Published Oct 26, 2023

Jonathan Frankle: From Lottery Tickets to LLMs

Jonathan Frankle, Chief Scientist at MosaicML, delves into his empirical AI research philosophy and the lottery ticket hypothesis, discusses AI policy challenges and MosaicML's evolution, and explores neural network pruning complexities, all while highlighting the importance of data-driven approaches and technologist involvement in shaping AI advancements.

Episode Highlights

Topics covered

Episode Highlights

Related Episodes

Hattie Zhou: Lottery Tickets and Algorithmic Reasoning in LLMs
Answers 383 questions
Zachary Lipton: Where Machine Learning Falls Short
Answers 383 questions
Joel Lehman: Open-Endedness and Evolution through Large Models
Answers 383 questions
Peter Lee: Computing Theory and Practice, and GPT-4's Impact
Answers 383 questions
Seth Lazar: Normative Philosophy of Computing
Answers 383 questions
Hugo Larochelle: Deep Learning as Science
Answers 383 questions
Luis Voloch: AI and Biology
Answers 383 questions
Subbarao Kambhampati: Planning, Reasoning, and Interpretability in the Age of LLMs
Answers 383 questions
Sara Hooker: Cohere For AI, the Hardware Lottery, and DL Tradeoffs
Answers 383 questions
Stuart Russell: The Foundations of Artificial Intelligence
Answers 383 questions
Terry Winograd: AI, HCI, Language, and Cognition
Answers 383 questions
Kristin Lauter: Private AI, Homomorphic Encryption, and AI for Cryptography
Answers 383 questions
François Chollet: Keras and Measures of Intelligence
Answers 383 questions
Manuel & Lenore Blum: The Conscious Turing Machine
Answers 383 questions
Preetum Nakkiran: An Empirical Theory of Deep Learning
Answers 383 questions

Dexa/Daniel Bashir

Jonathan Frankle: From Lottery Tickets to LLMs

Topics covered

Popular Clips

Learning Rate Insights

Empirical Foundations

Journey into AI

Efficiency in Hypothesis Testing

Two-Sided Questions

Optimization Insights

AI Policy Evolution

Pruning Neural Networks

Lottery Ticket Insights

Evaluating Metrics

Mosaic's Evolution

Lottery Ticket Hypothesis

Regulating AI Outcomes

Building Efficient Models

Episode Highlights

Jonathan's Research Philosophy

AI Policy and Implications

Evolution of MosaicML

Understanding Pruning

Related Episodes

Hattie Zhou: Lottery Tickets and Algorithmic Reasoning in LLMs

Zachary Lipton: Where Machine Learning Falls Short

Joel Lehman: Open-Endedness and Evolution through Large Models

Peter Lee: Computing Theory and Practice, and GPT-4's Impact

Seth Lazar: Normative Philosophy of Computing

Hugo Larochelle: Deep Learning as Science

Luis Voloch: AI and Biology

Subbarao Kambhampati: Planning, Reasoning, and Interpretability in the Age of LLMs

Sara Hooker: Cohere For AI, the Hardware Lottery, and DL Tradeoffs

Stuart Russell: The Foundations of Artificial Intelligence

Terry Winograd: AI, HCI, Language, and Cognition

Kristin Lauter: Private AI, Homomorphic Encryption, and AI for Cryptography

François Chollet: Keras and Measures of Intelligence

Manuel & Lenore Blum: The Conscious Turing Machine

Preetum Nakkiran: An Empirical Theory of Deep Learning

Jonathan Frankle: From Lottery Tickets to LLMs

Topics covered

Popular Clips

Episode Highlights

Jonathan's Research Philosophy

AI Policy and ImplicationsJonathan Frankle explores the evolution of AI policy and the challenges of regulating AI technologies. He emphasizes the importance of focusing on desired outcomes and the role of technologists in contributing to policy discussions.

AI Policy and Implications

Evolution of MosaicML

Understanding PruningThe discussion shifts to the complexities of pruning metrics and the challenges of early training in neural networks. Jonathan Frankle shares insights on the intricacies of evaluating pruning methodologies and the evolving landscape of deep learning research.

Understanding Pruning

Related Episodes