[11] Jacob Andreas - Learning from Language
The Thesis Review
00:00
How to Use Free Natural Language to Improve Reinforcement Learning
Gen. Oh had a bunch of work sort of concurrently with the policy sketches paper looking at those kinds of models. We are hopefully soon going to have some follow up work to that coming out and then also work from some folks at Google brain who are using you know not exactly natural language but like language parameterized sub goals. Yeah well hopefully we'll have time to talk about that a little bit in the context of the learning with latent language chapter as well. But no I mean I think that's very much more and you know in terms of can we get like a little bit more structure in these annotations without going all the way up to language and exploiting it?
Transcript
Play full episode
