This paper introduces Logit Calibration (LoCa) to enhance knowledge distillation by correcting teacher model predictions while preserving valuable information, improving student model performance without extra parameters.

https://arxiv.org/abs//2409.04778

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