AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
2min chapter
AI Trends 2023: Reinforcement Learning - RLHF, Robotic Pre-Training, and Offline RL with Sergey Levine - #612
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
RL With Language Models
The idea in our work was that instead of viewing RL with language models as this bandit problem where you optimize for the highest reward utterance, you can instead take it to the other extreme and view every single token as a decision. And the way this was done was actually... Was this across a series of dialogue exchanges or tokens within a single utterance or turn? Yeah, so it's across the entire dialogue. So if the next token is the model's choice, then it know that and if it ends the line, then it knows that the next token will come from the human.
00:00
Transcript
Episode notes
Get the Snipd
podcast app
Unlock the knowledge in podcasts with the podcast player of the future.
AI-powered
Discover
highlights
Listen to the best highlights from the podcasts you love and dive into the full episode
Save any
moment
Hear something you like? Tap your headphones to save it with AI-generated key takeaways
Share
& Export
Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more
AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Discover
highlights
Listen to the best highlights from the podcasts you love and dive into the full episode