Pierluca D'Oro and Martin Klissarov
TalkRL: The Reinforcement Learning Podcast
Analyzing the Gap between RL and Language Models
The speakers discuss the challenges of incorporating language models into reinforcement learning and how NetHack makes it easier. They explore different methods to bridge the gap, including designing hand-selected value functions and fine-tuning language models. The chapter also touches on the motivation behind using large language models for fine-tuning and how RL can build upon their high-level knowledge.
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.
