"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis cover image

Everything You Wanted to Know About LLM Post-Training, with Nathan Lambert of Allen Institute for AI

"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis

00:00

Contamination and Tuning in Language Models

This chapter explores the challenges of detecting contamination in language model training datasets, including prompt matches and the effects of synthetic data generation. It also discusses the post-training adjustments of model weights, the distinctions between instruction tuning and preference tuning, and the role of KL regularization in controlling model changes. Additionally, the chapter highlights the complexities of evaluation strategies and the impacts of different training techniques on model performance.

Transcript
Play full episode

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!
App store bannerPlay store banner
Get the app