AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
2min chapter
Advancing Deep Reinforcement Learning with NetHack, w/ Tim Rocktäschel - #527
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
How to Train a Student Agent to Navigate Mazes?
We started with the grit worlds, and we trained agents to basically generate mazes. And then we take it out of that kind of space and actually present it with handcrafted, quite tricky mazes. We see strong zero shot generalization to these kind of held out problems. Then we thoght ok. If that works for mazes, maybe we can also do this in a continuous control environment. So we actually moved over to car racing, and we let the teacher to basically overtime generate formula one tracks. I mean, not actual formula one tracks, but basically race tracks. And we have te stuen trying to get through these as quickly as possible.
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