Pre-trained language models can autoregressively complete complex token sequences, including spatial patterns. These models can be applied to robotics problems, such as extrapolating sequences of numbers and prompting reward-conditioned trajectories.

https://arxiv.org/abs//2307.04721

YouTube: https://www.youtube.com/@ArxivPapers

PODCASTS:

Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016

Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers