English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Recent studies have shown that CLIP has achieved remarkable success inperforming zero-shot inference while its fine-tuning performance is notsatisfactory. In this paper, we identify that fine-tuning performance issignificantly impacted by hyper-parameter choices. We examine various keyhyper-parameters and empirically evaluate their impact in fine-tuning CLIP forclassification tasks through a comprehensive study. We find that thefine-tuning performance of CLIP is substantially underestimated. Equipped withhyper-parameter refinement, we demonstrate CLIP itself is better or at leastcompetitive in fine-tuning compared with large-scale supervised pre-trainingapproaches or latest works that use CLIP as prediction targets in Masked ImageModeling. Specifically, CLIP ViT-Base/16 and CLIP ViT-Large/14 can achieve85.7%,88.0% finetuning Top-1 accuracy on the ImageNet-1K dataset . Theseobservations challenge the conventional conclusion that CLIP is not suitablefor fine-tuning, and motivate us to rethink recently proposed improvementsbased on CLIP. We will release our code publicly at\url{https://github.com/LightDXY/FT-CLIP}.

CLIP Itself is a Strong Fine-tuner: Achieving 85.7% and 88.0% Top-1 Accuracy with ViT-B and ViT-L on ImageNet