This paper investigates how effectively transformer models can learn new tasks in-context, both within and outside their pretraining distribution. Results show that while transformers excel at learning tasks within their pretraining data, they struggle with out-of-domain tasks.

https://arxiv.org/abs//2311.00871

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