The paper explores using off-the-shelf vision-language models as sources of rewards for reinforcement learning agents. It demonstrates how rewards for visual achievement of language goals can be derived from CLIP models, leading to more capable RL agents.

https://arxiv.org/abs//2312.09187

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

TikTok: https://www.tiktok.com/@arxiv_papers

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

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