https://openreview.net/forum?id=Hntp7s2YfF&referrer=%5Bthe%20profile%20of%20Cihang%20Xie%5D(%2Fprofile%3Fid%3D~Cihang_Xie3)
What If We Recaption Billions of Web Images with LLaMA-3? | OpenReview
Web-crawled image-text pairs are inherently noisy. Prior studies demonstrate that semantically aligning and enriching textual descriptions of these pairs can...
what if