Nathan Lambert on the rise of "thinking" language models
AI Summer
00:00
Evolving Language Models: From Scaling to Understanding
This chapter explores the transition of language models from traditional post-training methods to innovative approaches focused on understanding ethical distinctions. It examines the complexities involved in calibrating models and the significant impact of evolving reward signals on coding capabilities. The discussion also addresses the challenges in achieving artificial general intelligence and the ongoing debate about necessary research breakthroughs beyond just scaling up model sizes.
Transcript
Play full episode
