
Scaling Up Test-Time Compute with Latent Reasoning with Jonas Geiping - #723
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
The Evolution of Reasoning in Machine Learning
This chapter examines the evolution of reasoning methodologies in machine learning, highlighting the differences between verbalized and algorithmic reasoning. It discusses the potential benefits of incorporating recurrence and adaptive exits to enhance model performance and efficiency. The chapter also explores the specialization of tokens during training and testing, emphasizing the importance of dynamic computation based on complexity.
00:00
Transcript
Play full episode
Remember Everything You Learn from Podcasts
Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.