About CLIP ViT-L/14

[githubiubiu]

Thank you for your meaningful work. I would like to know if larger models (such as ViT-L/14 as the image encoder) will be released in the future, or how I can train a larger model.

[zkcys001]

We will release ViT-L/14 on the 400M-scale dataset with long caption~