a fascinating glimpse into the future of AI research.

Hugo speaks with three leading figures from the world of AI research: Sander Schulhoff, a recent University of Maryland graduate and lead contributor to the Learn Prompting initiative; Philip Resnik, professor at the University of Maryland, known for his pioneering work in computational linguistics; and Dennis Peskoff, a researcher from Princeton specializing in prompt engineering and its applications in the social sciences. This is Part 1 of a special two-part episode, prompted—no pun intended—by these guys being part of a team, led by Sander, that wrote a 76-page survey analyzing prompting techniques, agents, and generative AI. The survey included contributors from OpenAI, Microsoft, the University of Maryland, Princeton, and more. In this first part, we’ll explore the critical role of prompt engineering, & diving into adversarial techniques like prompt hacking and the challenges of evaluating these techniques. we’ll examine the impact of few-shot learning and the groundbreaking taxonomy of prompting techniques from the Prompt Report. Along the way, we’ll uncover the rich history of natural language processing (NLP) and AI, showing how modern prompting techniques evolved from early rule-based systems and statistical methods. we’ll also hear how Sander’s experimentation with GPT-3 for diplomatic tasks led him to develop Learn Prompting, and how Dennis highlights the accessibility of AI through prompting, which allows non-technical users to interact with AI without needing to code. Finally, we’ll explore the future of multimodal AI, where LLMs interact with images, code, and even music creation. Make sure to tune in to Part 2, where we dive deeper into security risks, prompt hacking, and more. LINKS The livestream on YouTube The Prompt Report: A Systematic Survey of Prompting Techniques Learn Prompting: Your Guide to Communicating with AI Vanishing Gradients on Twitter Hugo on Twitter Vanishing Gradients' lu.ma calendar Vanishing Gradients on YouTube

In this enlightening discussion, Dr. Chelle Gentemann, an Open Science Program Scientist at NASA, delves into NASA's groundbreaking efforts to integrate AI into the research lifecycle. She explores innovative metrics developed to measure the impact of open science, moving beyond traditional publication metrics. Gentemann also shares insights on AI applications in various NASA projects, from oceanography to the universe's origins, and discusses the challenges of implementing open science in government. Moreover, she highlights the need for reform in academic incentives to promote data sharing and collaboration.

Guests Ines Montani and Matthew Honnibal, founders of Explosion AI and creators of the widely-used spaCy library, discuss the evolution of natural language processing (NLP) in industry. They share insights on balancing large and small AI models, challenges in modularity and privacy, and the impact of regulation on innovation. Their transition to a smaller company highlights lessons learned in the AI startup world. The conversation touches on the importance of data quality and open-source tools while celebrating the practical applications of AI for data scientists and enthusiasts alike.

Hugo speaks with Dan Becker and Hamel Husain, two veterans in the world of data science, machine learning, and AI education. Collectively, they’ve worked at Google, DataRobot, Airbnb, Github (where Hamel built out the precursor to copilot and more) and they both currently work as independent LLM and Generative AI consultants. Dan and Hamel recently taught a course on fine-tuning large language models that evolved into a full-fledged conference, attracting over 2,000 participants. This experience gave them unique insights into the current state and future of AI education and application. In this episode, we dive into: The evolution of their course from fine-tuning to a comprehensive AI conference The unexpected challenges and insights gained from teaching LLMs to data scientists The current state of AI tooling and accessibility compared to a decade ago The role of playful experimentation in driving innovation in the field Thoughts on the economic impact and ROI of generative AI in various industries The importance of proper evaluation in machine learning projects Future predictions for AI education and application in the next five years We also touch on the challenges of using AI tools effectively, the potential for AI in physical world applications, and the need for a more nuanced understanding of AI capabilities in the workplace. During our conversation, Dan mentions an exciting project he's been working on, which we couldn't showcase live due to technical difficulties. However, I've included a link to a video demonstration in the show notes that you won't want to miss. In this demo, Dan showcases his innovative AI-powered 3D modeling tool that allows users to create 3D printable objects simply by describing them in natural language. LINKS The livestream on YouTube Educational resources from Dan and Hamel's LLM course Upwork Study Finds Employee Workloads Rising Despite Increased C-Suite Investment in Artificial Intelligence Episode 29: Lessons from a Year of Building with LLMs (Part 1) Episode 30: Lessons from a Year of Building with LLMs (Part 2) Dan's demo: Creating Physical Products with Generative AI Build Great AI, Dan's boutique consulting firm helping clients be successful with large language models Parlance Labs, Hamel's Practical consulting that improves your AI Hamel on Twitter Dan on Twitter Vanishing Gradients on Twitter Hugo on Twitter

Join Shreya Shankar, a UC Berkeley researcher specializing in human-centered data management systems, as she navigates the exciting world of large language models (LLMs). Discover her insights on the shift from traditional machine learning to LLMs and the importance of data quality over algorithm issues. Shreya shares her innovative SPaDE framework for improving AI evaluations and emphasizes the need for human oversight in AI development. Plus, explore the future of low-code tools and the fascinating concept of 'Habsburg AI' in recursive processes.