Question about the paper

facebookresearch / MetaCLIP

ICLR2024 Spotlight: curation/training code, metadata, distribution and pre-trained models for MetaCLIP; CVPR 2024: MoDE: CLIP Data Experts via Clustering

Other

1.17k stars 49 forks source link

Question about the paper #25

Closed pion-changho closed 10 months ago

pion-changho commented 10 months ago

Location: Page 2, end of the Introduction section.

Issue: The text mentions "400 image-text pairs," but I believe it should be "400M" (400 million). Please confirm and correct if necessary.

howardhsu commented 10 months ago

good catch, will revise for the next arxiv version.