Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Unimodal evaluation results using models pretrained on ImageNet and CCNews #157

Open
cuichenx opened this issue Jul 14, 2022 · 0 comments
Open
Assignees

Comments

@cuichenx
Copy link

Hi! First of all congratulations on the impressive paper and this very well written repo. I've been playing around with this model, and I have a question about Table 4 / C.1 in the paper.
In columns 1 and 2, you report unimodal evaluation results in NLP and vision, but the pretraining dataset used here is PMD.
If I understand correctly, the normal pretraining pipeline is to use ImageNet-1k to train the image encoder, and CCNews+BookCorpus for the text encoder.
I was wondering if you have GLUE finetuning and image linear probe results on (unimodal) models pretrained on ImageNet-1k and CCNews+BookCorpus?
Thanks a lot!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants