The paper presents a method for approximating training data from model weights, improving performance significantly on classification tasks using a gradient-based approach to select relevant public documents.

https://arxiv.org/abs//2506.15553

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